Aberrant Thyroid-Stimulating Hormone Receptor Signaling Increases VEGF-A and CXCL8 Secretion of Thyroid Cancer Cells, Contributing to Angiogenesis and Tumor Growth Free

Thyroid cancer is the most common endocrine malignancy, and its incidence has robustly increased over the past 4 decades (1). Although most thyroid cancers are well-differentiated carcinomas, which are slow to progress and have an excellent prognosis, up to 30% of patients experience recurrence within 10 years (2, 3). To prevent recurrence, the long-term use of levothyroxine (T4) to suppress the endogenous pituitary excretion of thyroid-stimulating hormone (TSH), so-called T4 suppression therapy, has been generally considered for thyroid cancer patients at intermediate or high risk for recurrence (4–7).

TSH is a well-known growth factor for thyrocytes that leads to TSH-dependent growth of the thyroid gland, and higher serum TSH concentrations have been found to increase the oncogenic risk or tumor aggressiveness in well-differentiated thyroid cancer (WDTC; refs. 8–10). In contrast, the suppression of serum TSH concentration by administering levothyroxine was found to improve relapse-free and overall survival in high-risk patients with WDTC (11, 12), which supports the proposal that TSH suppression exerts inhibitory effects on the progression of WDTC. Thus, T4 suppression therapy has long been used as an essential therapeutic strategy for the treatment of WDTC, especially in high-risk cases (4–7).

However, T4 suppression therapy should be considered after weighing the potential risks and benefits of TSH suppression (4–7), because long-term, aggressive suppression of TSH could be related to systemic adverse events, such as the exacerbation of ischemic heart disease, as well as an increased risk of atrial fibrillation (13) and osteoporosis (14, 15). The molecular characteristics of WDTC cells have been found to undergo alterations during disease progression; the loss of the TSH receptor (TSHR) and/or sodium/iodide symporter (NIS) is often observed in poorly differentiated thyroid cancers (PDTC; ref. 16). Subsequently, these changes attenuate the iodine uptake potential, resulting in iodine-refractory incurable diseases. In in vitro experiments, the response to TSH stimulation measured in terms of cAMP levels (17) or iodine uptake (18) was also markedly decreased in some thyroid cancer cell lines, and these cancer cells often showed TSH-independent tumor growth (19). Therefore, it is reasonable to deduce that the role of T4 suppression therapy might be negligible in this setting.

Meanwhile, several pieces of evidence have shown that TSH not only regulates tumor cell growth, but also simulates cytokine production in tumor cells, which could affect the tumor microenvironment (19–21). These findings led us to hypothesize that TSH would influence the tumor microenvironment via the paracrine secretions of tumor cells, which could potentially be associated with T4 suppression therapy having advantages not limited to the suppression of TSH-dependent cell growth. Moreover, TSH also binds to extrathyroidal cells including fibroblast or endothelial cells (22, 23), which are essential members of tumor stroma, and may have potential for directly modulating tumor microenvironments. The aim of this study was to investigate the role of TSH on the tumor microenvironment and its effects on tumor growth and progression.

To identify the effects of TSH on tumor growth and angiogenesis, the ectopic tumor models using PDTC cells harboring TSHR or a breast cancer cell line which did not express TSHR were treated with TSH. To obtain a mechanistic insight, proangiogenic effects of conditioned medium (CM) of TSH-treated PDTC cells were studied using human microvascular endothelial cells. Angiogenic factors were screened using cytokine array, and VEGF-A and CXCL8 were increased. To explore whether these factors mediated TSH effects on PDTC tumor angiogenesis, neutralizing antibodies were administered to TSH-treated ectopic tumors in vivo. Finally, clinical relevance was studied in PDTC tumor microarray with IHC.

Recombinant human TSH (rhTSH; Thyrogen) from Genzyme Therapeutics, MMI from Sigma-Aldrich, bevacizumab (anti-VEGF; Avastin) from Roche, and clodronate liposomes from FormuMax Scientific Inc. were used. Anti–VEGF-A antibody, anti-CXCL8 antibody, and ELISA kits for VEGF-A and CXCL8 were purchased from R&D Systems. The inter- and intra-assay coefficients of variation were 4.1% and 6.7% for VEGF-A and 4.6% and 6.7% for CXCL8. The linear ranges of VEGF-A (27 kDa) and CXCL8 (11 kDa) were 7.8 to 1,000 pg/mL and 15.6 to 2,000 pg/mL, respectively.

