Modified h7H human thyrocyte culture medium (HTCM) was used for all thyrocyte culture experiments as reported previously (Foley et al., 2024). Medium was sterile filtered through a 0.22 µm PES filter unit (ThermoFisher Scientific, Waltham, MA).

All methods were performed in accordance with the guidelines and regulations of LifeNet Health’s ethics committee. Informed consent was obtained for all donor tissue for research purposes by LifeNet Health. Thyrocytes from adult euthyroid donor tissues were isolated using modifications of a previously published isolation and cryopreservation protocol (Deisenroth et al., 2020). A panel of donors, male and female, used in this study were of varying ethnicity, age, and body mass index (BMI) (Table 1).

For functionality studies, 96-well white-wall clear-bottom microplates were used (Perkin Elmer, Waltham WA). For immunostaining, 96-well black-wall clear-bottom microplates were used (Corning, CA). All microplates were tissue culture-treated and either coated with 50 µL of Matrigel^® at the concentration of 9–10 mg/mL for 3D culture or left uncoated for 2D applications. Coated microplates were incubated for 1 h at 37°C CO₂ to allow Matrigel^® to solidify. Cryopreserved human thyrocytes were thawed in a 37°C water bath for ≥90 s until the cell suspension liquified and could slide freely when the vials are inverted (approx. 2 min max) before being added dropwise into 14 mL of human thyrocyte plating medium (HTPM), which is HTCM without bovine TSH. The cell suspension was centrifuged for 5 min at 200 x g in a swing bucket rotor. The pellet was resuspended in 1 mL of fresh plating medium, stained with 0.4% Trypan Blue (Gibco, Billings, MT) and counted using an automated cell counting system. Cells were further diluted with HTPM to the appropriate final volume, and 100 µL was added to each well excluding the outer wells at the final cell seeding densities specified in the Results section. Microplates were then incubated in a 37°C CO₂ incubator. On day 2, cell culture medium was replaced with HTCM containing bovine TSH (Sigma-Aldrich, St. Louis, MO) at the concentration indicated for each data set and maintained in the same medium until day 14 with medium exchange on days 5, 7, 9, and 12.

For H&E staining, microtissues on day 12 were fixed with 10% formalin prior to processing and paraffin embedding. Blocks were mounted and 4-µm cross sections were cut, deparaffined, and stained with Gill 2 Hematoxylin and Eosin Y (Richard Allan Scientific, San Diego, CA) for 2 and 3 min respectively. Slides were dehydrated and imaged using a Ziess Observer Z1 microscope (Zeiss, Dublin, CA).

For CK7 and FSP1 staining, thyrocytes were seeded at 100,000 cells per chamber on an 8-well chamber slide (ThermoFisher Scientific). After 18–24 h, the medium was aspirated and 250 µL of Cytofix/Cytoperm solution (BD Biosciences, San Jose, CA) was added to each well followed by a 30 min incubation at 4°C for cell fixation. Samples were then washed 2 times with 250 µL of 1X Perm/Wash buffer (BD Biosciences). Primary antibodies were added including monoclonal mouse anti-human CK7 (Clone OV-TL, Agilent Dako, Santa Clara, CA) (1:100 dilution) and rabbit polyclonal FSP1 (Abcam, Boston, MA) (1:100 dilution), and incubated overnight at 4°C. After overnight incubation, samples were washed with 1X Perm/Wash buffer followed by the addition of Goat Anti-mouse Alexa 488 IgG (H + L) (1:500) (Life Technologies, Carlsbad, CA) or Goat anti-rabbit Alexa-555 IgG (H + L) (1:500) (ThermoFisher). The chambers were then incubated at room temperature for 1 h in the dark. Samples were washed with 1X Perm/Wash buffer and followed by a wash with 1X DPBS (ThermoFisher Scientific). Finally, two drops of Fluoromount G with DAPI (ThermoFisher Scientific) were added into each chamber prior to imaging. Images were acquired and processed using Ziess Observer Z1 fluorescent microscope with Apotome-2 and Zen Image Acquisition Software.

For actin microfilament staining, thyrocytes were seeded at 7,500 cells per well and cultured as 3D microtissues for 9 days. On day 9, microtissues were fixed with Cytofix/Cytoperm as described above, and Alexa Fluor 488-labeled Phalloidin (ThermoFisher Scientific) (1:200 dilution) was added. After a 2-h incubation period at room temperature in the dark, samples were washed and stained with 300 nM DAPI (ThermoFisher Scientific) for 1 h in the dark at room temperature. The cells were then washed with 1X DPBS, and Fluoromount G with DAPI was added to each sample. Images were acquired and processed as described above using ×100 magnification on the Ziess Observer Z1 fluorescent microscope.

