TSP‐1 is downregulated and inversely correlates with miR‐449c expression in Cushing's disease

Abstract The pathogenesis of Cushing's disease, which is caused by pituitary corticotroph adenoma, remains to be studied. Secreted angioinhibitory factor thrombospondin‐1 (TSP‐1) is an adhesive glycoprotein that mediates cell‐to‐cell and cell‐to‐matrix interactions and is associated with platelet aggregation, angiogenesis and tumorigenesis. We have found that the expression of TSP‐1 is significantly lower in human pituitary corticotroph tumours compared with normal adenohypophysis. This study aims to elucidate the role of TSP‐1 in regulating the tumour function of pituitary adenomas. Forced overexpression of TSP‐1 in a murine AtT20 pituitary corticotroph tumour cell line decreased corticotroph precursor hormone proopiomelanocortin (POMC) transcription and adrenocorticotropic hormone (ACTH) secretion. Functional studies showed that TSP‐1 overexpression in pituitary adenoma cells suppressed proliferation, migration and invasion. We have demonstrated that TSP‐1 is a direct target of miR‐449c. Further study showed that miR‐449c activity enhanced tumorigenesis by directly inhibiting TSP‐1 expression. Low expression of lncTHBS1, along with low expression of TSP‐1, was associated with the high expression of miR‐449c in Cushing's disease patients. Furthermore, RNA‐immunoprecipitation associates miR‐449c with lncTHBS1 suggesting that lncTHBS1 might be a negative regulator of miR‐449c. Taken together, this study has demonstrated that lncTHBS1 might function as competing endogenous RNA for miR‐449c, which could suppress the development of Cushing's disease.

hypercortisolism. 6 Approximately one-to two-third(s) of corticotroph adenomas possess somatic mutations in the 14-3-3 protein binding motif of ubiquitin specific peptidase 8 (USP8), a deubiquitinating enzyme gene. 7,8 Mutations in USP8 may be associated with some of the clinical features of Cushing's disease such as tumour aggressiveness and ACTH hypersecretion. 9,10 However, the symptoms of Cushing's disease are similar to other conditions and often remain unrecognized, especially in individuals diagnosed with diabetes, hypertension or depression. 11 In fact, the mortality rates in patients with untreated Cushing's disease are reported to be up to nine times higher those of the general population. 12,13 Pituitary adenomectomy is generally performed in the first-line treatment of Cushing's disease, however, pituitary adenomas often recur after remission. 2 Therefore, more effective therapeutic and diagnostic approaches are needed to improve the survival prospects of patients with Cushing's disease.
Thrombospondin-1 (TSP-1) is a matricellular protein first found in human plasma and widely expressed in diverse tissue. 14 TSP-1 is a multi-domain glycoprotein that is encoded by the THBS1 gene and involved in cell-to-cell and cell-to-matrix interactions, in particular, it is associated with platelet aggregation, angiogenesis and tumorigenesis. [15][16][17] Using RNA sequencing (RNA-Seq), the transcriptome of 13 cases of CD and five normal human pituitaries (NHPs) were analysed in our previous study. 18 Marked downregulation of the TSP-1 encoding gene was identified in Cushing's disease. TSP-1 has been demonstrated to have a complicated role in human cancer and to exert stimulatory and inhibitory effects in different types of tumours. TSP-1 is known as an inhibitor of proliferation in endothelial cells 19 and induces apoptosis, 20 and suppresses the cell cycle. 21 TSP-1 is under-expressed in various tumours such as colorectal cancer, 22 clear cell renal carcinoma. 23 The upregulation of TSP-1 can suppress tumour growth in stroma 24 and oesophageal squamous cell carcinoma. 25 Moreover, it was reported that activated somatostatin receptor subtype2 (sst2) and bone morphogenetic protein 4 (BMP4) can also suppress the growth of solid tumours, such as in pancreatic and cervical cancers, via the induction of TSP-1. 26,27 Interestingly, use of an sst2 analog and BMP4 can also inhibit corticotroph tumour cells and ACTH secretion as well. 28,29 In addition, through the activation of TGF-beta, TSP-1 may inhibit angiogenesis and tumour growth in multiple cancers. 30 It is also known that the downregulation of the TGF beta signalling pathway and activation of the TGF pathway may decrease the secretion of ACTH and tumour cell proliferation in pituitary corticotrophinomas. 31,32 However, TSP-1 may, in contrast, be involved in the promotion of tumorigenesis in various cancers such as gastric cancer and human melanoma. 33,34 TSP-1 induced by TGFB1 is reported to promote the migration of oral squamous cell carcinoma and stimulate the expression of matrix metalloproteinases (MMPs) through integrin signalling. 35 However, the decreased expression of TSP-1 and its cause in Cushing's disease remains to be elucidated.
TSP-1 is proposed to influence the vascular endothelial growth factor (VEGF) pathway by binding to a high-affinity receptor CD47 and disrupting its association with VEGF receptor 2, which in turn downregulates the pro-angiogenic signals downstream of VEGF. 36 Ki67, VEGF and matrix metalloproteinase-9 (MMP9) are among the markers normally used to identify the biochemical characteristics of Cushing's disease. [37][38][39][40][41] MicroRNAs (miRNAs) are short non-coding RNA molecules with 22-24 nucleotides, which can affect the stability and translation of mRNAs through binding to targeted mRNA. Several miRNAs, such as miR-26a and miR-449a, also play an important role in the regulation of ACTH-secreting pituitary adenomas. 42,43 It has been hypothesized that glucocorticoids may induce the expression of miRNAs in the pituitary. The 3′untranslated region (UTR) of TSP-1 is a potential target of miR-449c.
Long non-coding RNAs (lncRNAs) are a group of miRNAs that could function as a miRNA sponge, often referred to as competing endogenous RNA (ceRNA), regulating the expression pattern and biological characteristics of miRNA. Other studies implicate lncRNAs in the regulation of pituitary adenomas and other cancers. [44][45][46] An elevated level of lncRNA H19 expression was found in invasive pituitary adenoma cells. 44 LncRNA CCAT2 has recently been found to be significantly upregulated in pituitary adenomas tissues. 45 Loss-and gain-of-function assays showed that CCAT2 positively regulated pituitary adenoma cell proliferation, migration and invasion, and interacted with PTTG1 to promote its stability.
To investigate dysregulated lncRNAs and miRNAs in Cushing's disease we selected miR-449c and lncTHBS1 and compared their expression in ACTH-secreting adenoma and normal pituitary tissue.
Northern blot analysis, luciferase assay and RNA-immunoprecipitation assay were used to determine the reciprocal relationship between miR-449c and lncTHBS1. We also investigated the expression of TSP-1 in ACTH-secreting adenoma and its effects on the expression of POMC, VEGF, Ki67 and MMP9.