To investigate effects of TSH in PDTC, two PDTC cell lines, BHP10-3SCp and FRO cells, were used. The BHP10-3SCp cell line was established as a tumorigenic subclone of BHP10-3 cells, a human papillary thyroid cancer (PTC) cell line harboring the RET/PTC rearrangement that was kindly provided by Dr. Soon-Hyun Ahn (Seoul National University College of Medicine, Seoul, Republic of Korea) and Dr. Gary L. Clayman (MD Anderson Cancer Center, Houston, TX; ref. 24). The original BHP10-3 cell line was authenticated by the short tandem repeat typing method in April 2015 and verified again after being established as a tumorigenic subclone in July 2015. The last authentication was performed in June 2016. The FRO cell line and two PTC cell lines, TPC-1 and BCPAP, were kindly provided by Dr. June-Key Chung (Seoul National University College of Medicine, Seoul, Republic of Korea). A human triple-negative breast cancer cell line MDA-MB-231, a kind gift from Dr. Hyo Soo Kim (Seoul National University College of Medicine, Seoul, Republic of Korea), was used to distinguish the direct effects of TSH on thyroid cancer cells and the surrounding endothelial cells.

BHP10-3SCp cells were seeded on 100-mm culture dishes with 3 × 10⁵ cells/mL with complete RPMI-1640 medium containing 10% FBS. When cells reached a >90% confluent monolayer, 20 ng/μL of rhTSH or saline was administered in serum-free medium. After 24 hours, the CM was harvested from each cell culture, filtered, and stored at −70°C. CMs from rhTSH-treated BHP10-SCp cells were referred to as TSH-CM/B, and saline-treated CMs from BHP10-SCp cells were referred to as control-CM/B. An angiogenesis antibody array (R&D Systems) was used to test the protein expression of angiogenic cytokines in TSH-CM/B.

Five-week-old female BALB/c nu/nu mice were purchased from Orient Bio Inc. and housed in a dedicated pathogen-free facility. The Institutional Animal Care and Use Committee of Seoul National University approved the protocols for the animal study (no. SNU-141224-2). All animals were maintained and used in accordance with the Guidelines for the Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, Seoul National University.

BHP10-3SCp (5 × 10⁶/50 μL of PBS), FRO (2 × 10⁶/50 μL of PBS), or MDA-MB-231 (2 × 10⁶/50 μL of PBS) cells were mixed with 50 μL of Matrigel (BD Biosciences) and implanted subcutaneously in the flank area of each mouse. Five days after tumor cell injection, when visible tumors began to appear, rhTSH was started to administer intraperitoneally (5 days/week) until day 28, and saline (0.9% NaCl) was used for the controls (n = 8 in each group). In the BHP-10-3SCp tumor mouse model, to determine the dose-dependent effect of TSH, rhTSH were administered at two different doses of 0.15 μg/g (low-TSH) and 1.5 μg/g (high-TSH), which were considered to lead to the physiologic and supraphysiologic responses in mouse (25). Because the high-TSH group showed more obvious difference from the control than the low-TSH group in the BHP-10-3Sp tumor model, in the FRO and MDA-MB-231 models, high doses (1.5 μg/g) of rhTSH were used. Tumor size was measured at least twice per week, and tumor volume was calculated using the following equation: volume = ½ × a × b², where a = the long tumor diameter and b = the short tumor diameter (26). On day 28, all mice were euthanized, and blood was collected 24 hours after the last injection of rhTSH or saline. For experiments involving treatment with MMI, MMI was dissolved in water (0.025%) and applied to mice from day 7 to day 28 after tumor cell injection. Tap water was used in the controls (n = 8 in each group).

