Hypothyroidism Impairs Human Stem Cell–Derived Pancreatic Progenitor Cell Maturation in Mice

Transplantation of cadaveric human β-cells can restore insulin-independence in patients with type 1 diabetes (T1D) (1,2) but is not widely available to most patients due to the inadequate supply of donor cells and burden of immunosuppression. Pluripotent stem cells are a highly scalable alternative cell source (3), and we have previously demonstrated that human embryonic stem cell (hESC)–derived pancreatic progenitor cells can reverse hyperglycemia in mouse models of streptozotocin-induced T1D (4–6) and high-fat diet–induced type 2 diabetes (7). However, glucose-responsive human insulin secretion was only achieved after a lengthy cell maturation period in vivo, and how environmental factors within the host may affect this maturation process remains unclear. ViaCyte Inc. has initiated phase 1/2 clinical trials involving transplant of macroencapsulated hESC–derived pancreatic progenitor cells into patients with T1D. Thus, understanding how variability in the physiology of transplant recipients may affect the development of progenitor cells in vivo is important.

Patients with diabetes have a significantly higher risk of developing thyroid disease than the general population (8). Up to one-third of patients with T1D also have thyroid dysfunction, which can exacerbate the impaired metabolic control and complications associated with diabetes, particularly when the thyroid disorder is undetected (8). Moreover, there is evidence to suggest that excessive or deficient levels of thyroid hormones may affect β-cell development and function. Maternal hypothyroidism caused impaired insulin secretion in neonatal rats as well as glucose intolerance and β-cell dysfunction in adult offspring (9). Systemic knockout of Dio3 (the enzyme required for intracellular inactivation of thyroid hormones) caused significantly reduced islet area and pancreatic insulin content compared with wild-type mice at birth as well as glucose intolerance and impaired insulin secretion during adulthood (10). Triiodothyronine (T3) has also been shown to have prosurvival effects on adult β-cells by protecting mice from streptozotocin-induced β-cell death and diabetes (11). Aguayo-Mazzucato et al. (12) demonstrated that daily T3 injections from postnatal days 1–7 in rats increased expression and nuclear localization of Mafa, a transcription factor essential for β-cell maturation, whereas inhibition of postnatal thyroid hormone synthesis decreased Mafa levels in neonatal rat islets. Moreover, treatment of isolated neonatal rat islets with T3 in vitro induced glucose-stimulated insulin secretion, an effect that was blocked in the presence of dominant-negative Mafa, suggesting that the effects of T3 on β-cell maturation are via Mafa regulation (12). Consistent with these findings, addition of T3 to differentiating hESCs increased gene expression of INS and MAFA and led to improved glucose-stimulated insulin secretion in vitro (13). Taken together, these studies suggest that thyroid hormone signaling may play an important role in β-cell development, maturation, survival, and maintenance of adult β-cell function.

Here, we investigated the effects of thyroid hormone dysregulation on the maturation of encapsulated hESC-derived pancreatic progenitor cells in vivo. We hypothesized that hyperthyroidism may accelerate the development of hESC-derived pancreatic progenitor cells into mature insulin-secreting cells, whereas hypothyroidism may hinder the maturation process. Our findings indicate that chronic hypothyroidism impairs the development of hESC-derived β-cells in vivo and instead promotes the formation of α- and ε-cells from pancreatic progenitor cells. Short-term exposure to hyperthyroidism for 4 weeks posttransplant did not affect the development of glucose-dependent human insulin production from pancreatic progenitor cells in vivo.

The H1 hESC line was obtained from WiCell Research Institute, Inc. (Madison, WI). All experiments at The University of British Columbia (UBC) with H1 cells were approved by the Canadian Stem Cell Oversight Committee and the UBC Clinical Research Ethics Board. Pluripotent H1 cells were differentiated into pancreatic progenitor cells for transplantation studies according to a 14-day, four-stage protocol, as previously described (4). Expression of key pancreatic progenitor cell markers was assessed before transplant using flow cytometry, as previously described (5); antibody information is provided in Supplementary Table 1. To determine the effect of T3 on hESC development in vitro, H1 cells were differentiated according to our recently published protocol (13), and T3 was added during stage (S)4 (500 nmol/L or 1,000 nmol/L final concentration). Differentiated cells were assessed by quantitative (q)PCR on S4 day 3 (S4D3), S5D3, and S6D3, as described below.