| Human tissue samples and cell culture
Human pituitary tumour samples were obtained by transsphenoidal surgery from 33 patients with Cushing's disease. Normal human pituitaries (NHP) (n = 7) were obtained from fresh autopsy specimens.
This study was approved by the Ruijin Hospital Ethics Committees and written informed consent was obtained from all patients. HEK-293T and the mouse AtT20 pituitary corticotroph tumour cell line were obtained from the American Tissue Type Collection (ATCC, Manassas, VA, USA). Cells were maintained in Dulbecco's modified Eagle's medium (DMEM, GIBCO, Carlsbad, CA, USA) with 10% foetal bovine serum (FBS), 2 mmol/L L-glutamine and 100 μg/mL penicillin/streptomycin in a humidified incubator with 5% CO 2 at 37°C.

| Plasmid construction
TSP-1 and negative controls were obtained from GenePharma (Shanghai, China). The full-length 3′-UTR of TSP1 was inserted into a pGL3-Control luciferase vector (Promega, Madison, WI, USA) following the manufacturer's instructions. All primers used in the study can be found in Table 1. The pRenilla-TK vector was used as an internal control for the dual-luciferase assay. Cells were grown to 60%-70% confluence then seeded into six-well plates 24 hours before infection. The cells in each well were transfected with 1 mL of culture medium containing 10 μL of lentivirus (1×10 9 TU/ml) and incubated for 72 hours, the infection rate was observed using a fluorescence microscope(Olympus, Center Valley, PA, USA).
Western blotting was used to confirm effective transfection.

| Immunohistochemistry
Tissues were fixed in 4% paraformaldehyde for at least 24 hours, de-  Table 1. Relative expression of mRNAs, lncTHBS1and miR-449c was calculated using the 2 −ΔΔCT method and normalized to GAPDH and U6, respectively.