In the BHP10-3SCp tumor mouse model, to confirm the effects of VEGF-A and CXCL8 induced by TSH, anti-VEGF (1 μg/g, 3 days/week) and anti-CXCL8 (50 μg/g, once a week) were i.p. injected in the TSH group (high-TSH; n = 8 in each group). For the in vivo macrophage depletion study, clodronate liposomes (35 μg/g, 2 days/week) or control liposomes were i.p. injected in both the control and TSH groups (n = 8 in each group). All animal experiments were performed at least twice independently.

Serum murine TSH levels were measured using a TSH ELISA kit (MPTMAG-49K, Millipore) according to the manufacturer's instructions. Serum-free T4 and human TSH levels were measured using an immunoradiometric assay kit (IRMA, Shin Jin Medics Inc.) following the manufacturer's instructions. Serum TSH and free T4 levels were measured in mouse serum samples obtained 24 hours after the last injection of rhTSH.

The tumors from xenograft models were fixed in 4% paraformaldehyde and embedded in paraffin. IHC staining was performed using anti-mouse CD31 (1:100, Santa Cruz Biotechnology), ERG (predilute, Ventana Medical Systems), VEGF-R2 (1:100, Cell Signaling Technology), F4/80 (1:200, eBioscience), and anti-human Ki-67 (1:500, Neomarkers).

Human microvascular endothelial cells (HMVEC), purchased from Lonza, were cultured in 0.5% BSA/EBM-2 medium (Lonza) and used for assays. More details are in Supplementary Materials and Methods.

BHP10-3SCp cells were treated with rhTSH (20 ng/μL), and the cell lysates were harvested at 0, 15, 30, and 120 minutes. The primary antibodies were pAKT (1:2,000, #4058), AKT (1:2,000, #9272), pERK (1:2,000, #4370), and ERK (1:2,000, #9102) from Cell Signaling Technology, and β-ACTIN (A5441) from Sigma-Aldrich. More details are in Supplementary Materials and Methods.

After 6 hours of starvation, BHP10-3SCp cells were pretreated with TSH receptor antibody or inhibitors of PI3K/mTOR and ERK at 2 hours after rhTSH administration (20 ng/μL). K1-70 (100 ng/μL, RSR Ltd.) was used as a TSH receptor blocking antibody. LY294002, RAD001, and PD98059 (10 μmol/L for all, Sigma-Aldrich) were used to inhibit PI3K, mTOR, and ERK activity, respectively. Seventy-two hours after the treatment, the cells were washed, and the mRNA was harvested with Trizol (Invitrogen). RT-PCR was done using an ABI Prism 7500 Sequence Detection System (Applied Biosystems). The sequences of the primers are listed in Supplementary Table S1.

Paraffin-embedded PDTC samples from 13 patients and advanced PTC samples defined as ≥2 cm-sized tumors from 35 patients were used for constructing tissue microarray as previously described (27). IHC staining of the tissue microarrays was performed using anti-human VEGF-A (1:30, R&D systems), CXCL8 (1:500, Novus Biologicals), CD31 (1:1,000, Cell Signaling Technology), ERG (predilute, Ventana Medical Systems), and CD163 (1:1,000, Cell Signaling Technology). All images were analyzed using the publicly available programs Image J (28), Immunoratio (29), and ImageScope (Aperio). All experiments were approved by the Institutional Review Board of Seoul National University Hospital (1107-060-369).

For statistical comparisons, the Mann–Whitney test or the two-tailed Student t test was used for two groups, and the Kruskal–Wallis test with the Bonferroni correction was used for three groups. Comparisons of categorical variables were performed with either Pearson χ² test or Fisher exact test. Relationships between each IHC marker in human PDTC samples were evaluated by the Pearson correlation coefficient. Statistical significance was defined as two-sided P values < 0.05.

To investigate the effects of TSH on tumor growth, BHP10-3SCp cells, a tumorigenic subclone of BHP10-3M cells originating from a PTC patient, were used. BHP10-3SCp cells did not show TSH-dependent cell growth (Supplementary Fig. S1A) or cAMP production (Supplementary Fig. S1B) in comparison with the normal rat thyrocytes, FRTL-5 cells. Thyroid differentiation–related genes such as thyroid transcription factor-1(TTF-1), TSHR, and paired-box gene 8 (PAX8) were rarely detected in BHP10-3SCp cells (Supplementary Fig. S2A), suggesting poorly differentiated status of it. Meanwhile, PAX8, not TTF1 nor TSHR, was upregulated in response to the 72-hour treatment of rhTSH (Supplementary Fig. S2B). Therefore, it is reasonable to deduce that BHP10-3SCp is a PDTC cell line dedifferentiated from a DTC cell line, as previously reported (30).