Male 7- to 8-week-old SCID-beige mice (C.B-Igh-1b/GbmsTac-Prkdc^scid-Lyst^bgN7; Taconic, Hudson, NY) were maintained on a 12-h light/dark cycle throughout the study. The first cohort of mice (7–8 weeks of age) was used to characterize the different models of thyroid dysregulation: chronic hypothyroid (n = 10), chronic hyperthyroid (n = 10), and euthyroid (n = 10); the treatment protocol is summarized in Fig. 1A and described in detail below. A subset of mice from this cohort was subsequently used for transplantation, as summarized in Fig. 2A. The chronic hypothyroid group was maintained on an iodine-deficient diet (n = 8), and the euthyroid group was divided into two subgroups: euthyroid (group A; n = 5) and acute hypothyroid (n = 4). A second group of mice (8 weeks of age) was used for a new hyperthyroid cohort (Fig. 2A): acute hyperthyroid (n = 8) versus euthyroid (group B; n = 8). Euthyroid groups A and B were analyzed separately for blood glucose and body weight tracking but subsequently combined for all further analysis. All experiments were approved by the UBC Animal Care Committee and performed in accordance with the Canadian Council on Animal Care guidelines.

All mice received ad libitum access to a standard irradiated diet (Teklad Diet #2918; Harlan Laboratories, Madison, WI) to allow for acclimatization after their arrival at UBC. At 12–13 weeks old, mice from the first cohort were randomly selected to undergo one of the following treatments (summarized in Fig. 1A): 1) euthyroid, iodine control diet (cat. #TD.08260; Harlan Laboratories) and normal drinking water; 2) hypothyroid, iodine-deficient diet with 0.15% propylthiouracil (PTU) (cat. #TD.08259; Harlan Laboratories) and normal drinking water; or 3) hyperthyroid, iodine control diet (cat. #TD.08260; Harlan Laboratories) and drinking water containing various concentrations of L-thyroxine sodium salt pentahydrate (T4) (cat. #T2501, Sigma-Aldrich): a) 12 mg/L: 50% PBS, 50% double-distilled [dd]H₂O; b) 6 mg/L: 25% PBS, 75% ddH₂O; or c) 3 mg/L: 12.5% PBS, 87.5% ddH₂O.

For the mice that received cell transplants, five treatment groups were monitored for 180 days posttransplant (summarized in Fig. 2A). Euthyroid mice (groups A and B) and the acute hyperthyroid mice received the iodine control diet for the duration of the study. The acute hyperthyroid group received T4 drinking water (3 mg/L) for 1 week before and 4 weeks after transplantation. The acute hypothyroid mice received the iodine-deficient diet with 0.15% PTU for 4 weeks after transplantation, and the chronic hypothyroid mice received the iodine-deficient diet for the duration of the study.

Euthyroid (group A), chronic, and acute hypothyroid mice received hESC-derived pancreatic progenitor cell transplants at 20–21 weeks of age (after 8–9 weeks of treatment regimens; Fig. 2A). Euthyroid (group B) and acute hyperthyroid mice received cell transplants at 10 weeks of age (Fig. 2A). All mice were anesthetized with inhalable isoflurane, and ∼5 × 10⁶ hESC-derived pancreatic progenitor cells were transplanted subcutaneously within a 20 μL TheraCyte macroencapsulation device (TheraCyte Inc., Laguna Hills, CA) on the right flank, as previously described (4).

All metabolic analyses were performed in conscious, restrained mice, and blood samples were collected via saphenous vein at the indicated time points. Specific assays used to measure plasma analytes and detailed protocols for metabolic testing are described in the Supplementary Data.

