Optimized Mass Spectrometry Detection of Thyroid Hormones and Polar Metabolites in Rodent Cerebrospinal Fluid

Abstract

Thyroid hormones (TH) are required for brain development and function. Cerebrospinal fluid (CSF), which bathes the brain and spinal cord, contains TH as free hormones or as bound to transthyretin (TTR). Tight TH level regulation in the central nervous system is essential for developmental gene expression, which governs neurogenesis, myelination, and synaptogenesis. This integrated function of TH highlights the importance of developing precise and reliable methods for assessing TH levels in CSF. We report an optimized liquid chromatography-mass spectrometry (LC-MS)-based method to measure TH in rodent CSF and serum, applicable to both fresh and frozen samples. Using this new method, we find distinct differences in CSF TH in pregnant dams vs. non-pregnant adults and in embryonic vs. adult CSF. Further, targeted LC-MS metabolic profiling uncovers distinct central carbon metabolism in the CSF of these populations. TH detection and metabolite profiling of related metabolic pathways open new avenues of rigorous research into CSF TH and will inform future studies on metabolic alterations in CSF during normal development.

Keywords: cerebrospinal fluid; development; mass spectrometry method; metabolomics; reverse-phase chromatography; rodent; thyroid hormone.

Figure 1

A method to quantify triiodothyronine (T3) and thyroxin (T4) by LC-MS. (A) Optimization of HESI parameters on Orbitrap mass spectrometer for the indicated metabolites using C18 chromatography and standards. Graphs represent integrated areas of chromatographic peaks under changing parameters for capillary temperature or S-lens. (B,C) Retention times and sensitivity comparison among four chromatography methods. Chromatographic runs were carried out on either “ZIC-pHILIC” (ZIC-pHILIC 150 × 2.1 mm, 5 µm particle size, EMD Millipore); “Accucore Amide” (Accucore™ 150 Amide HILIC (150 × 3 mm, 2.6 mm particle size; Thermo Fisher Scientific)); “LUNA NH₂” (Luna^® 3 µm NH₂ 100 Å, LC Column (150 × 2 mm, 3 µm particle size; Phenomenex, 00F-4377-B0)); or “C18” (Ascentis Express C18 HPLC column (2.7 µm × 15 µm × 2.1 mm; Sigma Aldrich)) with 22 min (polar chromatography) or 18 min (C18 reverse-phase chromatography) linear gradients, respectively. Overlaid peaks are shown for the indicated ranges of (B) polar and (C) reverse-phase chromatographic methods. Overlayed are corresponding chromatographic gradient compositions. Metabolite stock solutions were diluted in 10 µm ammonium hydroxide solution at 1:1 ratio of methanol/water. (D) Limit of detection (LOD) and linearity for individual T3 and T4 standards, respectively, on C18 reverse-phase chromatography. Presented is one experiment from two representative dilutions; R²—goodness of fit; “LOD”—linear limit of detection. (E) MS² spectra comparison to different collisions for T3 and T4 standards.

Figure 2

Optimization of extraction and reconstitution conditions for T4 and T3 from cerebrospinal fluid (CSF). (A) Optimization of solvents for extraction of T3 and T4 from CSF. Single-phase (“MB”: methanol-based; see methods for further details) vs. two-phase (lipid and polar phase, respectively) extractions are compared. Normalized peak integration areas are shown as the average and standard deviation for CSF collections in triplicate. (B) Fine-tuning of mass spectrometry detection parameters. Two narrower detection scans (tSIM vs. “narrow”, corresponding to 600–800 m/z) were compared for T3 and T4 extracted from either 5 μL or 1.4 μL of CSF. Normalized peak integration areas are shown as the average and standard deviation for CSF samples in triplicate. (C) Comparison of reconstitution efficiency for isotopically labeled T3 and T4 standards spiked in mouse CSF (see also Figure S4). Both lipid (bottom) and polar (top) phases from two-phase extraction were compared, where each was reconstituted in either water or 70% ACN. Normalized peak integration areas are shown as the average and standard deviation for two independent extractions. (D) Comparison of the partitioning efficiency of extracted T3 or T4 between the polar and lipid phase of a two-phase extraction for rat CSF or serum. Normalized peak integration areas are shown as the average and standard deviation for quadruplicate CSF collection and triplicate serum collection.

Figure 3

Characterization of T3 and T4 extracted from CSF and serum after different storage conditions. (A) Comparison of storage conditions and signal stability upon freezing and storage for rat CSF and serum T3 or T4. Freshly extracted CSF extract was compared to fresh, acutely frozen and then thawed (labeled “frozen”), and frozen and then stored at −80 °C for 24 h (labeled “stored”). Bottom and top phases from a two-phase extraction are further compared. Normalized peak integration areas are shown as the average and standard deviation for at least a triplicate CSF collection. (B) As in (A) but mouse serum levels of T3 and T4 were compared to two storage conditions of mouse CSF and a mock sample (empty tube processed as samples). CSF was collected from at least 8 mice with paired serum from 4 mice.

Figure 4

Levels of T3 and T4 differ between adult, embryo, and pregnant mice. (A) Comparison of T3 and T4 levels extracted from adult and embryonic mouse CSF and adult serum. Normalized peak integration areas are shown as the average and standard deviation for at least five CSF collections and four paired serum collections. (B) Comparison of T4 levels extracted from CSF of adult males, pregnant females, or embryos. Normalized peak integration areas are shown as the average and standard deviation for at least eight CSF collections. Missing values (zeros) were excluded from the analysis. (C,D) PCA (C) and heatmap (D) analyses of embryo, female, and male CSF polar metabolites using HILIC chromatography LC-MS. Detected metabolites were Pareto scaled and log transformed within the MetaboAnalyst online platform. (D) Top 25 changed metabolites are shown.