Next, BHP10-3SCp cells were transplanted into athymic nude mice to establish an ectopic tumor model, and rhTSH at two different doses (0.15 μg/g for low-TSH group; 1.5 μg/g for high-TSH group), or saline was administered intraperitoneally from day 5 to day 28. On day 10, the tumor volume of high-TSH group was larger than that of the control group, whereas there was no significant difference between low-TSH group and control group (Fig. 1A). On day 14, the tumor volumes of the low-TSH and high-TSH groups versus the controls were significantly different, by 1.9-fold (P = 0.021) and 2.7-fold (P = 0.043), respectively (Fig. 1A). The tumors in the low-TSH and high-TSH groups were approximately 2.3-fold larger at day 21 and 2.5-fold larger at day 28 than those of the controls (P < 0.05 and P < 0.001 at day 21, respectively; P < 0.05 and P < 0.01 at day 28, respectively). Tumor volume was not statistically significantly different between the low-TSH and high-TSH groups on day 21 or 28 (Fig. 1A). Human TSH levels measured in mouse serum were highest in the high-TSH group, followed by the low-TSH group, whereas the levels in the controls were undetectable (Fig. 1B). Free T4 levels were significantly higher in the high-TSH group than in controls, whereas no difference was found between the low-TSH and control groups (Fig. 1C). The density of Ki-67 staining, a marker of proliferation, was higher in the low-TSH and high-TSH groups than in the controls (Fig. 1D).

Because TSH had a minimal direct effect on the viability of BHP10-3SCp cells (Supplementary Fig. S1), we then hypothesized that the positive effects of TSH on PDTC tumor growth in vivo were due to its effects on the tumor microenvironment. Interestingly, IHC staining using anti-CD31 and anti-ERG demonstrated that tumors from the low-TSH and high-TSH groups showed a higher vascular density than those from controls (Fig. 1E). IHC staining (Fig. 1E) and Western blot analyses (Supplementary Fig. S3) showed that VEGF-R2 was highly expressed in the low-TSH and high-TSH groups compared with the controls. An ELISA assay of tumor lysates showed that VEGF-A expression likewise increased in the high-TSH group by 1.4-fold in comparison with the controls (Fig. 1F), and it was positively correlated with the serum TSH levels with marginal significance (r = 0.532, P = 0.075, Fig. 1G). Furthermore, the serum TSH levels showed positive correlation with CD31⁺ area of IHC, significantly (r = 0.594, P = 0.042, Fig. 1H).

For validation, the effects of TSH on tumor growth and angiogenesis were further evaluated in the mouse tumor model using FRO cells. Similar to BHP10-3SCp cells, TSH enhanced tumor growth (Supplementary Fig. S4A) and vascularity measured by the immunoreactivity score of CD31-positive cells (Supplementary Fig. S4B). Collectively, TSH treatment enhanced tumor growth in association with augmented tumor angiogenesis.

Because TSH stimulates thyroid hormone production, these effects of exogenous rhTSH might have resulted from the actions of both TSH and thyroid hormone. Although free T4 levels in the low-TSH group were not significantly higher than in the controls and resulted in as large of an increase in tumor growth as was observed in the high-TSH group, the effects of thyroid hormone were still possibly present. To exclude the effects of thyroid hormone from those of elevated TSH, an MMI-induced hypothyroidism model, which exhibits high TSH but low thyroid hormone, was used. BHP10-3SCp cells were s.c. injected to nude mice and MMI (0.025%) or tap water was administered by drinking water from day 7 to day 28. After 4 weeks, the tumor volume in the MMI group was larger than in the controls by 1.7-fold (P = 0.010), although the size was smaller than in the high-TSH or low-TSH groups (Supplementary Fig. S5A). Serum mouse TSH was higher (714.7 ± 342.7 pg/mL vs. 181.8 ± 55.3 pg/mL, P = 0.025, Supplementary Fig. S5B), and serum-free T4 was lower (0.52 ± 0.47 ng/dL vs. 1.11 ± 0.41 ng/dL, P = 0.038, Supplementary Fig. S5C) in the MMI group than in the control group. In the IHC staining of tumor tissues, positivity for CD31 (Supplementary Fig. S5D) and Ki-67 (Supplementary Fig. S5E) was significantly higher in the MMI group than in the controls. Moreover, the tumor lysates showed higher VEGF-A levels in the MMI group than in the controls (Supplementary Fig. S5F), suggesting that similar angiogenic changes were induced in this model to those found in the TSH injection model. Taken together, TSH enhanced tumor growth and angiogenesis independently of thyroid hormone in BHP10-3SCp PDTC tumor models.

Because a recent study showed that TSH stimulated angiogenic potentials in human endothelial cells (22), we hypothesized that TSH might directly affect endothelial cells in thyroid cancer microenvironments. To test this hypothesis, we used a triple-negative breast cancer cell line, MDA-MB-231, which showed extremely lower expression of TSHR in RT-PCR analysis (Supplementary Fig. S6A). Treatment of rhTSH did not change transcriptions of VEGF-A or CXCL8 (Supplementary Fig. S6B). Cells were transplanted in athymic nude mice, and rhTSH was injected from 1 to 4 weeks. Four weeks after cell transplantation, the tumor volumes of the TSH group showed no statistically significant difference compared with those of the control group (Supplementary Fig. S6C), although serum human TSH levels were significantly higher in the TSH group (Supplementary Fig. S6D).

Next, we hypothesized that treatment of thyroid cancer cells with TSH may lead to the creation of a proangiogenic environment. To test this hypothesis, the CM of TSH-treated BHP10-3SCp cells was harvested and administered to HMVECs. First, a cell viability assay was performed using a CCK-8 assay, and TSH-CM/B showed significantly increased cell viability at 24 hours in HMVECs by 1.3-fold (Fig. 2A). Treatment with TSH-CM/B significantly increased the cell migration potential of HMVECs in a transwell assay compared with the control-CM/B (Fig. 2B). Moreover, the tube formation potential was also enhanced in the TSH-CM/B treatment group in comparison with the control-CM/B group (Fig. 2C). To explore the TSH-mediated secretory factors that augmented the angiogenic potential of human vascular endothelial cells, TSH-CM/B was analyzed using a cytokine array containing 56 angiogenesis-related proteins. Among them, VEGF-A and CXCL8 expression showed the greatest increases in expression in the TSH-CM/B group compared with the control-CM/B group. An ELISA assay showed that VEGF-A (11.84 ± 0.59 ng/mL vs. 5.66 ± 0.28 ng/mL, P < 0.01, Fig. 2D) and CXCL8 (3.84 ± 0.47 ng/mL vs. 2.81 ± 0.33 ng/mL, P < 0.01, Fig. 2E) showed significant increases in the TSH-CM/B group compared with the control-CM/B group.

To obtain insights into the molecular mechanism of the effects of TSH on VEGF-A and CXCL8 production in BHP10-3SCp cells, AKT and ERK phosphorylation, which are known to be activated via a TSH receptor–activated PKC pathway, were evaluated (31). Because TSH did not increase cAMP production or cell growth in BHP10-3SCp cells (Supplementary Fig. S1), we then hypothesized that TSH influenced cytokine production via the PKC pathway. Interestingly, treatment with TSH stimulated the phosphorylation of AKT and ERK (Fig. 2F). Furthermore, TSH-mediated VEGF-A and CXCL8 expressions were significantly blocked by TSH receptor blocking antibody (K1-70) and inhibitors of PI3K (LY294002), mTOR (RAD001), and ERK (PD98059; Fig. 2G and H), indicating that TSH increased VEGF-A and CXCL8 secretion via PI3K/AKT/mTOR or ERK signaling in BHP10-3SCp cells.