TheraCyte devices were harvested at 27 weeks posttransplant from all mice for qPCR analysis. Devices were cut longitudinally, and one-half was preserved in RNAlater Stabilization Solution (Life Technologies, Carlsbad, CA) and stored at −80°C until use. The other half of each device was stored in 4% paraformaldehyde, as described below. The procedure for isolating RNA from engrafted tissue within devices and qPCR analysis has been described in detail elsewhere (7). Data were analyzed using Expression Suite 1.0.3 software (Thermo Fisher Scientific/Life Technologies) and normalized to adult human islets (n = 2 donors) using the ΔΔCt method. Gene expression in cultured hESC-derived cells was assessed as previously described (5) (n = 2 biological replicates per condition). A list of primers is provided in Supplementary Table 2.

Engrafted hESC-derived cells (half of TheraCyte device), pancreas tissue, and thyroid glands were harvested at 27 weeks posttransplant, fixed in 4% paraformaldehyde, and stored in 70% ethanol before paraffin-embedding. All paraffin sections (5-μm thickness) were prepared by Wax-it Histology Services (Vancouver, British Columbia, Canada). Immunofluorescent staining and imaging were performed as previously described (14), and primary antibody information is provided in Supplementary Table 3. Refer to Supplementary Data for details about quantification of the endogenous pancreas and engrafted tissue.

All statistics were performed using GraphPad Prism software (GraphPad Software Inc., La Jolla, CA). Details about individual statistical tests are provided in the Supplementary Data. For all analyses, P < 0.05 was considered statistically significant. Data are presented as mean ± SEM with individual data points.

Our first goal was to establish a protocol to induce hyperthyroidism or hypothyroidism in immunodeficient SCID-beige mice (outlined in Fig. 1A). Initially, plasma T3 levels were similar between groups, but T3 levels were significantly reduced after 14 days of feeding with an iodine-deficient diet compared with an iodine control diet (euthyroid) (Fig. 1B). In contrast, the addition of T4 to drinking water produced dose-dependent increases in plasma T3 levels. A dosage of 12 mg/L for 7 days, followed by 6 mg/L for 7 days, produced plasma T3 levels in the hyperthyroid mice that were ∼10- to 15-fold higher than the euthyroid group (Fig. 1B) but much higher than T3 levels reported in humans with clinical hyperthyroidism (euthyroid: 3.3 ± 0.7 ng/mL; hyperthyroid: 7.1 ± 1.0 ng/mL [15]). Therefore, the dosage of T4 was further reduced to 3 mg/L to achieve clinically relevant plasma T3 levels (Fig. 1B) and in an effort to prevent the severe hypoglycemia observed with higher T4 dosages (Fig. 1D and E).

Mice displayed an initial reduction in body weight after administration of the iodine-deficient diet, which recovered and stabilized after 12 days of treatment but remained significantly lower than euthyroid controls (Fig. 1C). Hyperthyroidism had no effect on body weight (Fig. 1C). Hypothyroidism caused significantly elevated blood glucose levels under fasting conditions (Fig. 1D) and after an oral glucose challenge (Fig. 1E), whereas hyperthyroidism resulted in significantly decreased blood glucose levels (Fig. 1D and E). Hyperthyroidism was also associated with significantly higher fasting plasma insulin levels compared with euthyroid controls (Fig. 1F). Unfortunately, after 24 days of T4 administration, 4 of 10 hyperthyroid mice died, and the remaining mice were consequently switched to normal drinking water, resulting in return to normoglycemia within 2 weeks (Fig. 1D).

We next assessed the effects of excessive or deficient thyroid hormone levels on the development of hESC-derived pancreatic progenitor cells in vivo. After the 14-day differentiation in vitro, hESC-derived cells were ∼99.5% PDX1⁺ and 70% NKX6.1⁺ before transplant (Supplementary Fig. 1), consistent with previous studies (5–7,13). The progenitor population also contained ∼14% endocrine cells, which coexpressed NKX2.2, but were largely NKX6.1⁻ (Supplementary Fig. 1). These pancreatic progenitor cells were transplanted subcutaneously within TheraCyte devices into mice with thyroid hormone dysregulation (as outlined in Fig. 2A). Our experimental groups enabled us to examine the effect of acute exposure to deficient or excessive thyroid hormone during the first 30 days posttransplant and chronic thyroid hormone deficiency for the duration of the study. We chose not to include a chronic hyperthyroid treatment group because even the lowest dosage of T4 tested (3 mg/L) caused dangerous hypoglycemia and death (Fig. 1D and E). To validate the efficacy of treatments, plasma T3 levels were measured at 28 days posttransplant (during the acute intervention) and were confirmed to be significantly lower in the acute hypothyroid mice and significantly higher in the acute hyperthyroid group compared with euthyroid controls (Fig. 2B). On day 42 (12 days after the cessation of the acute thyroid interventions), T3 levels were normal in the acute hyperthyroid and hypothyroid mice but remained significantly lower in the chronic hypothyroid mice, reflecting their ongoing treatment with iodine-deficient diet (Fig. 2B). As further validation of the model, the thyroid gland was examined at the end of the study. Thyroid weight (normalized to body weight) was significantly increased in both the chronic and acute hypothyroid groups compared with the euthyroid group, although this was most pronounced in the chronic group (Supplementary Fig. 2A). Moreover, we observed severe thyroid follicular atrophy and absence of colloid in the thyroid gland of mice with chronic hypothyroidism (Supplementary Fig. 2B).

Consistent with the first study (Fig. 1), acute hypothyroidism resulted in a transient decrease in body weight, which quickly recovered once the mice were taken off the iodine-deficient diet at 30 days posttransplant (Fig. 2C). The initial weight loss in the chronic hypothyroid group plateaued after ∼80 days, and body weight remained stable thereafter (Fig. 2C). The reduction in body weight was associated with a significant decrease in fat pad weight (relative to body weight) in the chronic hypothyroid mice at 27 weeks posttransplant (Supplementary Fig. 3). Chronic hypothyroidism also led to elevated fasting blood glucose levels, whereas acute hypothyroid mice remained normoglycemic throughout the study (Fig. 2C). Acute hyperthyroidism had no effect on body weight but caused a transient decrease in blood glucose levels during the T4 treatment period (Fig. 2D), consistent with the previous cohort of mice (Fig. 1D). From this point forward, all data from the two euthyroid groups were pooled because their body weight and blood glucose levels were not significantly different.

To assess the development of the hESC-derived grafts under conditions of thyroid dysregulation, human C-peptide and blood glucose levels were measured in response to various secretagogues after transplant. Chronic hypothyroid mice exhibited elevated blood glucose levels during an oral mixed-meal challenge at all ages examined (Fig. 3A and C), as well as during intraperitoneal glucose (Fig. 3D) and arginine (Fig. 3G) tolerance tests at 22 and 24 weeks posttransplant, respectively. Mice with acute hyperthyroidism had decreased blood glucose levels after the meal challenge at 4 weeks posttransplant (during their T4 treatment period; Fig. 3A) but mild hyperglycemia during the meal and glucose challenges at 8 and 22 weeks posttransplant, respectively (Fig. 3C and D). Acute hypothyroidism did not affect glycemia during any metabolic challenge (Fig. 3A, C, D, and G).

Beginning at 8 weeks posttransplant and persisting throughout the study, the chronic hypothyroid mice had significantly decreased human C-peptide levels compared with euthyroid control mice under fasting conditions (data not shown) and at 40 min after the oral mixed meal (Fig. 3B). In addition, mice with acute hypothyroidism had a transient decrease in meal-stimulated human C-peptide levels compared with euthyroid mice at 16 weeks posttransplant, which was recovered by 25 weeks (Fig. 3B). At 22 weeks posttransplant, chronic hypothyroid mice exhibited severely blunted human C-peptide secretion during an intraperitoneal glucose challenge (Fig. 3E) and also displayed altered C-peptide secretion kinetics compared with euthyroid mice (Fig. 3F). Peak human C-peptide levels were observed at 60 min after glucose administration in the chronic hypothyroid mice (∼2.5 times higher than basal), whereas the other groups had reached similar peak C-peptide levels at 30 min and were approaching basal levels by 60 min (Fig. 3F). The acute hyperthyroid group also had significantly reduced human C-peptide levels at 30 min after the glucose injection (Fig. 3E), but unlike the chronic hypothyroid group, they exhibited human C-peptide secretion kinetics similar to those of euthyroid control mice (Fig. 3F). Consistent with the meal and glucose challenges, the chronic hypothyroid group also displayed significantly reduced human insulin levels at fasting and 15 min after an arginine injection compared with euthyroid mice at 24 weeks posttransplant (Fig. 3H). Interestingly, these mice also had higher plasma glucagon levels postarginine (Fig. 3I) as well as elevated GLP-1 levels at fasting and postarginine (Fig. 3J) compared with euthyroid mice.

TheraCyte devices were harvested at 27 weeks posttransplant for qPCR and histology analysis. Although the chronic hypothyroid group displayed decreased plasma human insulin and increased plasma glucagon levels (Fig. 3H and I), INS and GCG mRNA levels in these hESC-derived grafts were not significantly different from the euthyroid group (Fig. 4). However, chronic hypothyroidism resulted in significantly increased levels of SST, GHRL, and ISL1, as well as a pronounced reduction in IAPP and G6PC2 mRNA in hESC-derived grafts compared with grafts from euthyroid controls (Fig. 4). Acute hyperthyroidism also caused a mild but significant reduction in IAPP and G6PC2 and elevated GCG mRNA levels relative to euthyroid grafts (Fig. 4).

Given that the most pronounced effects on graft function and gene expression were a result of chronic hypothyroidism (Figs. 3 and 4), we focused our detailed characterization of graft composition on the chronic hypothyroid versus euthyroid mice. Thyroid hormone deficiency did not affect the overall formation of endocrine (synaptophysin⁺) or ductal (CK19⁺) cells from hESCs (Fig. 5A) but appeared to shunt the cells toward α-cells at the expense of β-cells (Fig. 5B and C). Indeed, grafts from chronic hypothyroid mice contained a significantly lower fraction of insulin⁺ cells and more than twice as many glucagon⁺ cells compared with euthyroid grafts (Fig. 6A). Thus, there was a significantly higher ratio of glucagon-to-insulin immunoreactive cells in hESC-derived grafts from chronic hypothyroid compared with euthyroid mice, whereas the glucagon-to-insulin ratio in the endogenous pancreas was not affected by chronic hypothyroidism (Fig. 6B). Moreover, thyroid hormone deficiency did not affect the β-cell area per islet in the endogenous pancreas (Fig. 6C). At 27 weeks posttransplant, proliferating cell nuclear antigen⁺ (PCNA) endocrine cells were rare, and there were no obvious differences in the number of proliferating insulin⁺ or glucagon⁺ cells between treatment groups (Supplementary Fig. 4). Despite the significant increase in SST transcript levels in chronic hypothyroid grafts (Fig. 4), there was no significant difference in the fraction of somatostatin⁺ cells among groups (quantification not shown; Supplementary Fig. 5). Pancreatic polypeptide immunoreactivity was rare in grafts and did not appear to differ between groups, although this was not quantified (Supplementary Fig. 5).

Interestingly, chronic hypothyroid grafts had approximately an eightfold increase in GHRL transcript levels (Fig. 4) and a significantly higher proportion of ghrelin⁺ cells relative to DAPI⁺ cells (Figs. 5C and 6A) compared with euthyroid grafts. Moreover, while ghrelin⁺ cells were abundant in the hESC-derived engrafted tissue, they were only rarely detected in the endogenous pancreas (Fig. 6D). Chronic hypothyroid grafts contained approximately equal proportions of ghrelin⁺ and insulin⁺ cells, whereas euthyroid grafts had approximately four insulin⁺ cells for every one ghrelin⁺ cell (Fig. 6E). In the pancreas, the ratio of ghrelin-to-insulin immunoreactive cells was not affected by exposure to chronic hypothyroidism (Fig. 6E). The difference in circulating acylated (active) or unacylated (inactive) ghrelin levels among the groups was not significant, but the ratio of unacylated-to-acylated ghrelin was approximately two times higher in the chronic hypothyroid mice than in the euthyroid mice (Fig. 6F).

Chronic hypothyroidism resulted in a reduced number of hESC-derived β-cells (Figs. 5 and 6), which may explain the decreased absolute human C-peptide levels compared with euthyroid mice (Fig. 3B, E, and H). However, the disrupted human insulin secretion kinetics in chronic hypothyroid mice (Fig. 3F) suggested that the maturation status of individual hESC-derived β-cells may also be altered by thyroid hormone deficiency. On the basis of evidence that islet Mafa expression was regulated by thyroid hormones in neonatal rats (12), we examined expression of MAFA in grafts and the endogenous mouse pancreas. Although MAFA transcript levels were not affected in whole grafts (Fig. 4), we observed substantial heterogeneity in nuclear MAFA immunoreactivity within the hESC-derived insulin⁺ population from chronic hypothyroid mice but no difference in MAFA expression within the endogenous pancreas of hypothyroid versus euthyroid mice (Fig. 7A). Similarly, hypothyroidism also caused decreased NKX2.2 immunoreactivity in the grafts, but no differences were observed among the groups in the endogenous pancreas (Fig. 7B). In addition, there was less amylin immunoreactivity within the insulin⁺ cell population in the hypothyroid compared with euthyroid grafts (Fig. 7C), confirming the significant reduction in overall gene expression of IAPP in hypothyroid grafts (Fig. 4); amylin immunoreactivity was not affected in the β-cells from the endogenous pancreas (Fig. 7C).

The hESC-derived pancreatic progenitor cells were also treated in vitro with T3 during S4 to determine if the effects on graft maturation in vivo might be a result of direct action by thyroid hormones. Consistent with our in vivo study, we observed reduced GHRL, GCG, and ARX mRNA during S5 after treatment with 500 nmol/L or 1,000 nmol/L T3 during S4 (Supplementary Fig. 6). T3 treatment also increased mRNA levels of INS and mature β-cell markers, G6PC2, IAPP, and MAFA, during S6 (Supplementary Fig. 6).

The high incidence of thyroid disease in patients with T1D (8) means that hESC-derived pancreatic progenitor cells transplanted into these patients may be exposed to abnormal thyroid hormone levels in vivo. Notably, we found that chronic thyroid hormone deficiency had a detrimental effect on hESC-derived β-cell development and was associated with higher numbers of hESC-derived α- and ε-cells. Beginning at 8 weeks posttransplant and persisting throughout the study, grafts from chronic hypothyroid mice secreted less than half as much human C-peptide as grafts from euthyroid mice. Blunted human C-peptide secretion was also observed in mice acutely exposed to hypothyroidism, but this effect was transient and fully recovered at 25 weeks posttransplant. Chronic hypothyroid mice also displayed impaired glucose-stimulated insulin secretion kinetics, suggesting that the maturation status of β-cells was affected by thyroid hormone deficiency. The elevated plasma glucagon and GLP-1 levels in chronic hypothyroid mice pointed to preferential formation of α-cells in hESC-derived grafts exposed to thyroid hormone deficiency. Indeed, grafts harvested from chronic hypothyroid mice contained approximately three times fewer insulin⁺ cells and more than twice as many glucagon⁺ cells as grafts from euthyroid mice. It is unclear from these studies whether thyroid hormone deficiency altered the lineage commitment of differentiating endocrine cells, thus resulting in a fate-switch toward α-cells and ε-cells at the expense of β-cells, or whether perhaps thyroid hormone is required for survival and/or expansion of newly differentiated β-cells. T3-treated neonatal rats had increased β-cell proliferation but no measureable change in β-cell apoptosis relative to control rats (12). Proliferation of hESC-derived endocrine cells was not affected by thyroid hormone deficiency in our study after 27 weeks, but it is possible that impaired β-cell replication and/or increased β-cell apoptosis may have occurred in hypothyroid grafts at an earlier time point.

Thyroid hormone deficiency resulted in heterogeneous protein expression of nuclear MAFA in hESC-derived grafts as well as reduced NKX2.2 and amylin levels, important markers of mature β-cells. IAPP (amylin) and G6PC2 mRNA levels were also significantly reduced in grafts from hypothyroid mice compared with euthyroid controls, whereas the observed differences in insulin⁺, glucagon⁺, and MAFA⁺ cells were not reflected at the level of gene expression. Because mRNA levels represent the whole population of hESC-derived cells (including all endocrine cell types, ductal cells, etc.), examining protein expression in individual insulin⁺ cells by immunofluorescent staining is a more accurate assessment of the hESC-derived β-cell phenotype. The various lines of evidence pointing to a β-cell deficiency in hypothyroid mice are consistent with our previous observations in nude rats implanted with hESC-derived progenitor cells (16). Interestingly, nude rats had significantly higher circulating T3 levels than SCID-beige mice and also had improved glucose-stimulated human insulin secretion, a higher proportion of insulin-to-glucagon immunoreactivity in grafts, and more consistent nuclear MAFA expression in hESC-derived β-cells than SCID-beige mice (16). Neonatal rats with hypothyroidism also exhibited decreased islet Mafa expression (12), which is consistent with our findings and suggests a potential role for thyroid hormone in regulating β-cell maturation. This is further supported by evidence that thyroid hormones bind directly to the thyroid hormone response elements in the Mafa promoter to induce Mafa expression and promote glucose-stimulated insulin secretion in immature neonatal rat islets (12).

Ghrelin production by hESC-derived grafts also proved to be interesting in this study. Grafts from euthyroid mice contained substantially higher (200 times) ghrelin mRNA levels compared with human islets and a 4:1 ratio of insulin-to-ghrelin immunoreactive cells, whereas ghrelin⁺ cells were exceedingly rare (<0.5%) in the adult human and mouse pancreas. This discrepancy was even further amplified in the chronic hypothyroid grafts, which contained ∼1,250 times more ghrelin mRNA than human islets and an ∼1:1 ratio of insulin-to-ghrelin immunoreactive cells. In the human fetal pancreas, ghrelin⁺ cells constitute a relatively high proportion (∼10%) of endocrine cells (17). Therefore, the high ghrelin levels in chronic hypothyroid grafts could reflect graft immaturity or may simply indicate a shift in lineage commitment during development. Grafts from chronic hypothyroid mice also contained less NKX2.2 immunoreactivity, which was previously shown to be required for specification and maintenance of β-cell fate (18). Moreover, the absence of NKX2.2 in mice resulted in a dramatic expansion of ghrelin-producing cells at the expense of β-cells (18). Interestingly, patients with hypothyroidism were reported to have elevated serum ghrelin levels (acylated and unacylated), which was normalized by T4 treatment (19). Although the chronic hypothyroid mice in our study did not produce elevated circulating ghrelin levels, there was clearly an effect of thyroid hormone deficiency to induce ghrelin locally within hESC-derived grafts. Pancreatic ghrelin serves as a local regulator of insulin release even though it may not contribute to the circulating ghrelin levels (20). Therefore, ghrelin may be acting in a paracrine manner to impair insulin secretion in the hESC-derived grafts from chronic hypothyroid mice.

Although thyroid hormone deficiency caused clear effects on the transplanted human pancreatic progenitor cells, the endogenous pancreas was largely unaffected in our study. It is possible that developing endocrine cells are more susceptible to thyroid hormone dysregulation than adult islet cells or that human pancreatic cells are more susceptible than mouse cells, although we suspect the former is more likely. It is also possible that the effects on the differentiation of hESC-derived cells are due to the profoundly disrupted metabolic control in the mice with hyper- or hypothyroidism. However, we previously observed efficient graft maturation and robust human C-peptide production in the setting of chronic hyperglycemia (4) and thus the impaired glycemic control in hypothyroid mice is unlikely to be the underlying cause of impaired maturation in the current study. Thyroid hormones are also known to profoundly affect the stress response in vivo (12), which could also indirectly mediate the observed effects of hypothyroidism on β-cell maturation. Interestingly, dexamethasone treatment of neonatal rat islets blocked the T3-mediated induction of Mafa mRNA and glucose-stimulated insulin secretion ex vivo (12). Our previous discovery that T3 treatment at late stages of hESC differentiation enhanced β-cell maturation in culture (13) supports the notion that the hESC-derived grafts may be directly affected by thyroid hormone levels. Moreover, when S4 pancreatic progenitor cells were treated with T3 in vitro, we observed a transient decrease in GHRL, GCG, and ARX mRNA at S5, and elevated INS, G6PC2, IAPP, and MAFA mRNA levels at S6 of in vitro differentiation. These data are consistent with our in vivo study, in which reduced thyroid hormone levels caused increased α- and ε-cell formation along with impaired β-cell maturation in hESC-derived grafts, including reduced G6PC2 and IAPP mRNA. Further studies are required to determine whether thyroid hormones are acting directly on developing human β-cells or whether the indirect effects of thyroid hormones on metabolism and/or the hypothalamic-pituitary-adrenal axis may be mediating the observed changes in β-cell development in vivo.

Clinically, hyperthyroidism can lead to elevated blood glucose levels, possibly due to enhanced gluconeogenesis and increased absorption of sugars from the intestine (21), whereas the opposite typically occurs with hypothyroidism. Interestingly, we observed contradictory effects of thyroid hormones on blood glucose homeostasis in SCID-beige mice. Hyperthyroidism led to hypoglycemia, whereas hypothyroidism caused hyperglycemia compared with euthyroid controls. Hypothyroidism also caused decreased body weight in mice, whereas in humans, hypothyroidism generally leads to weight gain. Although our data differ from the clinical situation, these findings are consistent with previous studies examining hyperthyroidism and hypothyroidism in rodents. For example, rodents that received subcutaneous T3 injections had lowered fasting glucose levels compared with controls (12,22), and hypothyroidism in rats (induced via thyroidectomy and PTU or methimazole treatment) caused decreased body weight (12,23) and increased serum glucose levels (23). The differences between human and rodent thyroid physiology that account for these opposing phenotypes are currently unknown.

Stem cell–derived pancreatic progenitor cells are currently being transplanted into patients with T1D in a phase 1/2 clinical trial with the ViaCyte VC-01 cell product. Our findings raise the possibility that the maturation and ultimate function of the transplanted cells can be influenced by the hormonal and metabolic milieu of the cells. Specifically, we recommend that eligible patients are screened for thyroid dysfunction and treated accordingly to reduce the risk of off-target cell differentiation leading to compromised graft function. Alternatively, hESCs could be differentiated to a more mature stage of development in vitro, as we have recently described (13), to minimize the maturation period after transplantation. It is possible that more mature hESC-derived cells will be less susceptible to altered levels of thyroid hormones because they are closer to adult cells, although this remains to be examined.

Acknowledgments. The authors thank Diana Rosman-Balzer (Janssen R&D, LLC) for her technical assistance with qPCR experiments and Ali Asadi (UBC Department of Cellular & Physiological Sciences) for his assistance with histology.

Funding. This work was funded by the Canadian Institutes of Health Research (CIHR) Regenerative Medicine and Nanomedicine Initiative, the Stem Cell Network, and JDRF. J.E.B. was funded by a JDRF postdoctoral fellowship, a Canadian Diabetes Association (CDA) postdoctoral fellowship, the CIHR Transplantation Training Program, and a L'Oréal Canada for Women in Science Research Excellence Fellowship. N.S. received funding from the Stem Cell Network.

Duality of Interest. P.A. and A.R. are employees of Janssen R&D, LLC, and A.R. is also a shareholder. T.J.K. received financial support from Janssen R&D, LLC, for the research described in this article. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.E.B. and N.S. wrote the manuscript. J.E.B., A.R., and T.J.K. contributed to conception and design of experiments. J.E.B., N.S., S.O., J.K.F., M.M., P.A., A.R., and T.J.K. were responsible for acquisition, analysis, and interpretation of data; contributed to the manuscript revisions; and approved the final version of the manuscript. T.J.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.