Mus-Gapdh-A GGGGTCGTTGATGGCAACA
TA B L E 1 Primers used in the study

| Cell viability assays and colony formation test
To determine cell viability, AtT20 cells were seeded into 96-well plates (2000 cells/well) and incubated for 24 hours until they were attached. Cell viability was determined using a Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan) following the manufacturer's instructions. Colony formation was determined by culturing AtT20 cells in 6-well plates (10 3 cells/well) in DMEM culture medium containing 10% FBS, which was replaced every 4-5 days. After 21 days, colonies were fixed for 20 minutes in 4% paraformaldehyde in PBS containing 4% sucrose and then stained with 0.005% crystal violet for 30 minutes. After washing in PBS, colonies were counted using an inverted microscope.

| Cell apoptosis analysis (Annexin V assay)
Apoptosis was assessed by flow cytometry using an AnnexinV-

| In vivo xenograft tumour model
Six-week-old nu/nu male mice were inoculated subcutaneously with empty vector or TSP-1 overexpressing stable transfectant AtT20

| Northern blot analysis
Total RNA (20 μg) was prepared using TRIzol reagents (Life Technologies) was denatured with formaldehyde and loaded on to agarose gel containing 1.2% formaldehyde. After electrophoresis, the RNA was transferred to a nylon membrane and fixed with a UV crosslinker. The membrane was probed with a digoxigenin-labelled lncTHBS1 oligonucleotide probe with the following sequence:

| Dual-luciferase assay
The lncTHBS1 promoter was amplified from human genomic DNA and then subcloned into a pGL3 vector (Promega) to construct

| Statistical analysis
Comparisons were analysed using a two-tail unpaired Student's t test or one-way ANOVA. The Mann-Whitney U test was used for continuous variables. All statistical analysis was performed using SPSS version 16.0. The association between clinicopathological parameters and TSP-1 expression were evaluated by χ 2 test. Values are presented as mean ± standard deviation unless otherwise stated and P < 0.05 was considered to be statistically significant.
F I G U R E 3 TSP-1 inhibits AtT20 cell migration. A, Representative images of the wound healing assay in the TSP-1-transfected or vectortransfected AtT20 cells at 48 h. B, Transwell invasion assay was employed to assess the invasive potential of overexpressing TSP-1 in AtT20 cells. C, Relative expression of MMP2, MMP7 and MMP9 by qRTPCR. **P < 0.01

| TSP-1 is downregulated in human pituitary corticotroph adenoma
The non-invasive (n = 24). Moreover, microadenoma (n = 28) and macroadenoma (n = 5) are also classified based on a diameter less or more than 10 mm, respectively. The analysis of TSP-1 expression and association with these clinicopathologies is shown in Table 2.

| TSP-1 suppresses tumorigenicity in vitro and in vivo
To further investigate the influence of TSP-1 on cell proliferation, migration and invasion we transfected a murine AtT20 pituitary corticotroph tumour cell line with a TSP-1 expression vector (pcDNA-TSP-1). The relative expression of TSP-1 measured by qRT-PCR and WB was significantly lower in untransfected AtT20 cells but levels were elevated in cells transfected with pcDNA-TSP-1 (Figure 2A,B). We found that overexpression of TSP-1 decreased corticotroph precursor hormone POMC transcription and ACTH secretion in AtT20 cells and significantly decreased cell viability and colony formation ( Figure 2C-F). Moreover, flow cytometric analysis indicated that TSP-1 overexpression increased the ratio of apoptosis ( Figure 2G). TSP-1 overexpression was also found to inhibit AtT20 cell migration in a wound healing assay conducted over To determine the influence of TSP-1 overexpression on tumorigenicity, tumour growth was assessed in nude mice that were subcutaneously injected in the right flank with either control AtT20 cells or AtT20 cells overexpressing TSP-1. Tumour volume and weight were the lowest in TSP-1 overexpression mice ( Figure 4A-C). We also detected the expression levels of ACTH and corticosterone in blood samples of nude mice ( Figure 4D-E). HE and Ki67 expression were detected in tumours by IHC ( Figure 4F). TSP-1, MMP-9, VEGF expression were detected by Western Blot ( Figure 4G). Ki67, MMP-9 and VEGF levels are reduced. Taken together, these results indicate that overexpressing TSP-1 decreases tumour growth in nude mice.

| TSP-1 inversely correlates with miR-449c expression in human ACTH-secreting pituitary adenomas
The influence of miR-449c expression on the regulation of TSP-1 was next examined in pituitary corticotroph tumours. The level of miR-449c mRNA was found to be significantly elevated in tissue from patients with Cushing's disease (n = 33) compared with normal pituitary tissue (n = 7) (P < 0.01) ( Figure 5A). The results were found in AtT20 cells (P < 0.01 vs. normal mouse pituitary) ( Figure 5B). The predicted target sequence of miR-449c in the 3′UTR of TSP-1 was mutated and interactions were assessed by a Luciferase assay, which confirmed the binding of miR-449c to the target sequence in 3′UTR of TSP-1 ( Figure 5C,D). qRT-PCR and Western blotting analysis were then used to assess the effects of overexpressing miR-449c on the mRNA level of TSP-1 in AtT20 cells ( Figure 5E,F). The presence of miR-449c significantly reduced the expression of TSP-1 mRNA (P < 0.01) and the detectable levels of protein.
However, overexpression of TSP-1 reversed the effects of miR-449c in AtT20 cells ( Figure 6A-E). Cell viability, migration and invasion were all increased by the overexpression of miR-449c but decreased when TSP-1 was overexpressed. POMC expression and ACTH secretion were also increased in AtT20 cells transfected with miR-449c but decreased when TSP-1 was overexpressed in the cells. Taken together, these results imply that TSP-1 expression inversely correlates with miR-449c expression in human ACTH-secreting pituitary adenomas.

| lncTHBS1 and TSP-1 correlate through targeting by miR-449c and a reciprocal negative regulation existed between miR-449c and lncTHBS1
On  Figure 6F). Whereas, lncTHBS1 expression decreased miR-449c expression while inhibition of lncTHBS1 increased miR-449c expression ( Figure 6G). Moreover, miR-449c was identified in an lncTHBS1 complex using RNA-IP with anti-Ago2 antibody ( Figure 6H). These results indicate a positive correlation between TSP-1 and lncTHBS1 expression and demonstrate that lncTHBS1 targets miR-449c by directly binding to a miRNA response element.

| D ISCUSS I ON
ACTH-secreting pituitary adenomas, originating from pituitary corticotroph cells, are related to substantial morbidity and cause adrenal hypercortisolaemia, which is often referred to as Cushing's disease. 47 The pathogenesis of Cushing's disease is obscure and it remains difficult to treat successfully. [48][49][50] The present study aimed and MMP9 appear to be elevated in Cushing's disease. 38,40 Ki-67 is a marker of cell proliferation and is found to occur at significantly higher levels in invasive compared with non-invasive adenomas. 51 The growth factor VEGF can induce endothelial cell proliferation and the permeabilization of blood vessels, promote cell migration, inhibit apoptosis and is actively involved in the angiogenesis, vasculogenesis and endothelial cell growth of several cancers including colorectal cancer 52 and ovarian granulosa cell tumour. 53 In pituitary adenomas, VEGF expression is well preserved and might contribute to the vascular supply of tumours. 54 . The inhibitory effect of TSP-1 on VEGF-mediated angiogenesis is also proposed to involve the TSP-1 receptor CD36 and endothelial cell apoptosis pathways. 55  are associated with a greater reoccurrence and a shorter recurrence-free interval. 38 The same study found that PTTG, HMGA2 and Ki-67 expression were not significantly different between recurrent and non-recurrent ACTH-secreting pituitary tumours. In F I G U R E 7 Positive correlation between TSP-1 and lncTHBS1 expression. lncTHBS1 targets miR-449c by directly binding to a miRNA response element. A, The expression of lncTHBS1 mRNA in CD (n = 33) and normal pituitary tissue (n = 7). B, Correlation analysis of the relationship between lncTHBS1 expression and TSP-1 level. C, Northern blot assays were used to detect lncTHBS1 expression in CD. D, Schematic representation of the predicted binding sites for miR-449c, and the site mutagenesis design for the reporter assay. E, The relative luciferase activities were inhibited in the HEK-293T cells transfected with the reporter vector lncTHBS1-WT, but not with the reporter vector lncTHBS1-Mut. F, miR-449c expression decreased lncTHBS1 expression, and inhibition of miR-449c enhanced lncTHBS1 expression in the HEK-293T cells. G, lncTHBS1 expression decreased miR-449c expression whereas inhibition of lncTHBS1 increased miR-449c expression in HEK-293T cells. H, miR-449c was identified in the lncTHBS1 complex. miR-NC and miR-449c cell lysates were used for RNA-IP with anti-Ago2 antibody. Cells transfected with miR-449c mimics or miR-NC, followed by qRT-PCR to detect lncTHBS1. **P < 0.01 the present study, VEGF and MMP9 appeared to be upregulated expression. In addition, miR-449c was identified in an lncTHBS1 complex using RNA-IP with anti-Ago2 antibody. RNA-IP is a novel strategy that exploits the involvement of Argonaute proteins in the translational repression complex associated with microRNA. 63 The technique involves the immunoprecipitation of whole miRsilencing complexes containing miRs and associated target mRNAs. 64 Our results indicate a positive correlation between TSP-1 and lncTHBS1 expression and demonstrate that lncTHBS1 targets miR-449c by directly binding to a miRNA response element. Therefore, lncTHBS1 may function as a competitive endogenous RNA with miR-449c.
In conclusion, our present work reveals that TSP1 expression is downregulated in Cushing's disease.

ACK N OWLED G EM ENTS
This study was supported by grants from the National Natural Science Foundation of China (nos. 81770779 and 81700692).

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.