To inhibit the effects of TSH-stimulated VEGF-A and/or CXCL8, neutralizing antibodies for each molecule were treated respectively and/or coincidently. Transwell migration and tube formation potentials of TSH-CM/B–treated HMVECs were evaluated after treatment of anti-CXCL8 or anti-VEGF-A, or both. Treatment with the neutralizing antibody of CXCL8 inhibited the TSH-CM/B–mediated increases of cell migration potential (Fig. 3A and B), but did not inhibit TSH-CM/B–dependent tube formation (Fig. 3A and C). Meanwhile, treatment with the neutralizing antibody of VEGF-A blocked TSH-CM/B–mediated increases in cell migration and tube formation potential in HMVECs (Fig. 3A−C). Moreover, blocking VEGF-A and CXCL8 coincidently showed further inhibition of tube formation potential compared with what was observed when blocking VEGF-A alone (Fig. 3A−C). In in vivo experiments, nude mice injected with BHP10-3SCp cells subcutaneously were divided into five groups: control, TSH, TSH + anti-VEGF-A, TSH + anti-CXCL8, and TSH + anti-VEGF-A + anti-CXCL8. TSH increased tumor volume by 1.4-fold compared with the control. Conversely, tumor volume was significantly decreased by blocking VEGF-A, CXCL8, and both compared with the TSH group (volume reduction of 34%, 54%, and 60% at day 28, respectively; Fig. 3D). IHC staining of the tumors revealed that the ERG expression in vascular endothelial cells of tumors was increased by TSH treatment, whereas the TSH effect was diminished by treatment of anti–VEGF-A, anti-CXCL8, or both (Fig. 3E).

To explore the biological impact of TSH-induced angiogenesis in the tumor microenvironment, we evaluated macrophage densities in tumors of the mouse model with BHP10-3SCp cells, because tumor-associated macrophages have been shown to support tumor progression in thyroid cancers (32). IHC staining with F4/80 demonstrated that a high density of macrophages was observed in the high-TSH group compared with the controls, especially in the perivascular area (Fig. 4A). Transwell migration assays using human monocyte/macrophage (THP-1) cells demonstrated that treatment of TSH did not enhance cell migration potentials in THP-1 alone (Supplementary Fig. S7A) or in coculture with BHP10-3SCp cells (Supplementary Fig. S7B), suggesting that the recruitment of macrophages was increased via enhanced vasculature rather than the direct chemotactic effects of TSH. Finally, in in vivo model, when rhTSH or saline was administered with or without clodronate-liposome, a macrophage inhibitor, treatment of clodronate-liposome reduced TSH-dependent tumor growth (Fig. 4B). Thus, TSH-enhanced tumor angiogenesis resulted in tumor growth by increasing tumor-associated macrophage recruitment into the tumor microenvironment.

Taken together, TSH stimulated VEGF-A and CXCL8 secretion from thyroid cancer cells via the AKT/mTOR and ERK signaling pathways. Secreted VEGF-A and CXCL8 affected endothelial cells in the tumor vasculature and subsequently increased macrophage recruitment, which enhanced tumor growth of thyroid cancer (Fig. 4C).

To validate the clinical relevance of VEGF-A and CXCL8 expression in human PDTC, we performed IHC staining of VEGF-A, CXCL8, CD31, ERG, and a macrophage marker, CD163, on human tissue microarrays from 13 PDTC and 35 advanced PTC patients. VEGF-A expression in both cancer cells and vasculature coincided with CD31- or ERG-positive areas (Fig. 5A–C), whereas CXCL8 was dominantly expressed in cancer cells (Fig. 5D). CD163-positive macrophages were observed at stromal sites (Fig. 5E). In PDTC samples, VEGF-A expression showed positive correlations with CD31 (Fig. 5F), ERG (Fig. 5G), CD163 (Fig. 5H), and CXCL8 expression (Fig. 5I), although the correlation with ERG was marginally significant (r = 0.507, P = 0.077, Fig. 5G). Consistent to the PDTC data, 35 advanced PTC patients with a tumor size of ≥ 2 cm also showed that the VEGF-A expression showed positive correlations with CD163 and CXCL8 expressions (Supplementary Table S2).

Next, the clinical characteristics of PDTC patients according to preoperative serum TSH levels and VEGF-A expression were analyzed. Preoperative serum TSH levels were positively related with VEGF-A expression (r = 0.583, P = 0.036, Fig. 5J) and larger tumor size (r = 0.677, P = 0.011, Fig. 5K), but not significantly associated with CD31 (r = 0.183, P = 0.550). Tumor size also positively associated with VEGF-A expression (r = 0.526, P = 0.065, Fig. 5L). Interestingly, patients with higher VEGF-A expression showed significantly higher rates of distant metastasis and poor disease specific survivals (Table 1 and Fig. 5M).

The present study demonstrated that TSH stimulated secretion of angiogenic factors such as VEGF-A and CXCL8 in PDTC cells, which showed TSH/cAMP-independent cell growth, through TSH/AKT and ERK pathways. As a result, TSH-stimulated tumor angiogenesis enhanced tumor growth in association with enhanced macrophage infiltration. Because treatment of neutralizing antibodies blocking VEGF and/or CXCL8 or macrophage inhibitor reduced TSH-dependent tumor growth in vivo, we concluded that TSH-mediated tumor angiogenesis is essential for tumor growth of PDTCs. In human thyroid cancer tissue microarrays, preoperative levels of TSH were positively correlated with VEGF-A and tumor size, and the expression of VEGF-A was positively correlated with CD31, CD163, and CXCL8, as well as their clinical poor prognosis, suggesting that TSH might affect tumor progression by modulating tumor angiogenesis and microenvironments in PDTCs. Taken together, the present study supports the usefulness of T4 suppression therapy, even in PDTCs that lose some characteristics of differentiated cells, and thus do not show TSH-dependent growth.

Classically, TSH has been established to play a role in tumor cell growth through receptor signaling, especially at the time of thyroid cancer initiation (17). However, during cancer progression, TSHR is silenced in some tumor cells, and the differentiation status of each tumor cell becomes heterogeneous. In this stage, it is reasonable to assume that T4 suppression therapy no longer exerts effects on controlling cancer cell growth (33). However, TSH also modulates the transcription of several genes in these cells (19–21). The present study clearly demonstrated that TSH stimulated angiogenic cytokines, including VEGF-A and CXCL8, in a PDTC cell line, with minimal expression of TSHR. In the context of the alternative role of TSH signaling in gene transcription, T4 suppression therapy could still be beneficial in PDTCs, even those in which the expression of TSHR is very low, because T4 suppression therapy would help control the tumor microenvironment. The next question would be the cutoff value of TSH for these proangiogenic actions. The present study did not show dose-dependent response of TSH on tumor growth, although there were positive correlations between the serum TSH levels and CD31 expressions. Further studies with human-relevant model are needed to explore the protumorigenic threshold of TSH.

Growing evidence indicates that TSH has biological functions in various human cells other than thyrocytes, such as fibroblasts (23), immune cells (34), and endothelial cells (22). Moreover, a recent study suggested that TSH may directly stimulate endothelial cells and promote angiogenesis (22). Based on this evidence, one of the specific aims of this study was to investigate whether TSH has direct effects on endothelial cells and contributes to the enhancement of tumor angiogenesis in vivo or not. If the effects of TSH on endothelial cells in the tumor microenvironment are sufficiently strong to modulate tumor vasculature, TSH signaling might have a role in the tumor microenvironment of other human cancers. Elevated serum TSH, whether overt or due to subclinical hypothyroidism, is a common human pathology in the general population (35), especially in cancer patients as it relates to antithyroid treatment (36). However, the present study, using an ectopic tumor model of breast cancer cells, demonstrated that the preangiogenic effects of TSH on human cancers without TSHR expression were limited, but further studies using various human cancer cell lines other than breast cancer cells are needed to validate it.

An interesting finding of this study is that thyroid cancer cells that showed TSH/cAMP-independent growth still responded to TSH stimulation and changed their gene transcriptional activities. Although Gsα−cAMP signaling is the canonical pathway known to regulate the proliferation and differentiation of thyrocytes, the existence of other aberrant pathways, such as PI3K/AKT and PKC/ERK, and their signaling cross-talk has been intensively studied using various TSH receptor antibodies (31). Moreover, previous studies have shown that TSH regulated IL6 or secretory IL1 receptor antagonist production in fibrocytes through the PI3K/AKT pathway, independently of Gsα−cAMP signaling (37, 38). Here, we demonstrated that TSH alternatively activated the PI3K/AKT and ERK pathways, and subsequently regulated VEGF-A and CXCL8 transcription in thyroid cancer cells showing minimal TSHR expression. These data suggest that even in cancer cells that show limited expression of TSHR and lose their differentiation status, TSH can still affect the transcriptional activities of many genes, promoting tumor angiogenesis and growth by activating aberrant TSH signaling. One of the possible explanations for the effects of TSH on thyroid cancer cells that minimally express TSHR is that gain-of-function mutations could be involved (33, 39), although further studies are needed to clarify this.

In the present study, TSH increased CXCL8 secretion in conjunction with VEGF-A. Previously, several studies have demonstrated that TSH or tumor necrosis factor-α stimulated CXCL8 production in normal thyrocytes, playing a role in the immune process in thyroiditis (21, 40). In the thyroid cancer microenvironment, CXCL8, originating from mast cells or macrophages (41, 42), has been established as an angiogenic cytokine that modulates tumor vasculature (41, 43). Herein, we first demonstrated that TSH upregulated CXCL8 production in thyroid cancer cells. Furthermore, in microvascular endothelial cells, simultaneous inhibition of VEGF-A and CXCL8 using neutralizing antibodies blocked tube formation potential in a synergistic manner, suggesting a possible biological relevance of VEGF-A and CXCL8. A previous study demonstrated that CXCL8 stimulated VEGF-A production and reinforced an autocrine loop in VEGF-A/VEGF receptor 2 signaling in murine endothelial cells (44). Taken together, CXCL8 might support VEGF-A action in thyroid cancer microenvironments (Fig. 4C).

The associations among TSH, VEGF-A, and CXCL8 were validated in human PDTC tissues alongside an assessment of the density of endothelial cells and macrophages in tumor microenvironments. The sample size was small, still preoperative TSH levels were positively correlated with VEGF-A expression and tumor size, and finally VEGF-A was associated with poor clinical outcomes including poor disease-specific survival. Meanwhile, although VEGF-A expression was positively correlated with CD31, preoperative TSH levels were not correlated with CD31, suggesting that TSH might indirectly affect tumor angiogenesis. Several angiogenic factors including VEGF-A and CXCL8 or other cellular responses might mediate the link between TSH and tumor angiogenesis. Further human study with large sample size is needed to clarify it.

In conclusion, TSH signaling modulated tumor angiogenesis and macrophage infiltration, and enhanced tumor growth by stimulating VEGF-A and CXCL8 secretions from thyroid cancer cells. Higher preoperative TSH levels were associated with the higher expression of VEGF-A, which was related to higher vascular density, macrophage infiltration, and poor prognosis. Given these aberrant actions of TSH signaling on thyroid cancer cells, T4 suppression therapy might be beneficial not only in well-differentiated cancers that show TSH-dependent growth, but also in PDTCs.

No potential conflicts of interest were disclosed.

Conception and design: Y.S. Song, S.W. Cho, Y.J. Park

Development of methodology: Y.S. Song, S.W. Cho, Y.J. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.W. Cho, Y.J. Park

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.S. Song, S.W. Cho

Writing, review, and/or revision of the manuscript: Y.S. Song, M.J. Kim, H.J. Sun, H.H. Kim, H.S. Shin, Y.A. Kim, B.-C. Oh, S.W. Cho, Y.J. Park

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.S. Song, M.J. Kim, H.J. Sun, H.H. Kim, H.S. Shin, Y.A. Kim, B.-C. Oh, S.W. Cho

Study supervision: S.W. Cho, Y.J. Park

This work was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI, HI14C1277), and a grant from the National R&D Program for Cancer Control (HA17C0040) funded by the Ministry of Health and Welfare, Republic of Korea and by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2013R1A1A3007152).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.