Human TG and T₄ detection and quantification were carried out using the EHTG and the EIAT4C ELISA kits (ThermoFisher Scientific). Assays were performed according to the manufacturer’s instructions using medium collected from cell culture wells at day 7 for TG and day 14 for T₄ ELISA. A CLARIOStar (BMG Labtech) microplate reader was used for data collection. Medium collected for TG assay was diluted at a ratio of 1:200 for 2D cultures and 1:100 to 1:200 for 3D cultures. Medium collected for T₄ assay was undiluted. MARS (version 3.31) software was used for data analysis according to the manufacturer’s manual.

Anti-human Epithelial Cell Adhesion Molecule (EpCAM) (CD326, clone: VU-1D9)- FITC (Stem Cell Technologies, Vancouver, BC), anti-human CD90 (Thy-1, clone: 5E10)- Alexa Fluor 700 (ThermoFisher Scientific), and anti-human CD144 (clone: 55-7H1)- PerCPCy5.5 (BD Biosciences) were used for thyrocyte staining at 1:100 dilution. After viability and cell count measurements were determined post-thaw, cells were centrifuged at 250 x g for 5 min at room temperature, and the supernatant was discarded. Cell pellets were resuspended in 600 µL of BD Pharmingen Stain Buffer (BD Biosciences), and a 100 µL volume was aliquoted into labelled FACS tubes for unstained, FMO, and full stain samples appropriately. UltraComp eBeads^TM Plus Compensation beads (ThermoFisher Scientific) were prepared for single stain compensation controls. The bead vial was vortexed for 30 s and one drop was added into each labeled control FACS tube. Each tube was vortexed for 30 s then incubated for 30 min at 4°C in the dark. An Attune Nxt Flow Cytometer (ThermoFisher Scientific) was used for data acquisition, and De Novo Software–FCS Express 7 RUO Edition (7.12.0005) was used for data analysis. Between 5,000–10,000 events were recorded per single stain control, and ≥50,000 events were recorded per sample. Cells of interest were identified and gated based on forward side scatter area (FSC-A) against the side scatter area (SSC-A) from which single cells were gated using FSC-A and forward side scatter height (FSC-H). Cells positive for the markers of interest were identified based on the gating of each marker against SSC-A. Background fluorescence and gating boundaries were determined using unstained and FMO control tubes.

To morphologically assess the cells, they were imaged on the designated days using a BX41 microscope (Olympus, Tokyo, Japan). ImageJ was used to measure microtissue diameter using images collected using a ×4 objective on a BX41 microscope (Olympus). To ensure the accurate measurement of the microtissues, the scale bar on the image was measured through the software. The diameters of the microtissues were measured in µm from ≥5 images taken on days 1, 2, 5, 7, 9, 12, and 14. For each condition, a minimum of 10 microtissues were measured. The average diameter and standard deviation were calculated using Microsoft Excel.

The inhibitors, 6-propyl-2-thiouracil (CASRN 51–52-5) and Methimazole (CASRN 60–56-0), were obtained from MilliporeSigma (Burlington, MA). Thyrocytes were seeded on Matrigel^®-coated plates at a density of 7,500 cells per well (0.32 cm²) to form microtissues and were treated with 1 mIU/mL bovine TSH on days 2, 5, 7, 9, and 12. For each studied compound, cells were dosed on days 9 and 12 with concentrations ranging from 0.0001 to 10 μM at log intervals. T₄ and ATP levels were measured as described. All measured data were normalized to DMSO vehicle control.

In order to assess whether the reference chemicals posed any toxic effects to the thyrocytes, CellTiter-Glo assay was used to measure ATPase activity of samples on day 14 to determine cell viability. After collecting medium samples for T₄ measurement, backing tape was put on the back of the 3D culture 96-well plate, and 100 µL volume of a 1:1 mixture of CellTiter-Glo^® Luminescent Cell Viability Assay (Promega, Madison, WI) and HTCM was added into each well of the 3D cell culture plate. The plate was covered and shaken for 2 min on an orbital plate shaker at 500 rpm followed by a 10 min incubation in the dark without shaking. Luminescence reading was recorded using a CLARIOStar (BMG Labtech) microplate reader.

Images are shown from representative donor lots. Values were normalized to the number of cells seeded per well. Significance was calculated in MiniTab (State College, PA) using either Student’s t-test between two groups or One-way ANOVA with Tukey post hoc testing when comparing three groups or more to determine statistical significance with 95% confidence and p < 0.05 for statistical significance. GraphPad Prism v.9 (GraphPad Software, Boston, MA) was used to determine EC₅₀ and EC₉₀ values. Four-parameter nonlinear regression fit was used to determine the half-maximal inhibition concentration IC₅₀ using GraphPad Prism v.9. Microsoft Excel was used to determine R² values.

Cryopreserved p’1 primary human thyrocytes were seeded onto Matrigel^®-coated 96-well plates to determine if cells would self-aggregate and form microtissues that resemble their native follicle-like structure in vivo (Figure 1). Cells began to aggregate, starting 24 h after seeding, and distinct, stable individual microtissues formed by day 5 (Figure 1A). These microtissues had a smooth, round shape with clearly defined borders throughout the 14-day culture period. The morphology of these microtissues by day 7 showed distinct follicle-like structures with empty pockets absent of cells, which can be seen from the top and middle sections of H&E stained microtissues on day 12 (Figure 1B). Actin staining of these microtissues using fluorescently-labeled phalloidin showed a similar distribution of cells surrounding an inner lumen devoid of cells. The outer ring of cells exhibited a typical peripheral localization of microfilaments at the outer and inner membrane edges (Figure 1C).

Once microtissues formed, their functionality was measured by determining TG and T₄ secretion. Cells cultured in 2D were used as a control to demonstrate sensitivity to TSH stimulation. Thyrocytes in both 2D and 3D cultures secreted TG when stimulated with TSH (Figure 1D). Cells in 2D culture treated with TSH secreted more TG compared to microtissues in 3D culture (7,181 ± 3,291 vs. 3,658 ± 1,437 ng/mL). Cells in 2D treated with TSH secreted significantly higher levels of TG compared to non-treated cells (327 ± 214 ng/mL). When the corresponding T₄ levels were measured in the medium samples, microtissues in 3D culture treated with TSH secreted significantly higher levels of T₄ compared to 2D cultures, which secreted little to no T₄ with or without TSH stimulation (3.5 ± 1.1 vs. ND ng/mL) (Figure 1E).

Because the formation of 3D microtissues involves the aggregation of primary cells, it was important to determine the types of cells forming them (Figure 2). Flow cytometry was performed on five donor lots to characterize which cell types are present when these cells are seeded in 3D culture (Figure 2A). Three markers of cell lineage were used, including EpCAM, CD90, and CD144 for epithelial, stromal, and endothelial cell lineages, respectively. Four of the five donor lots had ≥80% EpCAM-positive cells. The fifth donor lot, 2217737, had ∼58% EpCAM-positive cells. This donor lot also had the highest number of CD90-positive cells (∼58%). Donor lot 2120653 had the lowest number of CD90-positive cells (∼3.7%). All five donor lots had <5% CD144-positive cells. Immunostaining for CK7 and FSP1, markers of epithelial and stromal cell lineages, was performed on 2D cultures of cells isolated from donor lots 2217737 and 2214495 (Figure 2B). Both donor lots stained positive for CK7, while donor lot 2217737 appeared to have a distinctly higher pattern of FSP1-positive cells compared to donor lot 2214495.

When cells from these donor lots were seeded in 3D culture, each of the lots had microtissues of varying sizes and numbers (Figure 2C). For example, cells from donor lot 2120653 appeared to form the smallest microtissues. Cells from donor lot 2217737 appeared to have the fewest number of microtissues per well compared to the other four donor lots tested along with microtissue size visually comparable to lot 2020525. When the microtissues were stimulated with TSH, cells from donor lots 2120653 and 2217737 secreted the lowest levels of T₄ compared to cells from the other donor lots (3.2 ± 0.3 and 13.2 ± 3.7 ng/10⁶ cells/48 h) (Figure 2D). The T₄ levels from the other donor lots ranged from 53.6 to 39.6 ng/10⁶ cells/48 h.

Microtissue formation in 3D culture is a necessary component for T₄ secretion; therefore, the effect of the number of cells seeded per well on microtissue formation was examined (Figure 3). Cells were seeded at different densities from 24–20,000, 16–15,000, 12–10,000, 8,000, 6,000, and 3,500 cells per well (Figure 3A). Microtissues of varying sizes and shapes formed at the different seeding densities. At the highest seeding density (24–20,000 cells per well), one to two large, non-symmetrical microtissues formed per well, which visually appeared to be much larger than the other microtissues seeded at the lower seeding densities. Microtissues appeared to decrease in size as the seeding density was lowered; however, cells were still able to self-organize into microtissues at the lowest seeding density of 3,500 cells per well in 3D culture.

TG and T₄ production rates were measured from microtissues in 3D culture formed at each seeding density. As the seeding density decreased, the rate of TG production increased from microtissues stimulated with TSH (Figure 3B). However, there was a drop off in the TG production rate (8,079 ± 1,125 ng/10⁶ cells/48 h) at the lowest seeding density of 3,500 cells per well. Cells that were seeded at 6,000 cells per well formed microtissues that secreted the highest level of TG (28,399 ± 21,870 ng/10⁶ cells/48 h). Significantly lower TG levels were measured from microtissues formed from cells seeded at the highest seeding densities of 24–20,000 (5,283 ± 3,538 ng/10⁶ cells/48 h) and 16–15,000 cells per well (8,151 ± 4,224 ng/10⁶ cells/48 h) compared to microtissues seeded at 6,000 cells per well.

A similar trend was seen for T₄ production rates at the different seeding densities (Figure 3C). As the number of cells seeded per well decreased, the rate of T₄ production increased until reaching a maximum at 8,000 cells per well (68.0 ± 6.8 ng/10⁶ cells/48 h). However, a drop off in secretion was seen from microtissues formed at the 6,000 and 3,500 cells per well seeding densities (51.1 ± 16.4 ng/10⁶ cells/48 h and 22.2 ± 3.0 ng/10⁶cells/48 h). Likewise, T₄ production levels were significantly lower from microtissues formed at 24–20,000 cells per well (27.6 ± 5.7 ng/10⁶ cells/48 h) compared to the 8,000 cells per well seeding density. The lowest rate of T₄ production was from microtissues seeded at 3,500 cells per well (22.2 ± 2.9 ng/10⁶ cells/48 h).

As further confirmation of the optimized seeding density, two seeding densities from donor lot 2313606 were tested at seeding densities of 15,000 and 7,500 cells per well. Both seeding densities formed microtissues that visually looked similar in 3D culture (Figure 3D). However, the microtissues formed from cells seeded at the 7,500 cells per well had significantly higher rates of T₄ production compared to the microtissues formed from cells seeded at 15,000 cells per well when stimulated with TSH (105.2 ± 31.28 vs. 62.6 ± 2.8 ng/10⁶ cells/48 h, respectively) after normalizing to seeded cell number (Figure 3E).

Relative differences in size were observed between non-stimulated versus TSH stimulated microtissues in 3D culture (Figure 4). Representative images of microtissues treated without or with TSH on days 5, 7, and 14 are shown (Figure 4A). The results showed that larger microtissues are present when stimulated with TSH. To confirm this, microtissue diameter was determined between the two groups on days 5, 7, 9, 12, and 14 (Figure 4B). Larger microtissues on average were measured starting on day 5 (39.8 ± 14.8 vs. 32.4 ± 8.8 µm), peaking by day 7 (56.0 ± 34.5 vs. 31.7 ± 13.4 µm), and remaining until day 14 (49.6 ± 16.5 vs. 34.6 ± 15.6 µm) when treated with TSH versus non-stimulated. Overall, the average range of diameters is consistently larger in the TSH stimulated microtissues.

Because microtissues increase in diameter due to TSH stimulation, the relationship between microtissue size and T₄ production rate was determined in microtissues from four donor lots seeded at the same density (7,500 cells per well) (Figure 4C). There were visual differences in microtissues between each donor lot. Donor lot 2314417 appeared to form the smallest microtissues, while donor lot 2313606 appeared to form the largest. When microtissue diameter was measured, donor lots 2314417 and 2312660 formed microtissues averaging 36.1 ± 3.7 and 53.8 ± 9.7 µm in diameter (Figure 4D), while larger microtissues were measured from donor lots 2313873 and 2313606 (118.7 ± 36.7 and 143.9 ± 42.5 µm).

T₄ production rate was found to be correlated to microtissue size as microtissue diameter increased, T₄ levels increased proportionally (Supplementary Figure S1A). No detectable quantity of T₄ was measured from donor lot 2314417 where average microtissue diameter was <40 µm. The remaining donor lots all secreted T₄ with donor lot 2312660 having the lowest level (21.5 ± 14.0 ng/10⁶ cells/48 h). Donor lots 2313873 and 2313606 secreted the highest T₄ levels (74.4 ± 9.9 and 105.2 ± 31.3 ng/10⁶ cells/48 h).

TG and T₄ synthesis from microtissues in 3D culture are dependent upon TSH stimulation, but full concentration-response profiles have not been determined in this model system previously. The sensitivity of the microtissues to TSH concentration was determined on both TG and T₄ production rates (Figure 5). For all the data sets, the cell seeding density was 7,500 cells per well. TSH concentrations were tested in half-log intervals between 0.0003 and 3 mIU/mL. There were no significant differences found in TG production when microtissues were stimulated between 0.001 mIU/mL (93.8 ± 37.7 µg/10⁶ cells/mL) to 3 mIU/mL (135.8 ± 33.2 µg/10⁶ cells/mL) TSH (Figure 5A). Significantly lower TG production was measured when microtissues were stimulated with 0.0003 mIU/mL TSH (42.2 ± 11.7 µg/10⁶ cells/mL) compared to the 0.01, 0.03, 0.3, 1, and 3 mIU/mL concentration. The EC₅₀ and EC₉₀ for TG secretion was 0.0008 and 0.003 mIU/mL, respectively. Non-stimulated microtissues (0 mIU/mL) secreted measurable levels of TG (31.7 ± 9.5 µg/10⁶ cells/48 h) that were not significantly different compared to the 0.0003 mIU/mL stimulated levels.

A concentration-dependent increase in T₄ production was observed between TSH concentrations 0.003 and 0.01 mIU/mL (Figure 5B). No significant differences in T₄ levels were measured from microtissues stimulated with TSH between 0.01 and 3 mIU/mL (247.09 ± 2 to 239.75 ± 39 ng/10⁶ cells/48 h). TSH concentration 0.3 mIU/mL (271.71 ± 19 ng/10⁶ cells/48 h) stimulated the highest T₄ levels. The EC₅₀ and EC₉₀ for T₄ secretion was 0.002 and 0.007 mIU/mL, respectively. Microtissues stimulated with TSH concentrations between 0.0003 (14.68 ± 2 ng/10⁶ cells/48 h) and 0.003 (177.73 ± 11 ng/10⁶ cells/48 h) mIU/mL secreted significantly less T₄ compared to microtissues treated with the higher TSH concentration (0.01, 0.03, 0.1, 0.3, 1 and 3 mIU/mL). Although non-stimulated microtissues produced T₄ (17.63 ± 1 ng/10⁶ cells/48 h), the level was not significantly different compared to those measured for the 0.0003 mIU/mL TSH stimulated microtissues.

Data presented previously showed that microtissue diameter increased with TSH stimulation, and T₄ production rates were correlated with a minimum diameter of >40 µm for T₄ synthesis. Higher concentrations of TSH were tested to determine the influence on microtissue size and corresponding T₄ synthesis (Figure 5C). Microtissues were stimulated with either 0.3 or 1 mIU/mL TSH starting on day 2 of culture, and their average diameters were determined on days 5, 7, 9, 12, and 14. Both TSH concentrations caused microtissues to increase in diameter throughout the 14-day culture period. On day 7, microtissues had an average diameter of 32.9 ± 18.2 and 39.7 ± 26.4 µm on day 7 with 0.3 and 1.0 mIU/mL TSH stimulation, respectively. By day 14, TSH 0.3 and 1.0 mIU/mL stimulated microtissues increased in diameter to 42.7 ± 18.4 and 52.4 ± 41.1 µm. Non-stimulated microtissues ranged in diameter from 27.6 ± 11.6 to 35.0 ± 17.3 µm on days 5–14. When T₄ production rates were measured, there was no significant difference in values between 0.3 and 1.0 mIU/mL TSH stimulated microtissues (15.8 ± 2.6 and 17.8 ± 3.2 ng/10⁶ cells/48 h, respectively), which were significantly higher compared to the levels from non-stimulated microtissues (9.0 ± 3.1 ng/10⁶ cells/48 h) (Figure 5D).

Established disruptors of thyroid hormone production were selected to determine the concentration-response profiles in 3D cultures for TDC screening purposes (Figure 6). Inhibition of T₄ synthesis was determined after a 120-h treatment between days 9 and 14 with 6-propyl-2-thiouracil, a TPO and DIO-1 inhibitor (Figure 6A), and methimazole, a TPO inhibitor (Figure 6B), at log concentration intervals between 1.0 × 10⁻¹⁰ to 1.0 × 10⁻⁵ M. No significant difference in viability was observed for either inhibitor (1.0 × 10⁻¹⁰ to 1.0 × 10⁻⁵ M) for donor lots 2021159 and 2021598. The concentration of both inhibitors was not toxic to the cells at the highest concentration of 1 × 10⁻⁵ M (Figures 6C,D). Average half-maximal inhibitory concentrations (IC₅₀) for 6-propyl-2-thiouracil and methimazole were 6.6 × 10⁻⁷ M and 7.7 × 10⁻⁷, 2.2 × 10⁻⁷ and 7.9 × 10⁻⁷ M, respectively (Table 2). No differences in microtissue size were observed between treatment groups and vehicle control (data not shown).