Functions and Mechanisms of Tumor Necrosis Factor-Α and Noncoding Rnas in Bone-Invasive Pituitary Adenomas

Published OnlineFirst July 6, 2018; DOI: 10.1158/1078-0432.CCR-18-0472

Cancer Therapy: Clinical Clinical Cancer Research Functions and Mechanisms of Tumor Necrosis Factor-a and Noncoding RNAs in Bone-Invasive Pituitary Adenomas Haibo Zhu1, Jing Guo1, Yutao Shen1, Wei Dong1, Hua Gao1, Yazhou Miao1, Chuzhong Li1,2,3,4, and Yazhuo Zhang1,2,3,4

Abstract

Purpose: To explore the molecular mechanism and prog- pathways. Pathway act work showed that osteoclast differen- nosis of bone-invasive pituitary adenomas (BIPA). tiation pathway was significantly implicated in the pathway Experimental design: A total of 274 patients with pituitary network. BIPAs had higher expression of TNFa than that of adenomas were followed up. Transcriptomic microarrays NBIPAs on IHC. In vitro, TNFa could induce RAW264.7 cells to analysis was performed on 10 pituitary adenomas, including differentiate into mature osteoclasts, leading to bone destruc- five BIPAs and five non-bone-invasive pituitary adenomas tion. NR_033258, lncRNA SNHG24, miR-181c-5p, and miR- (NBIPA). The targeted molecular markers were validated by 454-3p can regulate TNFa expression. qRT-PCR, IHC, ELISA, and osteoclast differentiation. Conclusions: BIPAs had worse PFS than did NBIPAs in Results: Clinical variable analyses revealed a significant the NGTR and NFPA groups. Inflammatory and immune correlation between bone invasion and female sex, large factors play an important role in BIPAs. TNFa can directly tumor volume, non-gross total resection (NGTR), and tumor induce osteoclast differentiation in BIPAs. NR_033258, regrowth. BIPAs had worse progression-free survival (PFS) lncRNA SNHG24, miR-181c-5p, and miR-454-3p can reg- than did NBIPAs in the NGTR and nonfunctional pituitary ulate TNFa expression. TNFa anditsrelatedlncRNAs adenoma (NFPA) groups. Gene ontology functional and and miRNAs represent potential therapeutic targets for KEGG pathway analyses showed that the biological processes bone-invasive pituitary adenomas in the future. Clin Cancer and pathways were primarily immune and inflammatory Res; 1–10. 2018 AACR.

Introduction analyze the prognosis, clinical features, and molecular mechanism of BIPAs to design and implement appropriate therapeutic Pituitary adenomas are benign tumors that originate in the interventions. anterior pituitary cells and account for approximately 8% to 15% Common diseases that lead to excessive bone destruction of intracranial tumors (1, 2). Large pituitary adenomas are typ- include tumors (bone metastases of cancer, osteosarcoma, giant ically invasive and infiltrate the surrounding dura, sinus, brain, cell tumors of the bone, neuroblastoma, chordoma, meningi- and bone tissue. Existing reports focus mainly on the suprasellar oma, etc.), periodontal disease (oral squamous cell carcinoma, and lateral-invasion of pituitary adenomas. There are few studies etc.), rheumatoid arthritis, and others (4–8). Tumor cells alone focused on the prognosis and mechanisms of bone-invasive cannot directly lead to bone destruction but can promote the pituitary adenomas (BIPA); most studies are clinical and imaging differentiation and maturation of osteoclasts to play a role in (3). Because of its serious destruction of the surrounding bone bone destruction. The differentiation and maturation of osteo- mass and large size, the surgical resection of BIPAs is a significant clasts is typically mediated by cellular components such as challenge and may increase the potential risk of cerebrospinal immune cells, the tumor cells and cytokines such as IL1/6, fluid leaks. Based on these factors, it is important to explore and TNFa,PTHrP,TGFb, CXCL13, CXCL12, etc. (9, 10). TNFa is a dominant cytokine that plays a critical role in the promotion of pathologic osteoclast formation leading to inflammatory bone 1Beijing Neurosurgical Institute, Capital Medical University, Beijing, China. destruction (11). In this study, we identified TNFa as a key 2Department of Neurosurgery, Beijing Tiantan Hospital affiliated to Capital 3 molecule involved in BIPAs. Medical University, Beijing, China. Beijing Institute for Brain Disorders Brain miRNAs are small, single-stranded, noncoding RNAs of Tumor Center, Beijing, China. 4China National Clinical Research Center for Neurological Diseases, Beijing, China. approximately 22 nucleotides that mediate homologous sequence-dependent gene silencing in cells. Research has dem- Note: Supplementary data for this article are available at Clinical Cancer onstrated that some miRNAs, such as miR-34c, miR-124, miR- Research Online (http://clincancerres.aacrjournals.org/). 223, etc., are involved in osteoclast differentiation (12–14). Long Corresponding Author: Yazhuo Zhang, Beijing Neurosurgical Institute, Beijing noncoding RNAs (lncRNA) are RNAs with transcript lengths of 100050, China. Phone: 86-10-67096763; E-mail: [email protected] >200 nucleotides that regulate gene expression through multiple doi: 10.1158/1078-0432.CCR-18-0472 mechanisms. A recent study showed that the lncRNA molecules 2018 American Association for Cancer Research. MAYA and lncRNA MALAT1 are involved in cancer cell–induced

www.aacrjournals.org OF1

Zhu et al.

January 1, 2008, to December 31, 2015. All patients were Translational Relevance classified according to preoperative images, including plain Bone-invasive pituitary adenomas (BIPA) are a rare cause of and enhanced head MRI, thin layer skull base CT scanning and pituitary adenomas, but their mechanisms and prognosis three-dimensional reconstruction. The patients were divided remain unknown. Because of its serious destruction of the into three groups: BIPAs, invasive pituitary adenomas (IPA) surrounding bone mass and large size, surgical resection of but not invasive to the bone, and noninvasive pituitary ade- BIPAs is a significant challenge and may increase the potential nomas (NIPA). The criteria of bone invasion included severe risk of cerebrospinal fluid leaks. So, it is important to explore destruction of the clivus or anterior skull base on CT and bone and analyze the prognosis, clinical features, and molecular destruction observed during operation or pathological reports mechanism of BIPAs to design and implement appropriate showing bone tissue infiltration (Fig. 1). This study was therapeutic interventions. In this study, we found that NGTR conducted in accordance with established ethical guidelines and invasion were independent risk factors of pituitary ade- as outlined in the Declaration of Helsinki. We obtained nomas, BIPAs had worse PFS than did NBIPAs in the NGTR written informed consent from each subject, and the Ethics and NFPA groups. Also, we identified TNFa as a key molecule Committee of Beijing Tiantan Hospital approved this study. involved in BIPAs. NR_033258, lncRNA SNHG24, miR-181c- 5p, and miR-454-3p can regulate TNFa expression. TNFa and Tissue samples and histology its related lncRNAs and miRNAs represent potential therapeu- All samples were obtained from the Beijing Tiantan Hospital tic targets for bone-invasive pituitary adenomas in the future. Neurosurgery. Fresh tumor tissue samples were immediately snap-frozen in liquid nitrogen and stored at 80C. A total of 74 pituitary adenomas samples were included in this study, of which 10 frozen tissue samples were subjected to transcriptome osteoclast differentiation and bone resorption (15, 16). In addi- microarrays (experimental group, five BIPAs and five NBIPAs; tion, lncRNAs are involved in regulating mRNA stabilization and Supplementary Table S1), 59 paraffin tissue samples were sub- transport as well as miRNA sponging (17). In this study, we found jected to IHC (confirmation group, including 27 BIPAs and 32 that NR_033258, lncRNA SNHG24, miR-181c-5p, and miR-454- NBIPAs), and five invaded bone and normal bone samples from 3p can regulate TNFa expression. BIPAs were subjected to HE staining and SEM.

Materials and Methods ceRNA microarrays Patients Total RNA was extracted and puriﬁed by using the mirVana We retrospectively reviewed the patients with pituitary ade- miRNA Isolation Kit without phenol (Cat # AM1561, Ambion, noma in a single ward of Beijing Tiantan Hospital from Austin, TX, US). RNA samples from each group were then used to

Figure 1. The imaging performance of three groups: A, BIPAs; B, IPAs but not invasive bone; C, NIPAs; D, Intraoperative image: bone destruction of clivus; E, HE staining of the BIPA: pituitary adenoma inﬁltrating bone tissue; F, electron microscope of the BIPA: diffused distribution of GH, PRL, and FSH granules.

OF2 Clin Cancer Res; 2018 Clinical Cancer Research

The Prognosis and Mechanism of BIPAs

generate fluorescence-labeled cRNA targets for the SBC human ¼ CT (gene of interest) CT (GAPDH)]. All qRT-PCR analyses ceRNA array V1.0 (4 180 K). The labeled cRNA targets were then were performed in triplicate. Student t tests were applied, and a hybridized with the slides. After hybridization, the slides were P-value <0.05 was considered significant. The primer sequences scanned on the Agilent Microarray Scanner (Agilent Technolo- are presented in Supplementary Table S2. gies). The data were extracted with Feature Extraction software 10.7 (Agilent Technologies). Raw data were normalized by the HE staining and IHC Quantile algorithm using the limma package of the R program. The invaded bone was fixed in 10% paraformaldehyde, dec- The microarray experiments were performed according to the alcified in formic acid, and embedded in paraffin. All samples protocol of Agilent Technologies, Inc., at Shanghai Biotechnology were sectioned at 5 mmol/L. The invaded bones were stained with Corporation. Ratios were calculated between the bone invasion HE. PAs were stained with HE and processed for IHC staining with group and nonbone invasion group. Then, differentially anti-TNFa (ab 8348, 1:1,200; Abcam) and anti-RANKL (ab 9957, expressed genes were identified by using the t-test with a cut-off 1:1, 000; Abcam) antibodies by using the immunoperoxidase criteria of P < 0.05 and fold-change >2or<0.5. The selected technique (Leica BOND III System). The staining intensity was mRNAs were grouped in functional categories based on the Gene scored as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong). An Ontology database (GO: http://www.geneontology.org/), and H-score was obtained by multiplying the staining intensity by a functional pathways (KEGG and BIOCARTA) were also analyzed constant to adjust the mean to the strongest staining [score ¼ 1.0 by using an online SAS analysis system. (%weak) þ 2.0 (%moderate) þ 3.0 (%strong)]. The final staining score >100 was defined as high expression, and 100 was defined miRNA microarrays as low expression. Total RNA was extracted and purified by using the mirVana miRNA Isolation Kit without phenol (catalog no. AM1561; RAW264.7 cell culture and experimental grouping Ambion). Human miRNA microarrays from Agilent Technologies GH3 and RAW264.7 cells were authenticated by China Infra- (860 K), containing probes for 2549 human miRNAs from the structure of Cell Line Resource and tested negative for mycoplas- miRbase V21.0 database, were adopted. Total RNA (100 ng) ma by China Infrastructure of Cell Line Resource, bone slices were extracted from each serum sample was used as inputs for sample obtained from mouse skulls. The following experimental groups labeling and hybridization preparation in accordance with the were used: group 1: GH3 cells, group 2: RAW264.7 cells, group 3: manufacturer's protocol (Agilent Technologies). The microarray GH3 cells þ TNFa 50 ng/mL (315-01A-5; Peprotech), and group image information was converted into spot intensity values using 4: RAW264.7 cells þ TNFa 50 ng/mL. Every group included six Scanner Control Software Rev. 7.0 (Agilent Technologies). Raw wells (three coverslips and three bone slices, 3 103 cells/well). data were normalized by using the Quantile algorithm, included The bottom of the 24-well plate was covered with coverslips and in the R package AgiMicroRna (Lopez-Romero, P. BMC Geno- bone slices. These cells were cultured in DMEM with 10% FBS at mics, 2011). The microarray experiments were performed accord- 37 C in a humidified 5% CO2 atmosphere, and the culture ing to the protocol of Agilent Technologies, Inc., at Shanghai medium was adjusted to a total volume of 1 mL. After 2 hours Biotechnology Corporation. Then, differentially expressed miR- of incubation, the coverslips and bone slices were removed and NAs were identified using the t test with the cut-off criteria of P < thoroughly rinsed in DMEM with 10% FBS. After removing the 0.05 and fold-change >2or<0.5. nonadherent cells, we placed the coverslips and bone slices into a new 24-well plate, adjusted the medium to 1 mL and replaced the Construction of the Cytoscape pathway act network medium and various factors every 2 days. Path-Net is the interaction network of the most common pathways associated with the differentially expressed genes and Osteoclast differentiation and identification was constructed according to the interaction among pathways of Cell cultures of selected bone slices and coverslips were termi- the KEGG database to directly and systemically determine the nated after incubating for 24 hours to confirm that there were no interactions among the significant pathways. We used the soft- mature OCs present in the RAW264.7 cultures. On the seventh day ware Cytoscape software (V2.8.0; http://www.cytoscape.org) to of culture, the cell cultures on the remaining bone slices and construct a pathway act network for graphical representations of coverslips were terminated. The coverslips were used for TRAP central pathways by using the genes enriched in the significant staining (TRAP, 387A-1KT; Sigma-Aldrich), and TRAP-positive, canonical pathways of KEGG (P < 0.05). multinucleated cells containing more than three nuclei, were counted as osteoclasts. Bone slices were examined to detect the qRT-PCR validation formation of bone resorption lacuna. Total RNA was extracted using Trizol reagent (Invitrogen) andthenreversedtranscribed using the HiFiScript gDNA Scanning electron microscope Removal cDNA Synthesis Kit (CWBio) according to the man- The invasive bones and normal bones of the bone invasion ufacturer's instructions. Subsequently, we performed qRT-PCR group and bone slices in cell experiments were placed in NH4OH using SYBR Green assays in a total reaction volume of 10 mL. for half an hour, sonicated for 10 minutes to remove surface cells, GAPDH was used as a reference gene. The levels of miRNAs subjected to gradient alcohol dehydration, naturally air-dried, were measured by qRT-PCR using the miDETECT A Track surface coated with gold powder and were then examined by miRNA qRT-PCR Kit containing a miRNA-specificforward scanning electron microscope. primer (RiboBio) and performed on an ABI 7500 System (Applied Biosystems). For the quantitative analysis, relative Western blot analysis expression levels were calculated based on CT values (corrected RAW264.7 cells treated with TNFa (0 and 100 ng/mL) for 30 D for GAPDH expression) according to the equation: 2 CT [DCT minutes (NFATc1 for 24 hours) were lysed in nondenaturing lysis

www.aacrjournals.org Clin Cancer Res; 2018 OF3

Zhu et al.

buffer (Applygen). For Western blotting, the protein samples (30 Statistical analysis mg) were separated by 10% SDS-PAGE and then transferred to Statistical analysis was performed by using SPSS software polyvinylidene difluoride membranes. Different blots were incu- (version 20.0, IBM). The x2 test was used to analyze the associ- bated with antibodies against JNK (9252P, 1:1,000; Cell Signaling ation among patient sex, age, hormone secretion, resection Technology), phospho-JNK (4668P, 1:1,000; Cell Signaling Tech- degree, and regrowth with clival invasion. The two independent nology), p38 (8690P, 1:1,000; Cell Signaling Technology), phos- samples t test was used to examine the association between tumor pho-p38 (4511P, 1:1,000; Cell Signaling Technology), p42/44 volume and history duration and clival destruction. Variables that (4695P, 1:1,000; Cell Signaling Technology), phospho-p42/44 showed an association with regrowth of pituitary adenomas were (4370P, 1:2,000; Cell Signaling Technology), TRAF6 (ab 33915, used in univariate and multivariate analyses. The regrowth-free 1:2,000; Abcam), NFATc1 (ab 25916, 1:400; Abcam), or GAPDH curves were generated by the Kaplan–Meier method. P values of (G1020V, 1:10,000; Beijing GuanXingYu Sci & Tech Co., Ltd.), <0.05 were defined as a significant difference. followed by incubation with secondary antibodies tagged with horseradish peroxidase (Santa Cruz Biotechnology). The blots Results were visualized by enhanced chemiluminescence, and densitom- Clinical features etry was performed with an imaging apparatus (Amersham Imag- A total of 1,477 patients with pituitary adenoma from January er 600; GE). GAPDH was used as a loading control. 1, 2008, to December 31, 2015, were retrospectively analyzed; of these, 95 (6.4%) patients met the criteria of bone invasion. All Construction of the TNF-related ceRNA network bone invasion patients underwent endoscopic endonasal tumor The interactions of miRNAs–TNF were predicted by using resection. The average age was 48.5 11.6 years (ranging from 18 TargetScan (http://www.targetscan.org/). The interactions of miR- to 73 years), with 58 female and 37 male patients, including 20 NAs–lncRNAs were predicted using miRcode (http://www. ACTH, 7 FSH, 4 GH, 6 plurihormonal adenoma, 50 NFPA, and 8 microde.org/). All mRNAs, lncRNAs, and microRNAs were dif- PRL cases. A total of 274 patients were followed up (Supplemen- ferentially expressed between the two groups. The coexpression tary Table S4) over a median clinical follow-up of 30.7 months of TNF-lncRNAs was calculated with Pearson's correlation (range, 8 to 54 months). Clinical variable analyses revealed a coefficient (PCC) set at 0.5 (only a positive correlation was significant correlation between bone invasion and female sex (P ¼ retained). Overlap of the same miRNA seed sequence binding 0.044), large tumor volume (P < 0.0001), tumor resection degree site on both lncRNAs and mRNAs predicted an lncRNA–miRNA– (P < 0.0001), and tumor regrowth (P < 0.0001), and there was no TNF interaction. significant correlation between bone invasion and age, course of disease or FPA (Supplementary Table S5). These observations Short hairpin RNAs, miRNAs inhibitors, and transfection of were confirmed using a Cox proportional hazard regression HEK 293T cells analysis. Univariate analysis showed that larger tumors (P < Short hairpin RNA (shRNA) against lnc-EIF3H-3:1 (sh-lnc- 0.0001), invasion (P < 0.0001), and NGTR (P < 0.0001) were EIF3H-3:1), lncRNA SNHG24 (sh-lncRNA SNHG24), and associated with tumor regrowth, but there was no significant NR_033258 (sh-NR_033258) and the negative control (sh-NC) difference in sex (P ¼ 0.89) and age (P ¼ 0.08). Multivariate were used and synthesized by GenePharma. miR-150-5p inhibi- analysis showed that NGTR (P ¼ 0.001) and invasion (P ¼ 0.017) tors, miR-181c-5p inhibitors, miR-454-3p inhibitors, and their were independent risk factors for tumor regrowth (Table 1). negative controls were used and synthesized by RiboBio. Supple- Kaplan–Meier curve (Fig. 2A) of BIPAs and NBIPAs showed that mentary Table S3 contains detailed sequences information for the BIPAs had worse PFS than did NBIPAs (P < 0.0001). Kaplan– shRNAs, miRNA inhibitors, and their negative controls. Expo- Meier curve (Fig. 2B) of the three groups showed that BIPAs had nentially growing HEK 293T cells (2 105) were seeded onto six- worse PFS than did IPAs (P ¼ 0.02), IPAs had worse PFS than did well plates overnight, and then transfected were performed by NIPAs (P ¼ 0.009). For NFPAs and FPAs (Fig. 2C and D), the using Lipofectamine3000 transfection reagent (Thermo Fisher Kaplan–Meier curve showed that BIPAs had worse PFS than did Scientific; final concentration: shRNA or negative control: 1.25 NBIPAs in the NFPAs group and had a significant difference (P ¼ mg/mL, miRNA inhibitors or negative control: 100 nm). The 0.0002). In FPAs group, there was no significant difference (P ¼ transfection efficiency was determined by qRT-PCR at 48 hours 0.08), but BIPAs had the tendency of worse PFS than did NBIPAs. after transfection. The cell culture supernatant was collected to Separate Kaplan–Meier curves were generated for both the NGTR measure the TNFa concentration. and GTR subgroups (Fig. 2E and F). The Kaplan–Meier curve showed that BIPAs had worse PFS than did NBIPAs in the NGTR Measurements of TNFa protein production group and that the difference was significant (P ¼ 0.03). In the The TNFa concentration of the culture supernatants was mea- GTR group there was no significant difference (P ¼ 0.6). We sured by using commercial ELISA kits purchased from SenBeiJia considered that the lack of significance may be due to a too-small Biological Technology. quantity of BIPAS in the FPAs and GTR groups.

Table 1. Factors associated with pituitary adenoma regrowth by Cox regression analysis Univariate Cox regression Multivariate Cox regression Clinic character HR (95%) P value HR (95%) P value Sex (male vs. female) 1.048 (0.538–2.039) 0.890 Age (<46 years vs. 46 years) 0.541 (0.273–1.075) 0.080 Diameter (>30 vs. 30) mm 0.277 (0.138–0.557) 0.000 Invasion (BIPA/IPA vs. NIPA) 2.665 (1.719–4.130) 0.000 1.842 (1.115–3.043) 0.017 Resection (GTR vs. NGTR) 5.983 (2.864–12.500) 0.000 4.789 (2.208–10.386) 0.001

OF4 Clin Cancer Res; 2018 Clinical Cancer Research

The Prognosis and Mechanism of BIPAs

Transcriptome microarrays identified differentially expressed and coagulation cascades, cell adhesion molecules, NF-kB sig- mRNAs, circRNAs, lncRNAs, and miRNAs naling pathway, and osteoclast differentiation pathway were core Compared with the NBIPAs, 518 mRNAs, 721 circRNAs, 684 nodes of the predicted pathways in the network. In addition, the lncRNAs, and 98 miRNAs were significantly differentially osteoclast differentiation pathway play an important role in bone expressed, including 405 upregulated and 113 downregulated destruction. We selected the osteoclast differentiation pathway for mRNAs, 458 upregulated and 263 downregulated circRNAs, 427 further verification. upregulated and 257 downregulated lncRNAs, and 66 upregulated and 32 downregulated miRNAs (Supplementary Table S6). Validation of microarray data by qRT-PCR Hierarchical clustering showed that the expression patterns of the Based on bioinformatics prediction, we selected 10 mRNAs of mRNAs, circRNAs, lncRNAs, and miRNAs between two groups the osteoclast differentiation pathway, and the results showed were obviously distinguishable [Fig. 3 (A) mRNA, (B) lncRNA, that the expressions levels of CCL3, CXCL12, CCL2, PIK3CG, (D) miRNA, (E) circRNA]. SOCS3, and TNFa were significantly different between the two groups (Fig. 4B). In total, nine lncRNAs and five miRNAs asso- GO and KEGG analysis of differentially expressed mRNAs ciated with TNFa were validated via qRT-PCR analysis, and three GO analysis was performed to study the biological processes, lncRNAs (Lnc-EIF3H-3:1, NR_033258, lncRNA SNHG24) and cellular components, and specific molecular functions of all three miRNAs (miR150-5p, miR-181c-5p, and miR-454-3p) were differentially expressed mRNAs. We performed GO analysis of significant between the two groups. Most of these results were the mRNAs that were differentially expressed between the two consistent with those of RNA microarrays. groups. The results showed that the biological processes were mainly about migration and chemotaxis of inflammatory IHC of TNFa and RANKL cells and immune cells (Fig. 3C). KEGG pathway analysis We selected TNFa and RANKL for IHC verification. The samples targeting differentially expressed mRNAs (the top 30 pathways were scored for cytoplasm and extracellular expression of TNFa with the highest enrichment scores were selected) revealed that and cytomembrane expression of RANKL. Compared with the differentially expressed mRNAs were mainly involved in nonbone invasion group, the bone invasion group had high TNFa inflammatory, immune, and chemokine-related signaling path- expression (x2 ¼ 10.31; P ¼ 0.001; Supplementary Table S7), as ways and the osteoclast differentiation pathway (Fig. 3F). Based shown in Fig. 4C, which is consistent with the results of micro- on the pathway enrichment analysis results, we speculated that array data and qRT-PCR analysis. The expression of RANKL was inflammatory and immune factors may play an important role low in both groups (Fig. 4C) and showed no significant difference in BIPAs. between two groups.

Cytoscape pathway act network Osteoclast identiﬁcation We used Cystoscope to construct a pathway act network accord- There were no TRAP-positive multinucleated cells or bone ing to the overlap of common differentially expressed molecules resorption lacuna formation when the cell culture was terminated in all signiﬁcant canonical pathways (Fig. 4A). The results showed after 24 hours. When the cell culture was terminated after 7 days, that apoptosis, Toll-like receptor signaling pathway, complement no TRAP-positive cells or bone resorption lacuna were found in

Figure 2. PFS analysis of different subgroups: A, BIPAs and NBIPAs; B, BIPAs, IPAs, and NIPAs; C, NFPA group; D, FPA group; E, NGTR group; F, GTR group.

www.aacrjournals.org Clin Cancer Res; 2018 OF5

Zhu et al.

Figure 3. Heat map showing the expression proﬁles of mRNAs, lncRNAs, miRNAs, and circRNAs between the two groups. Differentially expressed lncRNAs, miRNAs, circRNAs, and mRNAs (fold change > 2or<0.5 and P < 0.05) between the two groups were analyzed using hierarchical clustering. A, mRNAs; B, lncRNAs; D, miRNAs; E, circRNAs; each row represents a single mRNA, lncRNA, miRNA, or circRNA, and each column represents one sample. Red indicates high relative expression and green indicates low relative expression. GO analysis and KEGG pathway analysis. C, Go annotation of differentially expressed mRNAs with the top 30 signiﬁcant enrichment covering domains of biological processes, cellular components, and molecular functions. F, KEGG pathway analysis of mRNAs enriched in the top 30 pathways according to the P value.

group 1 (GH3 cells), group 2 (RAW264.7 cells), or group 3 (GH3 DLK1-6:9, lncRNA SNHG24, lnc-CLSTN2-1:1, lnc-PLXDC2-3:1, þ TNFa 50 ng/mL), whereas a large number of TRAP-positive and NR_033258) and five miRNAs (miR-130b-3p, miR-150-5p, multinucleated cells and bone resorption lacuna were present in miR-181c-3p, miR-181c-5p, and miR-454-3p) associated with group 4 (RAW264.7 cells þ TNFa 50 ng/mL; Fig. 4D and E). TNFa expression between the two groups (Fig. 5A). Three lncRNAs and three miRNAs were significant difference between the two HE staining and SEM of invasive bone and normal bone groups (Fig. 5B and C). From the TNFa ceRNA network, except for Under the light microscope (40), the normal bone slice miR-150-5p, which was positively correlated with TNFa and (piriform bone) had a smooth surface, whereas the surface of lncRNAs, the miRNAs were negatively correlated with TNFa and the invaded bone slice of BIPAs was uneven, forming grooves of lncRNAs, and all lncRNAs were positively correlated with TNFa. different sizes, in which the tumor cells infiltrate. Under SEM (1,000), the normal bone slice (piriform bone) also had a NR_033258, lncRNA SNHG24, miR-181c-5p, and miR-454-3p smooth surface (Fig. 4F), but the surface of invasive bone and can regulate TNFa expression bone slices cultured with RAW264.7 and TNFa showed significant Considering the crucial role of TNFa in the differentiation of bone lacunae. RAW264.7, we focused on this protein. To further evaluate the regulatory effect of lncRNAs and miRNAs validated by qRT-PCR TNF-induced osteoclastogenesis via MAPK pathway on the expression of TNFa, HEK 293T cells were transfected The MAPK pathway plays an important role in cell growth, with sh-lnc-EIF3H-3:1, sh-lncRNA SNHG24, sh-NR_033258, differentiation, inflammation, and other important cellular phys- miR-150 inhibitors, miR-181c-5p inhibitors, miR-454-3p inhi- iologic and pathologic processes. Western blot analysis showed bitors, and their respective negative controls. qRT-PCR and that the expression levels of TRAF6, phospho-p38, phospho-p42/ ELISA analyses showed that TNFa expressionatbothmRNA 44, phospho-JNK, and NFATc1 significantly increased after treat- (Fig. 5D) and protein (Fig. 5F) levels was significantly reduced ment with TNFa at 100 ng/mL (Fig. 4G). in HEK 293T cells transfected with sh-lncRNA SNHG24 and sh- NR_033258. The TNFa expressionatbothmRNA(Fig.5E)and The TNF-related ceRNA network protein (Fig. 5F) levels was significantly increased in miR-181c- In this study, we constructed a ceRNA network based on coex- 5p inhibitors and miR-454-3p inhibitors-introduced HEK 293T pressed miRNAs–mRNAs, miRNAs–lncRNAs, and lncRNAs– cells. Together, these data demonstrated that NR_033258, mRNAs. There were nine upregulated lncRNAs (lnc-EIF3H-3:1, lncRNA SNHG24, miR-181c-5p, and miR-454-3p can regulate lncRNA H19, ENST00000559321, ENST00000516179, lnc- TNFa expression.

OF6 Clin Cancer Res; 2018 Clinical Cancer Research

The Prognosis and Mechanism of BIPAs

Figure 4. TNFa directly induces osteoclast differentiation in BIPAs. A, Cytoscape pathway act network: pathway act network according to the overlap of common differentially expressed molecules in top 30 signiﬁcant canonical pathways. The size of the circle represents the degree of association. B, qRT-PCR: the relative expression levels of 10 mRNAs. C, IHC (200)ofTNFandRANKL.D, Osteoclast differentiation assays in RAW264.7 cell and RAW264.7þ TNFa 50 ng/mL. E, SEM (1,000) of bone slices: group 1: GH3, group 2: RAW264.7, group 3: GH3þ TNFa 50 ng/mL, group 4: RAW264.7þ TNFa 50 ng/mL. F, HE staining (40)andSEM(1,000) of invasive bone and normal bone. G, Western blot analysis of TRAF6, p38, JNK, ERK1/2, and NFATc1 expression after treatment with TNFa (0 and 100 ng/mL). NC, no control; , P < 0.05; , P < 0.01.

Discussion pituitary adenomas is not dependent on the RANKL/RANK/ Cavernous sinus invasion was previously reported to range OPG pathway. from 30% to 63% based on radiology alone but only 9% to Horowitz and colleagues (23) reported that multiple soluble 30% in a series based on operative evidence (18, 19). A large mediators of immune cells, including cytokines, chemokines, and number of studies have shown that cavernous sinus invasion is growth factors, also regulate osteoblast and osteoclast activity. It is an important factor affecting the prognosis of patients with likely that cytokines are critically responsible for the changes in pituitary adenomas (20, 21). However, the clinical features and bone metabolism that occur under postmenopausal osteoporosis prognosis of BIPAs are unclear because these anomalies are and inflammatory conditions such as rheumatoid arthritis, peri- very rare. Chen and colleagues (3) reported that clival invasion odontal disease, tumors and inflammatory bowel disease. The by pituitary macroadenoma was detected by preoperative CT in tumor cells themselves cannot directly cause bone destruction but 8.21% of patients compared with 6.4% of patients in this can promote the maturation of osteoclasts through the secretion study. Chen and colleagues also reported that female sex and of related proinflammatory factors. large tumor volume were risk factors for clival invasion by This study is the first to examine invaded bone by transcrip- multivariate analysis. In this study, female sex, large tumor tional microarrays and to characterize the molecular mechanisms volume, tumor resection degree, and tumor regrowth were all of BIPAs. The results showed that the biological processes differ- associated with bone invasion. Because of the extensive inva- entially associated with the BIPAs and NBIPAs primarily involved sion of the skull base, it is difficult to completely resect the the migration and chemotaxis of inflammatory and immune cells. tumor for BIPAs. In this study, multivariate analysis showed KEGG pathway analysis revealed that the differentially expressed that NGTR (P ¼ 0.001) and invasion (P ¼ 0.017) were mRNAs were mainly involved in inflammatory, immune, and independent risk factors for pituitary adenomas regrowth, and chemokine-related signaling pathways and the osteoclast differ- Kaplan–Meier analysis showed that bone invasion increased entiation pathway. Based on pathway enrichment analysis, the risk of regrowth. we speculated that inflammatory and immune factors may play The RANKL/RANK/OPG pathway is a critical signal for an important role in bone-invasive pituitary adenoma. In addi- the differentiation of mature osteoclasts from precursor cells tion, Wu and colleagues (24) reported that the expression (9, 22). RANKL can combine with RANK and directly induce of inflammatory activity-associated proteins was associated with the terminal differentiation of osteoclast precursor cells to bone-invasive clivus chordoma by an iTRAQ-based quantitative stimulate and maintain resorption in mature cells. In this study, proteomic strategy. Lorenzo and colleagues (25) reviewed the expression of RANKL mRNA and protein on immunohis- the important role that immune cells and many proinflammatory tochemistry was low in BIPAs, and there was no significant factors play in bone metabolism, including bone destruction. difference between the groups. Thus, the bone destruction of Many proinflammatory factors play important roles in

www.aacrjournals.org Clin Cancer Res; 2018 OF7

Zhu et al.

A BCBIPAs BIPAs NBIPAs NBIPAs 5 150

100 4

50 3 15 2 10 5 1 Relative expression of IncRNAs (fold change) Relative expression of IncRNAs (fold change) 0 0

NR_033258 IncRNA H19 miR 454-3p Inc-DLK1-6:9 miR 150-5pmiR 181c-5pmiR 181c-3p Inc-EIF 3H-3:1 miR 130b-3p Inc-PLXDC2-3:1 IncRNAInc-CLST SNHG 24 N2-1:1

IncRNA CTD-2147F2.1 IncRNA RNU4ATAC18P

NC NC DEFsh-Lnc-EIF3H-3:1 miR 150 inhibitor NC sh-NR_033258 miR 181c inhibitor TNFα sh-Lnc-SNHG24 miR 454 inhibitor 400 1.5 2.0 300

1.0 1.5 200 1.0

0.5 α (fold change) 100 0.5 Elisa (TNF α , pg/mL)

and TNF α (fold change) 0.0 0 and TNF 0.0 α Relative expression of IncRNAs Relative expression of miRNAs TNF TNFα LncRNAs miRNAs

sh-NR_033258 sh-Lnc-SNHG24 sh-Lnc-ElF3H-3:1 miR 150 miRinhibitor 181c inhibitormiR 454 inhibitor

Figure 5. NR_033258, lncRNA SNHG24, miR-181c-5p, and miR-454-3p can regulate TNFa expression. A, TNF-related ceRNA network: red nodes indicate an increased level of expression, while green nodes indicate a decreased level of expression. lncRNAs and microRNAs are indicated as oval and diamond symbols, respectively. The red line represents positive correlation and the green line represents negative correlation. B, qRT-PCR: the relative expression levels of nine lncRNAs. C, qRT-PCR: the relative expression levels of ﬁve miRNAs. D, The relative expression of lncRNAs and TNFa after treatment with shRNAs or NC. E, The relative expression of miRNAs and TNFa after treatment with miRNA inhibitors or NC. F, The relative protein expression of TNFa after treatment with shRNAs, miRNA inhibitors or NC. , P < 0.05; , P < 0.01; , P < 0.001.

osteoclastogenesis, such as IL1, IL6, TNFa, IL8, and other che- and colleagues (37) reported that progranulin plays crucial roles mokines (CCL3, CCL9, CCL2, and CXCL12; refs. 26–28). In this in preserving bone mass by inhibiting TNFa-induced osteoclas- study, the expression of TNFa, CCL3, CXCL12, and CCL2 was togenesis and promoting osteoblastic differentiation in mice. higher in the BIPAs than in the NBIPAs. Zhao and colleagues (38) reported that the transcription factor TNFa is an important inflammatory factor, with a variety of RBP-J can inhibit TNF-induced osteoclastogenesis and inflamma- biological functions, including regulating inflammatory tory bone resorption. TNFa may be a potential therapeutic target responses, immune responses, apoptosis, and antiviral for BIPAs in the future. responses. TNFa stimulates osteoclast differentiation in vitro miRNAs are a class of noncoding RNAs that regulate gene (29) and in vivo (30). The ability of TNFa to stimulate osteo- expression by binding to the 30-untranslated region (UTR) of clast formation in vitro culture systems is independent of RANK/ their target messenger RNAs (mRNA) and conduct translational RANKL (31). TNFa can induce osteoclast differentiation by repression or degradation of mRNA (39). miRNAs play an impor- activating MAPK and NF-kB pathways (32). TNFa can promote tant role in the regulation of animal development and are the expression of osteolytic cytokines, such as M-CSF and involved in human diseases, including developmental timing, RNAKL (33). This study showed that the expression level of growth control, differentiation, embryogenesis, programmed cell TNFa in BIPAs was higher than that in NBIPAs and that TNFa death, insulin secretion, immunodeficiency, and cancer develop- can directly induce osteoclast differentiation. ment (40). There are few reports on miRNA and bone invasion TNFa has been highlighted as a target for the treatment of (12–14). This study was the first to examine microRNAs and inflammatory bone diseases, and it is expected that an under- BIPAs and showed that miR-181c-5p and miR-454-3p can play an standing of molecules that mediate TNFa signals will provide important role in osteoclastogenesis by regulating TNFa. Zhang clues to develop new therapies for such diseases. TNFa recognizes and colleagues (41) reported that miR-181c-5p could directly two receptors, TNF-R1 (p55) and TNF-R2 (p75) (34), in both target the 30-UTR of TNFa mRNA, suppressing its mRNA and humans and mouse. TNF-R1 promotes osteoclastogenesis (35) protein expression after ischemia/hypoxia and microglia-medi- and the stimulatory effect of TNFa can be completely prevented ated neuronal injury. Zhang and colleagues (42) also showed that by the anti-p55 antibody. Blocking of TNF-R2 with the anti-p75 miR-181c-5p suppresses TLR4 by directly binding to its 30-UTR, antibody only partially inhibits osteoclastogenesis (36). Noguchi thereby inhibited NF-kB activation and the downstream

OF8 Clin Cancer Res; 2018 Clinical Cancer Research

The Prognosis and Mechanism of BIPAs

production of proinflammatory mediators, such as TNFa, IL1b, Disclosure of Potential Conflicts of Interest and iNOS. Wu and colleagues (43) reported that miR-454-3p can No potential conflicts of interest were disclosed. target BTG1 to render tumor cells sensitive to radiation, and Shao and colleagues reported that plasma miR-454-3p can be a poten- Authors' Contributions tial prognostic indicator in human glioma (44). Conception and design: H. Zhu, H. Gao, C. Li, Y. Zhang lncRNA are involved in diverse biological processes, includ- Development of methodology: H. Zhu, W. Dong, C. Li ing the regulation of proliferation, apoptosis, and invasiveness Acquisition of data (provided animals, acquired and managed patients, of tumors and reprogramming of induced pluripotent stem provided facilities, etc.): H. Zhu, Y. Miao Analysis and interpretation of data (e.g., statistical analysis, biostatistics, cells (45). Recent studies have reported that lncRNAs play computational analysis): H. Zhu important roles in cancer development, metastasis and chemo- Writing, review, and/or revision of the manuscript: H. Zhu, C. Li, Y. Zhang therapy resistance (46). Studies of the role of lncRNAs in bone Administrative, technical, or material support (i.e., reporting or organizing invasion are rare. Liu and colleagues reported that lncRNA data, constructing databases): H. Zhu, J. Guo, Y. Shen, W. Dong MALAT1 was upregulated in NSCLC tissues with bone metas- Study supervision: Y. Zhang tasis and in lung cancer cell lines with high bone metastatic abilities (15). Li and colleagues reported that a ROR1–HER3– Acknowledgments LncRNA signaling axis modulates the Hippo–YAP pathway to This study was supported by the National Natural Science Foundation of regulate bone metastasis (16). Wang and colleagues (47) indi- China (81771489); supported by Beijing Municipal Science & Technology Commission (Z171100000117002). We thank Dr. Chengcheng Wang (Etiol- cated that MEG3 regulated the expression of miR-133a-3p and ogy Laboratory, National Cancer Center/Cancer Hospital Chinese Academy of inhibited the osteogenic differentiation of BMSCs induced by Medical Sciences) for support with experiments. We thank Dr. Zhexuan Li PMOP. This study showed that NR_033258 and lncRNA (Epidemiological Laboratory, Beijing Cancer Hospital/Beijing Institute For SNHG24 can regulate TNFa expression. However, there are Cancer Research) for support of statistics. We thank Mr. Lei Gong and Mrs. fewer reports on the three lncRNAs. Hongyun Wang (Cell Laboratory, Beijing Neurosurgical Institute) for support In conclusion, BIPAs had worse PFS than did NBIPAs in the with technique. NGTR and NFPA groups. Inflammatory and immune factors play The costs of publication of this article were defrayed in part by the payment of an important role in BIPAs. TNFa can directly induce osteoclast page charges. This article must therefore be hereby marked advertisement in differentiation in BIPAs. NR_033258, lncRNA SNHG24, miR- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 181c-5p, and miR-454-3p can regulate TNFa expression. TNFa and its related lncRNAs and miRNAs represent potential thera- Received February 11, 2018; revised May 7, 2018; accepted July 2, 2018; peutic targets for bone-invasive pituitary adenomas in the future. published first July 6, 2018.

References 1. Fernandez A, Karavitaki N, Wass JA. Prevalence of pituitary adenomas: a 12. Cong F, Wu N, Tian X, Fan J, Liu J, Song T, et al. MicroRNA-34c community-based, cross-sectional study in Banbury (Oxfordshire, UK). promotes osteoclast differentiation through targeting LGR4. Gene Clin Endocrinol (Oxf) 2010;72:377–82. 2017;610:1–8. 2. Melmed S. Pathogenesis of pituitary tumors. Nat Rev Endocrinol 2011; 13. Lee Y, Kim HJ, Park CK, Kim YG, Lee HJ, Kim JY, et al. MicroRNA-124 7:257–66. regulates osteoclast differentiation. Bone 2013;56:383–9. 3. Chen X, Dai J, Ai L, Ru X, Wang J, Li S, et al. Clival invasion on multi-detector 14. Sugatani T, Hruska KA. MicroRNA-223 is a key factor in osteoclast differ- CT in 390 pituitary macroadenomas: correlation with sex, subtype and entiation. J Cell Biochem 2007;101:996–9. rates of operative complication and recurrence. AJNR Am J Neuroradiol 15. Liu M, Sun W, Liu Y, Dong X. The role of lncRNA MALAT1 in bone 2011;32:785–9. metastasis in patients with non-small cell lung cancer. Oncol Rep 4. Granchi D, Corrias MV, Garaventa A, Baglio SR, Cangemi G, Carlini B, et al. 2016;36:1679–85. Neuroblastoma and bone metastases: clinical significance and prognostic 16. Li C, Wang S, Xing Z, Lin A, Liang K, Song J, et al. A ROR1-HER3-lncRNA value of Dickkopf 1 plasma levels. Bone 2011;48:152–9. signalling axis modulates the Hippo-YAP pathway to regulate bone metas- 5. Takayama Y, Mori T, Nomura T, Shibahara T, Sakamoto M. Parathyroid- tasis. Nat Cell Biol 2017;19:106–19. related protein plays a critical role in bone invasion by oral squamous cell 17. Willingham AT, Orth AP, Batalov S, Peters EC, Wen BG, Aza-Blanc P, et al. A carcinoma. Int J Oncol 2010;36:1387–94. strategy for probing the function of noncoding RNAs finds a repressor of 6. Moon HS, Jung S, Jung TY, Cao VT, Moon KS, Kim IY. Possible role of matrix NFAT. Science 2005;309:1570–3. metalloproteinase in osteolytic intracranial meningiomas. J Korean Neu- 18. Woodworth GF, Patel KS, Shin B, Burkhardt JK, Tsiouris AJ, McCoul ED, rosurg Soc 2010;47:11–6. et al. Surgical outcomes using a medial-to-lateral endonasal endoscopic 7. Capietto AH, Faccio R. Immune regulation of bone metastasis. Bonekey approach to pituitary adenomas invading the cavernous sinus. J Neurosurg Rep 3:600. 2014;120:1086–94. 8. Ylitalo EB, Thysell E, Jernberg E, Lundholm M, Crnalic S, Egevad L, et al. 19. Dehdashti AR, Ganna A, Karabatsou K, Gentili F. Pure endoscopic endo- 2017; subgroups of castration-resistant prostate cancer bone metastases nasal approach for pituitary adenomas: early surgical results in 200 patients defined through an inverse relationship between androgen receptor activity and comparison with previous microsurgical series. Neurosurgery and immune response. Eur Urol 2014;71:776–87. 2008;62:1006–15; discussion 1015–1007. 9. Walsh MC, Kim N, Kadono Y, Rho J, Lee SY, Lorenzo J, et al. Osteoimmu- 20. Ratnasingam J, Lenders N, Ong B, Boros S, Russell AW, Inder WJ, et al. nology: interplay between the immune system and bone metabolism. Predictors for secondary therapy after surgical resection of nonfunctioning Annu Rev Immunol 2006;24:33–63. pituitary adenomas. Clin Endocrinol (Oxf) 2017;87:717–24. 10. