Reducing Uncertainties in Quantitative Adverse Outcome Pathways by Analysis of Thyroid Hormone in the Neonatal Rat Brain

Abstract

A number of xenobiotics interfere with thyroid hormone (TH) signaling. Although adequate supplies of TH are necessary for normal brain development, regulatory reliance on serum TH as proxies for brain TH insufficiency is fraught with significant uncertainties. A more direct causal linkage to neurodevelopmental toxicity induced by TH-system disrupting chemicals is to measure TH in the target organ of most concern, the brain. However, the phospholipid-rich atrix of brain tissue presents challenges for TH extraction and measurement. We report optimized analytical procedures to extract TH in brain tissue of rats with recoveries >80% and low detection limits for T3, rT3 and T4 (0.013, 0.033, and 0.028 ng/g) respectively. Recovery of TH is augmented by enhancing phospholipid separation from TH using an anion exchange column coupled with a stringent column wash. Quality control measures incorporating a matrix-matched calibration revealed excellent recovery and consistency across a large number of samples. Application of optimized procedures revealed age-dependent increases in neonatal brain T4, T3, and rT3 on the day of birth (postnatal day, PN0), PN2, PN6, and PN14. No sex-dependent differences in brain TH were observed at these ages, and similar TH levels were evident in perfused vs. non-perfused brains. Implementation of a robust and reliable method to quantify TH in the fetal and neonatal brain will aid in the characterization of the thyroid-dependent chemical interference on neurodevelopment. A brain- vs a serum-based metric will reduce uncertainties in assessment of hazard and risk on the developing brain posed by thyroid disrupting chemicals.

Introduction

Endocrine disrupting chemicals (EDC) are synthetic chemicals that can alter the endocrine system’s function and cause adverse effects on the health of an organism or its progeny (IPCS 2002). Thyroid hormone system disrupting chemicals (THSDC) represent one class of EDCs of particular interest because of the essentiality of adequate thyroid hormones (TH) for growth and development of the nervous system (Bernal 2015; Morreale de Escobar et al. 2004; Williams 2008; Zoeller and Rovet 2004). The complexity of the thyroid system coupled with the intricacies of brain development results in many potential target sites whereby chemical interference with thyroid signaling may negatively impact brain development. Most xenobiotics are identified as potential THSDC by their action to reduce levels of TH in the blood of laboratory animals (Brucker-Davis 1998; Gilbert et al. 2020; US EPA 2014), yet the exact relationship between serum hormone changes and brain function remains unknown. The absence of clear phenotypic profiles of THSDC-dependent developmental neurotoxicity has complicated the evaluation of risks associated with exposure to THSDC. To begin to address these challenges, regulatory bodies in North America and Europe have embraced the Adverse Outcome Pathway (AOP) Framework to assist in translating serum hormone changes in response to chemical exposure in laboratory animals to predict adverse neurodevelopmental outcomes stemming from exposure to THSDC.

Despite a number of international efforts to identify sensitive and clinically relevant phenotypic readouts of chemically-induced perturbations of thyroid signaling in the brain the regulatory community is largely left with serum hormone measures upon which to make their assessments of hazard and risk (Gilbert et al. 2020; Kortenkamp et al. 2020). There are a number of developmental studies using classic hormone synthesis inhibitors such as methimazole (MMI) or propylthiouracil (PTU), that demonstrate decreases circulating levels of TH that are linearly related to changes in structure, function, or thyroid signaling in brain (Axelstad et al. 2008; Darbra et al. 2003; Gilbert and Sui 2006; Gilbert et al. 2022; Gilbert et al. 2014; Gilbert et al. 2007; Hassan et al. 2017; Sharlin et al. 2008; Sui and Gilbert 2003). This correlation however does not always hold for agents that reduce serum TH by mechanisms distinct from synthesis inhibition [see (Gilbert et al. 2020; Gilbert et al. 2021; Ramhoj et al. 2020; Ramhøj et al. 2022). As THs must enter the brain from the blood to enact their effects on brain development, brain hormone concentrations may represent a critical missing link. Brain hormones represent a proximal intermediary in the pathway from serum to TH action that drive hormone-modulated neurodevelopmental processes. Assessment of this key event will aid in the construction of AOPs and evaluation of potential neurotoxicity of THSDCs. It is against this backdrop that we have developed and refined an assay to measure THs in brain tissue from fetal and neonatal rats.

Sensitive and widely used immunoassay-based methods exist for measuring serum TH (Morreale de Escobar et al. 1987; Takaguchi et al. 2022). Tissue concentrations of THs in thyroid gland, brain, liver and placenta have been assessed with radioimmunoassay (RIA), enzyme-linked immunoassay (ELISA), isotope-dilution gas chromatography-mass spectrometry, and liquid-chromatography mass spectrometry (LCMS) (Bastian et al. 2010; Chomard and Autissier 1991; Hornung et al. 2015; Kunisue et al. 2010; Li et al. 2018; Mayerl et al. 2014; Morreale de Escobar et al. 1987; O’Shaughnessy et al. 2018; Obregon et al. 1978; Thienpont et al. 1994; Visser et al. 1978). LC-MS/MS is the preferred method for hormone analysis in any tissue as it can simultaneously and accurately detect multiple TH analytes with the sensitivity and specificity necessary for quantitative precision (Nagao et al. 2011). However, brain TH has proven particularly challenging due to difficulties in efficiently extracting the hormones from the high phospholipid content of brain tissue. The rich lipid brain matrix also interferes with ionization in the LC-MS/MS methods, reducing signal of target THs in the MS electrospray and reducing assay sensitivity (Li and Bartlett 2014). Finally, relative to blood, the concentrations of TH within the brain are quite low, particularly in the fetus and newborn, such that highly sensitive instrumentation and optimal extraction efficiencies are essential for accurate quantification.

The goal of the present study was to optimize a method for brain TH measurement to minimize phospholipid matrix confounds, reduce variability, and achieve a low detection limit. We report an improved solid-phase extraction method using a mixed-mode strong anion exchange cartridge that retains the acidic functional group of THs, permitting more aggressive washes, reducing phospholipid breakthroughs, and lowering detection limits. As proof of concept, we implemented this optimized method and reported serum and brain TH concentrations over a range of ages in neonatal rats.

Methods

Chemicals and Reagents

The analytes 3,3’,5-triiodo-L-thyronine (T3), 3,3’,5’-triiodo-L-thyronine (rT3), thyroxine (T4), 3,3’,5-triiodo-L-thyronine[¹³C₆] hydrochloride (T3-¹³C₆), 3,3’,5’-triiodo-L-thyronine-[diiodophenyl-ring-¹³C₆] hydrochloride (rT3-¹³C₆), and thyroxine-[L-Tyr-ring-13C6] hydrochloride (T4-¹³C₆) all at 100 μg/mL were purchased from IsoSciences (Ambler, PA, USA). Secondary source standards (100 μg/mL each) used for calibration verification were purchased from Millipore Sigma (St. Louis, MO, USA). Internal standards (3,3’,5-triiodo-L-thyronine [ring-¹³C₁₂,99%] (T3-¹³C₁₂) and thyroxine [ring-¹³C_12, 98%] (T4-¹³C₁₂)) were purchased from Cambridge Isotope Laboratories, Inc (Tewksbury, MA, USA). Propylthiouracil (PTU) was purchased from Millipore Sigma (St. Louis, MO, USA). All solvents and reagents used for sample preparation were HPLC grade. Solvents and additives used for mobile phase preparation were Honeywell UHPLC grade and purchased through Government Scientific Source (Reston, VA, USA).

Standard solutions

Each unlabeled standard (T3, rT3, and T4) and each surrogate standard (T3-¹³C₆, rT3-¹³C₆, and T4-¹³C₆) was prepared at 100 ng/mL in methanol from stock internal standards solutions. Internal standards (T3-¹³C₁₂ and T4-¹³C₁₂) were prepared at 100 ng/mL in methanol from stock surrogate standards. A stock standard mix (T3, rT3, T4, T3-¹³C₆, rT3-¹³C₆, and T4-¹³C₆) were diluted to produce an intermediate calibration curve ranging from 0.1 ng/mL to 256 ng/mL in methanol. All standards were stored at −20°C until use. A working solvent calibration curve was prepared by pipetting 10 μL of each intermediate calibration curve point into separate 2-mL high recovery autosampler vials. The calibration curves range from 0.005 ng/mL to 25.6 ng/mL. The internal standard mix was added (5 μL each) into each calibration curve vial resulting in 5 ng/mL each. The solvent in each vial was dried to completion on a TurboVap (Biotage Inc., NC, USA) at 40°C. The dried standards were then re-dissolved in 0.1% acetic acid in 40% methanol. Matrix-matched curves were prepared using 10 μL of each intermediate calibration curve point added to 0.1g of the control brain, and the procedure used for sample preparation was carried out to generate a standard curve ranging in concentration from 0.005 ng/g to 25.6 ng/g.

Analytical Instrumentation

The LC-MS/MS system consisted of an AB Sciex ExionLC AC HPLC system coupled to an AB Sciex 6500+ QTRAP mass spectrometer (AB Sciex, MA, USA). The HPLC autosampler sample temperature was held at 10°C and 2 μL of brain extract was injected. Separation was performed on a Thermo Scientific Accurcore RP-MS column (2.1 × 100mm, 2.6μm; Thermo Fisher Scientific) held at 50°C. The flow rate for the entire analytical run was 0.4 mL/min. Mobile phase A consisted of 0.1% acetic acid in water, and mobile phase B consisted of methanol without additives. The gradient program began at 40% mobile phase B, held for 0.5 min, ramped to 60% B for 3.5 min and then ramped to 98% in 0.5 min. The gradient was held at 98% mobile phase B for 1 min, returned to 40% in 0.1 min and re-equilibrated at 40% for 2 min for a total run time of 6 minutes.

The mass spectrometer detected THs by positive electrospray ionization. Optimization of the source parameters and mass transitions for THs were performed by direct infusion with mobile phase addition (Table 1). Scheduled multiple reaction monitoring (sMRM) was used to enhance sensitivity. Calibration curves were generated as the relative response ratio of the analyte area to the corresponding internal standard area using Multiquant Software (AB Sciex, MA, USA). The curve fit was determined based on the coefficient of determination (r) and the percent difference of the calculated amount versus the theoretical amount.

Table 1.

Optimized ion transitions, retention time (Rt) declustering potential (DP), entrance potential (EP), collision energy (CE), and collision cell exit potential (CXP) for AB Sciex 6500+ Qtrap.

Analyte	Q1 Mass	Q2 Mass	DP	EP	CE	CXP	Analyte Status
T3.1	651.8	605.8	120	10	30	38	Quantitative
T3.2	651.8	478.8	120	10	50	30	Qualitative
T3-13C6	657.8	611.7	95	10	33	38	Surrogate
T3-13C12	663.8	617.8	100	10	32	40	Internal Std
rT3 Quant	651.8	605.8	120	10	30	35	Quantitative
rT3 Qual	651.8	507.8	120	10	30	35	Qualitative
rT3-13C6	657.8	611.7	120	10	33	40	Surrogate
T4 Quant	777.6	731.6	120	10	31	50	Quantitative
T4 Qual	777.6	350.9	120	10	65	35	Qualitative
T4-13C6	783.5	737.7	105	10	35	50	Surrogate
T4-13C12	789.7	743.6	100	10	36	40	Internal Std

Liquid-Liquid Extraction Procedures.

In preparation for solid phase extraction optimization, all brain samples were first processed using standard liquid-liquid extraction procedures adapted from Morreale de Escobar et al. (1987) and further described in (Hornung et al. 2015; O’Shaughnessy et al. 2018). Frozen brain tissue samples (50 – 150 mg) were re-weighed and placed in a 15-mL falcon tube containing 2.8 mm ceramic beads. A volume of methanol containing 1 mM propylthiouracil equal to four times the weight of the samples was added to inhibit deiodinase activity and aid in homogenization. Samples were homogenized on a Bead Ruptor 24 (Omni International, GA, USA) at 5.0 m/s for 1 cycle of 20 seconds. A volume of chloroform was added to each sample at two times the volume of methanol containing 1 mM propylthiouracil added. The samples were vortex mixed for 30 seconds and centrifuged at 2,000 rpm for 15 minutes. The supernatant was transferred to a clean labelled 15-mL falcon tube.

The remaining pellet was extracted two additional times with chloroform:methanol mixture (2:1, v:v), centrifuged at 3000 rpm for 15 minutes, and the supernatant from each extraction was combined with the initial supernatant. The combined volume was recorded, and 1 mL of 0.05% calcium chloride solution was added for every 5 mL of combined extract. The samples were vortex mixed for 30 seconds and centrifuged at 2000 rpm for 5 minutes to separate the organic layer from the aqueous layer. The upper aqueous layer was transferred to a clean 15-mL falcon tube, and the volume was recorded. The remaining organic layer was re-extracted with chloroform: 1 mM propylthiouracil in methanol:0.05% CaCl₂ (3:49:48, v:v:v) at equal volumes of the aqueous layer previously transferred, vortex mixed and centrifuged at 2000 rpm for 5 minutes. The supernatants from the two additional extractions were combined with the first aqueous extraction. The combined volume was recorded, reduced by half in a TurboVap LV (Biotage, NC, USA) using nitrogen at 40°C, ensuring only water remains. After solvent reduction, 1 mL of LC/MS grade water was added, and the volume was recorded. An equal volume of 4% formic acid in water:acetonitrile (80:20, v:v) was added. The sample was vortex mixed prior to solid phase extraction.

Study 1. Solid Phase Extraction Optimization

Solid Phase Extraction – A Comparison of Extraction Cartridges

The rich phospholipid matrix that constitutes brain tissue makes for poor TH recovery, limiting accuracy and reproducibility. The performance of four solid phase extraction cartridges was examined for their efficacy in sequestering hormones from the fatty matrix of brain tissue. Cartridges from Waters Oasis HLB, Phenomenex Strata-X-Pro, Biotage Evolute Express CX, and the Biotage Evolute Express AX were selected for this purpose (Biotage, NC). These comparisons were conducted on brain tissue from an adult Long Evans male rat, homogenized in methanol and PTU, and subjected to liquid-liquid extraction procedures described above. Solid-phase extraction generally followed the generic instructions and application notes provided by the manufacturer for each cartridge summarized in Table 2. In attempts to augment phospholipid removal, samples assessed using the Biotage Evolute Express CX and AX cartridges were subjected to an additional wash in 2% formic acid in dichloromethane. Phospholipid carryover in brain tissue was monitored using the liquid chromatography mass spectrometry used for hormones, but with the mass spectrometer set to monitor precursor mass of 184 m/z indicative of phosphatidylcholine. Phosphatidylcholine was the readout used to evaluate the efficiency of the phospholipid removal from THs extracts.

Table 2.

Comparison of solid phase extraction media for the concentration of thyroid hormones from brain tissue and the removal of matrix interference.

Step	Evolute Express CX	Evolute Express AX	Oasis PRiME HLB	Strata X Pro
Pre-Treatment	Dilute volume of 2% Formic Acid:ACN (60:40) v/v	Dilute with equal volume of 5% NH4OH:ACN (1:1) v/v	Dilute with equal volume of 2% Formic Acid	Dilute with equal volume of 2% Formic Acid
Conditioning	3 mL MeOH	3 mL MeOH	3 mL MeOH	-
Equilibration	3 mL 2% Formic Acid	3 mL H2O	3 mL H2O	-
Load	Pre-treaded Sample	Pre-treaded Sample	Pre-treaded Sample	Pre-treaded Sample
Wash 1	3 mL 50mM NH4OAC, pH 6	3 mL 50mM NH4OAC, pH 9	3 mL 30% MeOH	3 mL 30% MeOH
Wash 2	3 mL 2% Formic Acid	2 mL MeOH	-
Wash 3	3 mL MeOH	3 mL 2% Formic Acid in DCM	-
Elution	1 mL 2.5% NH4OH in MeOH	1 mL MeOH	1 mL MeOH	1 mL ACN/MeOH (90:10)

Study 2. Method Validation – Calibration, Recovery, Accuracy, Precision, Sensitivity

In order to validate the brain TH method, a standard set of procedures were followed as outlined below and summarized in Table 3. Validation studies were conducted using stable isotope-labeled standards as internal surrogates. The assay characteristics were probed using these tracers by contrasting calibration curves prepared in solvent and brain matrices, evaluating the efficacy of additional stringent column washes, examining the precision across different runs on different days, and finally defining the lower limits of detection and quantification.

Table 3.

Critical Features of Validation of the Analytical Method for Thyroid Hormone (TH) in Brain Tissue in Study 2.

1. MATRIX EFFECTS and CALIBRATION

TH are embedded in the rich phospholipid matrix of brain tissue. The degree to which this matrix can interfere with the accuracy of the measurement of TH needs to be minimized. Calibration curves were generated using 8 concentrations of each surrogate analyte with 5 replicates of solvent-based standards, matrix-matched standards, and blank standards. The solvent standards were spiked with unlabeled surrogates. Matrix and blank standards undergo the entire extraction procedure. After the extraction process, blank matrix standards were spiked with labeled surrogate TH analytes to generate the ‘spiked standards’. Each set of replicates was analyzed for T3, T4 and rT3 (Figure 3) and efficiencies of recovery, precision and matrix effects calculated as described below.

2. ACCURACY/RECOVERY

Accuracy of the analytical method for TH is reflected in the percent recovery of the T3, T4 and rT3 analyte from the matrix - greater recovery equating to better accuracy. Accuracy was determined by spiking 20 replicates of clean matrix with a concentration at or below the mid-point of the calibration curve and results are summarized in Table 4. The closer the measured analyte is to the spiked volume, the greater the accuracy of the method.

3. PRECISION

Precision of the analytical method refers to within- and between- day variability of the measurement of the TH analyte. Precision was determined by analyzing 5-replicate spiked samples at 3 different concentrations over 3 days. The replicate samples were prepared using blank matrix and spiked at concentrations that spanned the calibration range. The relative standard deviation (RSD) of the concentrations observed for the three sets of replicates provides the estimate of precision as reported in Tables 4 and 5.

4. SENSITIVITY

Sensitivity of the analytical method refers to the Method Detection Limit (MDL) and Method Quantitation Limit (MQL) for each TH analyte. The MDL is the lowest level that can be detected in the chromatogram, the MQL is the lowest level that can be accurately quantified in the chromatogram for each TH analyte. These values are reported in the text.

Calibration Curves.

Calibration curves were prepared by spiking each surrogate standard in solvent and a homogenate of adult rat brain. Brain tissue contains endogenous THs; therefore, surrogate matrices suitable to match this substrate are not readily available. Surrogate standards, T3-¹³C₆, rT3-¹³C₆, T4-¹³C₆, and the labeled internal standards, T3-¹³C₁₂ and T4-¹³C₁₂, were spiked in each to sample type permit the direct assessment of extraction efficiency. A third calibration curve was generated using unlabeled standards in solvent (i.e., neat standard) and all three 8-point curves were generated at concentrations ranging from 0.005 ng/g to 25 ng/g of each surrogate standard.

Recovery and Matrix Effects.

Recovery and matrix effects were evaluated by pre-spiking surrogate standards (T3-¹³C₆, rT3-¹³C₆, and T4-¹³C₆) at 5.00 ng/g into brain tissue, extracting hormones using the liquid-liquid method and the optimized SPE method. Post-spiked samples were prepared by adding the surrogates to post-extracted brain tissue. Based on the results of Study 1, the Biotage Evolute Express AX-60 mg was selected for further study. Moreover, a comparison was conducted with and without acidified dichloromethane wash. The responses of the pre-spiked extracted brain samples were compared to the average responses of the post-spiked extracted brain sample to generate the hormone recovery estimates. The SPE overall recovery was calculated as follows:

$$\mathrm{\%}\text{Recovery}=\frac{\text{Peak}\text{Area}\text{Response}\text{Pre}\text{Spiked}}{\text{Peak}\text{Area}\text{Response}\text{of}\text{Post}\text{Spiked}}\times 100$$

where the Peak Area Response Pre-Spiked is the area of the chromatogram peak of the surrogate TH spiked prior to liquid-liquid extraction and SPE. Peak Area Response Post-spiked is the area of the chromatogram peak of the surrogate TH that was spiked into the control brain matrix after extraction.

Matrix effects were determined by comparing the chromatogram peak areas of the post-extracted samples to those of the solvent spiked neat standard. The equation for the percent matrix effects used in this study is as follows:

$$\mathrm{\%}\text{Matrix}\text{Effect}=\left(\frac{\text{Peak}\text{Area}\text{Response}\text{Post}\text{Spiked}\text{Sample}}{\text{Peak}\text{Area}\text{Response}\text{Neat}\text{Standard}}-1\right)\times 100$$

where the numerator ‘Peak Area Response Post-Spiked Sample’ is the averaged chromatogram peak are for the surrogate TH that was spiked after extraction of control brain matrix. The denominator ‘Area Response Neat Standard’ is the average chromatogram peak area of the injection of surrogate standards prepared in solvent.

Method Precision and Accuracy.

Precision and accuracy were evaluated by comparing pre-and post-spiked samples fortified at 5.0 ng/g of ¹³C₆ labeled THs using the AX60 cartridge with the additional wash step. The injection was repeated 5 times for each spiked sample and over 3 days to calculate intra- and inter-day recoveries.

Sensitivity - Method Detection Limit.

The method detection level was performed according to EPA’s 40 CFR 136-Definition and Procedure for the Determination of the Method Detection Limit-Revision 1.11. Seven matrix extracts at the 0.01 ng/g were analyzed over three days, and the method detection limit was calculated using the following calculation:

$$\mathrm{M}\mathrm{D}\mathrm{L}=t_{\left(n-\mathrm{1,1}-\alpha =0.99\right)}\times \mathrm{S}\mathrm{D}$$

where: $\mathrm{M}\mathrm{D}\mathrm{L}$ = the method detection limit; ${\mathrm{t}}_{(\mathrm{n}-\mathrm{1,1}-\mathrm{\alpha }=0.99)}$ = The students’ t value appropriate for a 99% confidence level and a standard deviation estimate with n-1 degrees of freedom; and $\mathrm{S}\mathrm{D}$ = standard deviation of the replicate analyses. The method quantitation limit (MQL) was determined by multiplying the calculated MDL by 3.18 when seven replicate analyses are used.

Study 3: Assessing Suitability of the Method for Neonatal Rat Brain

It is assumed that, as for serum TH, brain TH will likely be at lower concentrations in the fetus and neonate than in the adult. It is undoubtedly true that tissue mass increases dramatically over developmental time such that more limited samples with lower levels of hormone present are available from the fetus and newborn. Although we used homogenates of the adult whole brain to optimize methods for brain TH quantification, the goal of the study was to develop a means to quantify THs in the fetal and neonatal rat brain. This study was conducted to test if the method was sufficiently sensitive to reliably measure THs in the brains of young animals. Brain tissue was collected from Long Evans rat pups (n=6) on postnatal day (PN) 2. Animals were euthanized by decapitation, the brains quickly removed from the skull and rinsed in cold phosphate-buffered saline. In this study, the forebrain was isolated for hormone analysis by gently lifting the cortical mantle to separate it from the midbrain/brainstem. It was rinsed, weighed and flash frozen in liquid nitrogen. Samples were analyzed according to the optimized procedures described above using the AX Evolute Express-60 mg extraction cartridge with acidified dichloromethane wash.

Study 4. Proof on Concept – Ontogeny of Brain Hormones Animals and Tissue Collection

Animals.

Pregnant Long-Evans (LE) rats (n=6) were obtained from Charles River (Raleigh, NC) on gestational day (GD) 2 and housed individually in standard plastic hanging cages. All animal rooms were maintained on a 12:12 light:dark schedule, and animals were permitted free access to food (Purina 5008 rat chow) and purified tap water. All procedures were conducted with prior approval from the United States Environmental Protection Agency’s Institutional Animal Care and Usage Committee (IACUC) and were carried out in an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) approved facility. On the day of birth (denoted as PN0), pups were sexed, counted, and assigned by age group for collection, saline perfusion, or tissue collection without perfusion. To the degree possible, sex was balanced across perfusion states, and only a single pup from each litter was in any given age or perfusion state. For perfusions, pups were euthanized with an overdose of Euthasol (50mg/kg. ip) and perfused with cold phosphate-buffered saline (PBS). A hypodermic needle was inserted into the heart, the aorta clipped, and ~25ml? PBS slowly perfused via gravity feed until fluid ran clear. In the non-perfused condition, pups were euthanized by decapitation, and trunk blood was saved, pooling across sex, for serum hormone analysis. A similar procedure was followed on PN2, PN6, and PN14, one pup from each litter was perfused or euthanized by decapitation, counterbalancing sex across litters. Blood samples were placed on ice and allowed to clot for 30 minutes before centrifugation at 4°C, 3000 × g for 30 minutes to isolate serum. Serum THs were measured using LC-MS/MS as described in (Hassan et al. 2017). Whole brains were removed from the skull, the olfactory bulbs and spinal cord at the level of the cerebellum removed, and rinsed in cold PBS, weighed, and flash frozen in liquid nitrogen. Brains collected from pups on PN6 and PN14 were rinsed and cut along the midline, each hemisphere individually weighed and frozen in liquid nitrogen. A minimum of 5 brains from independent litters were available for testing in each of the perfused and non-perfused state at each age.

Sample Preparation.

The solid-phase extraction was performed using an Evolute AX strong anion exchange cartridge (60 mg/3mL) from Biotage, Inc. with the additional 2% formic acid in dichloromethane wash. The AX cartridges are mixed mode, meaning they have anion exchange and reverse phase capabilities. The cartridge was initially conditioned with 3 mL of methanol followed by equilibration with 3 mL of water. The sample from the liquid-liquid extraction was then applied to the cartridge, and the column was washed with 3 mL of 50 mM ammonium acetate (pH 9). A 3mL methanol wash was applied to the cartridge while the THs were adhered to the cartridge by anion exchange. The cartridge was then washed with 3 mL 2% formic acid in dichloromethane and allowed to dry for 2 minutes. The use of the formic acid in dichloromethane releases the hormones from their adhesion by ion exchange making them available for elution while simultaneously removing remaining phospholipids. There is no loss of THs with this wash as hormones are not soluble in dichloromethane. Retained THs were then eluted from the cartridge with 3 aliquots of 0.8 mL of methanol and the final extract was transferred to a salinized polypropylene tube and evaporated to dryness in a TurboVap LV (Biotage, NC, USA) with nitrogen at 40°C. The dried sample was re-dissolved in 100 μL of 0.1% acetic acid in 40% methanol and placed on the instrument for hormone measurement as described above.

Statistical Analysis

Statistical analyses were performed using PRISM Graph Pad and SAS using t-tests, one-way, or repeated measures ANOVA where appropriate.

RESULTS

Results of Study 1: Solid Phase Extraction Cartridge Comparison

Each of the four solid phase extraction cartridges resulted in measurable quantities of TH from brain tissue; however, significant differences were observed in recovery. Figure 1 presents representative chromatograms of unlabeled THs added at a low (0.125 pg) and high (32 pg) concentration. Peaks for T3, T4, and rT3 are clearly resolved between 2.4 and 3.0 minutes even at very low concentrations (Figure 1A). Efficiency of phospholipid removal from the SPE eluent in response to different SPE cartridges was visualized by monitoring the presence of phosphatidylcholine with ion scanning set to 184m/z. It can be seen from the chromatograms in Figure 2 incomplete removal of phospholipids (phospholipid breakthrough) during the extraction process can significantly interfere with the detection of TH analytes. A comparison of the four SPE cartridges used on adult rat brain tissue reveals significant phospholipid breakthrough is evident in the Waters Oasis PRiME HLB and Phenomenex Strata X Pro cartridges, both of which were outperformed by the Biotage Evolute Express CX and AX cartridges. The best extraction was achieved with Biotage AX, but recoveries of labeled surrogates for each hormone analyte still fell below 50% resulting in suboptimal efficiency (Table 4A). An acidified dichloromethane column wash was used on the two Biotage cartridges to improve matrix interference removal. This action had minimal effect on the recoveries in the CX cartridge (data not shown) but completely eliminated phospholipid carryover in the final eluent from the AX cartridge (Table 4B and 4C and also depicted in the green spectrum in Figure 2). Combining AX cartridge with an acid wash ultimately minimized ion suppression in the LC-MS/MS analysis and significantly improved signal detection (Table 4C).

Figure 1.

Representative chromatogram of unlabeled thyroid hormones T3, rT3, and T4, injected on column at a (A) low amount (0.125 pg) and a (B) high amount (32 pg). Peaks for each hormone analyte are clearly resolved between 2.0 and 3.0 minutes. The chromatogram on the left with low intensity range on the y-axis shows that the signal-to-noise ratio is greater than 3.

Figure 2.

Investigation of phospholipids reduction in brain matrix from three different extraction procedures. Phosphatidylcholine was used as a readout of phospholipids using precursor ion scanning 184 m/z product ion. Solid-phase extraction with the Evolute Express AX cartridge (green) with the addition of a 2% formic acid in dichloromethane wash outperformed the other SPE cartridges, virtually eliminating phospholipid breakthrough.

Table 4. Optimizing Solid Phase Extraction to Improve Hormone Recovery.

Matrix Effects Study using ¹³C₆ labeled thyroid hormones as surrogates, pre- and post- spiked samples at 1.00 ng/g in A, B and C. Recovery %, Matrix Effect % and Process Efficiency % are as defined in text.

A. Evaluation of Evolute Express AX 60 mg cartridges using the manufacturer’s recommended method. Recovery was poor, <50% for all 3 hormone analytes.
	T3-13C6	rT3-13C6	T4-13C6
Recovery %	46	41	35
Matrix Effects % (ME)	99	110	112
Process Efficiency (%)	45	45	37
B. Recovery and process efficiency increases using Evolute Express AX 60 mg cartridges with an optimized wash step (2.0% formic acid in dichloromethane) and nitrogen evaporation in salinized glass tubes.
	T3-13C6	rT3-13C6	T4-13C6
Recovery %	62	58	63
Matrix Effects % (ME)	98	91	100
Process Efficiency (%)	70	62	73
C. Further enhancement of recovery using Evolute Express AX 60 mg cartridges with an optimized wash step (2.0% formic acid in dichloromethane) and nitrogen evaporation in low retention polypropylene tubes versus salinized glass tubes used in B. Recovery improved relative to conditions tallied in B.
	T3-13C6	rT3-13C6	T4-13C6
Recovery %	80	78	82
Matrix Effects % (ME)	98	91	100
Process Efficiency (%)	88	82	92

Results of Study 2. Methods Validation.

Methods validation was carried forward using the Biotage AX-60 mg exchange column coupled with a stringent anion column wash. Calibration curves generated by spiking solvent and matrix-matched samples with labeled T3, rT3, and T4 at concentrations ranging from 0.05 to 25 ng/g are presented in Figure 3. Measured calibration points fell within 20% of their predicted value, with r² values for all curves exceeded 0.995. The surrogate solvent curve closely approximated the matrix-matched calibration curve indicating that a solvent-based curve is sufficient for quantification.

Figure 3.

Calibration curves were generated by spiking solvent and matrix-matched samples with labelled T3, rT3 and T4 at concentrations ranging from 0.005 ng/g- 25ng/g. Measured calibration points were within ± 20% of their predicted value. The coefficient of determination, r2 for all curves was ≥ 0.995. The surrogate solvent curve closely approximated the matrix matched curve.

Precision and accuracy were evaluated by comparing pre-and post- spiked samples fortified at 5ng/g of ¹³C₆ labeled THs. Excellent recovery was obtained (>94%), well above the acceptable cut-off limits of <20 and >80%. The mean and standard deviations of technical replicates within and across days are summarized in Table 5.

Table 5.

Precision and Accuracy Results using samples fortified at 5.00 ng/g of ¹³C₆ labelled thyroid hormones. The samples were processed using the Evolute Express AX 60 mg cartridges with the additional wash step. High recovery with low intra- and inter-day variability was evident. RSD=Relative Standard Deviation of averaged concentrations of replicates within and across days.

	Intra-day (n=5)		Inter-day (n=15)
Analyte	Measured (ng/g)	%RSD	Measured (ng/g)	%RSD	% Recovery
T3-13C6	4.75 ± 0.088	1.85	4.74 ± 0.220	4.65	94.9 ± 4.4
rT3-13C6	5.02 ± 0.589	11.7	5.14 ± 0.744	14.5	102 ± 4.9
T4-13C6	5.09 ± 0.146	2.87	5.00 ± 0.293	5.85	100 ± 5.9

Finally, the detection and quantification limits (MDL and MQL) were derived by assessing seven matrix extracts at 0.05 ng/mL The experimental MDLs for T3, rT3, and T4 were 0.013, 0.033, and 0.028 ng/g, respectively. The experimental MQLs were 0.105, 0.040, and 0.090 ng/g for T3, rT3, and T4.

Results of Study 3. Assessing Suitability of the Method for Neonatal Rat Brain

The results of brain hormones measured in forebrain of PN2 rat pups (n=6) are presented in Figure 4A. Mean concentrations of T3 and T4 fell between 1.0–1.5 ng/g, well above the MDL for the method. Concentrations of rT3, were significantly lower (mean=0.051 ng/g) but still within the detectable range. Recoveries for each analyte were well above 80% signifying a robust and stable extraction procedure. A summary of all quality control results showing recoveries of labeled ¹³C₁₂-T3 and ¹³C₁₂-T4 is presented in Figure 4B and 4C. Of the 44 quality-control runs acquired with this optimized method, only two fell outside of the acceptable range of +/− 20%, the majority showing recoveries being within 5–10% of expected. These data signify that this method, developed using adult rat brain tissue, is suitably sensitive and efficient for the quantification of THs in the brains of neonatal rat pups.

Figure 4.

A) To assess the suitability of the methods optimized using the Biotage AX-60 exchange column to reliably detect hormones in the brains of young rat pups, forebrains collected from rat pups (n=6) on postnatal day 2 were assessed for thyroid hormones. T3 and T4 were detected in the 1–2 ng/g range and rT3 at much lower concentration, but well above the minimal detectable limit of 0.033ng/g. Quality control is exemplified in the summary of recoveries in analytical runs across a range of studies conducted with this method as shown in B) for ¹³C₁₂-T3 and C) for ¹³C₁₂-T4. In each analysis only 2 of 44 runs (4.5%) fell outside the lower control limits of 80%.

Study 4: Proof of Concept - Thyroid Hormones in Serum and Brain Increase With Age

Serum T3 and T4 increased as a function of age in samples collected from newborn pups on the day of birth (PN0), PN2, PN6, and PN14. T4 was typically 10X higher than T3, expressed in ng/mL for T4 and ng/dL for T3 (Figure 5A). It is possible that residual blood in brain tissues may interfere with the quantification of THs in the brain. To examine this directly, littermates were perfused with physiological saline to clear the brain of blood and brain hormone concentrations contrasted between perfused and non-perfused brain at each age. Mirroring the effects in serum, age-dependent increases in brain hormones were observed, with no differences in hormone concentrations evident as a function of sample preparation (Figure 5B). These observations were supported by results of ANOVA where significant main effects of Age [F(3,49) p<0.0001] were identified for T4, T3, and rT3, with no main effect of Perfusion [F(1,42)=1.50] or Age X Perfusion [F(3,42)=0.70, p>0.55] interaction present. Distinct from serum hormone, where large differences in concentrations are evident between concentrations of T3 and T4, brain T3 and T4 levels were quite similar, with brain T3 slightly higher than T4 at all ages tested.

Figure 5.

Serum and brain thyroid hormones in neonatal rat pups. A) Serum collected from male and female pups was pooled at each age. T3 and T4 increased with age (n=5/age group). B) Brain T3 and T4 increased with age and no differences were seen between perfused vs non-perfused brains (n=5–7/group). C) No significant differences in brain T3 or T4 were seen between male and female pups at the ages tested.

As no differences were observed as a function of perfusion condition, within this dataset, it was possible to examine the potential effects of sex on brain hormones. Brain hormone concentrations were similar in male and female rat pups at all ages (n=5–6/sex/age), increasing with age in a parallel fashion (Figure 5C). Results of ANOVA supported these conclusions with a main effect of Age [F(3,42), p<0.0001 for T3 and T4, p< 0.034 for rT3] with no effect of Sex or Age X Sex interaction (all p’s >0.55).

DISCUSSION

Here we report the optimization of LCMS procedures to measure TH in brain tissue of rats. Hormone recovery was enhanced by reducing phospholipid carryover using established LLE procedures, an anion exchange column, a stringent column wash, and enhanced evaporation procedures. MDL and MQLs were determined with this optimized method using both solvent and surrogate spiked brain matrix samples, with similar results. MDL for quality control measures revealed excellent recovery and consistency across sample runs. Detection limits for T3, rT3 and T4 of 0.013, 0.033, and 0.028 ng/g, respectively. In neonatal pup brains T3 and T4 ranged from ~0.5–5 ng/g, while rT3 was generally not detected. In combination, the implementation of these procedures produced reliable age-dependent measures of brain T4 and brain T3.

Consistent with previous reports, serum TH increased with age, and no sex differences were detected at the ages tested (Christian and Trenton 2003; Gilbert and Sui 2008). Mirroring that seen in serum, brain TH also increased from PN0 to PN14, and no differences were detected between brains from male and female pups.

It has been widely debated that residual blood in brain tissue may interfere with the quantification of THs in brain (Reyns et al. 2003; Taves et al. 2011). Whole body saline perfusion has been used routinely by some laboratories in quantifying small molecules, TH, and other steroids in brain (Barez-Lopez et al. 2014; Mayerl et al. 2014; Taves et al. 2011). In our study, we compared TH concentrations in brain samples collected from neonatal rats that were whole body perfused or euthanized by decapitation without perfusion at four postnatal ages. Consistent with the findings of Taves et al. (2011), perfusion to remove blood contaminates (1% blood, v/w) did not increase or decrease the brain TH concentrations.

Although T3 and T4 brain concentrations were similar in absolute values, slightly higher levels of T3 over T4 were consistently seen at all ages tested. This observation was somewhat surprising given the very large differences in these hormones in the periphery, where T4 in serum exceeds concentrations of T3 by >10-fold. Although surprising, the observation is not unique to our study. On review of the literature, it is apparent that T4:T3 ratios vary as a function of laboratory, measurement method (most common has been RIA), age (most common has been adult), and brain region. Many laboratories perfused animals routinely, others report results from nonperfused rats and mice. For whole brain estimates, higher concentrations of T4 over T3 were reported (T4/T3 ratios ranged from 1.0–3.0) in fetal and adult rat brain (Calvo et al. 1990; Grijota-Martinez et al. 2011; Kunisue et al. 2010; Morreale de Escobar et al. 1985). In other studies, as in our own, T3 exceeded T4 concentrations (T4/T3 ratios ranging from 0.22–0.87) in whole brain or forebrain/cortex in newborn, weanling, and adult rats and mice (Berbel et al. 2010; Campos-Barros et al. 1997; Galton et al. 2014; Morse et al. 1993; Morse et al. 1996). The largest differences in both absolute and ratio measures for brain THs were from studies comparing THs in different brain regions, yet in our albeit limited review of this literature, no clear pattern of species, region, or age emerged (Baumgartner et al. 1997; Campos-Barros et al. 1997; Galton et al. 2014; Galton et al. 2009; Joffe et al. 1994; Morreale de Escobar et al. 1985). THs gain access to the brain from the blood with a family of transporter proteins, some that preferentially carry T3, others with selectivity for T4. A set of more promiscuous transport proteins also show affinity for shuttling both T3 and T4 (Groeneweg et al. 2020). In addition to transport, tight control over T3 concentrations, the ‘active’ vs the “pro” -hormone T4, is achieved by deiodinating enzymes in brain. Both of these regulatory systems governing influx and metabolism of THs exhibit distinct ontogenies that vary temporally and spatially in the developing brain. As regional and temporal differences are well described for TH requirements during brain development, it is not surprising that considerable variability exists in reported estimates of hormone analytes in different brain regions, in different rodent species, across distinct lifestages. However, as we have reported here, it is likely that varying degrees of extraction efficiency in TH measurements represent yet another source of variability.

In conclusion, in many AOP schemes, tissue hormone concentrations occupy a critical key transition node in the sequence of biological steps spanning chemical interaction at a target site within the thyroid system to subsequent impairments in brain development (Crofton 2018; Gilbert et al. 2020; Hassan et al. 2017; Rolaki et al. 2019). Serum hormones, although important clinically and routinely assessed in the regulatory setting, do not always faithfully reflect tissue hormone concentrations (Reyns et al. 2003). Brain hormones, immediately downstream from serum THs, provide a direct link from serum to brain, but are rarely assessed due to the technical challenges associated with brain tissue. Here we report improved methods for the quantification of THs in brain tissue. However, it must be cautioned that even if robust estimates of brain THs can be achieved, whole brain or even brain region estimates of THs remain crude measures. Concentrations of T3 and T4 inside the cell are the drivers of receptor-mediated action of THs on neurodevelopmental processes. Intracellular concentrations of THs can vary widely from brain region to brain region, from cell type to cell type within a brain region and exhibit dramatic changes over the course of development time (Dentice et al. 2013; Hernandez et al. 2010). Despite these limitations, estimates of whole brain THs remain valuable metrics to inform and refine evaluations of the potential impact of THSDC on brain development. With improved methods, incorporation of robust and reliable quantification of THs in fetal and neonatal brain will augment thyroid-dependent neurodevelopmental quantitative AOPs to reduce uncertainties in assessment of hazard and risk of THSDC.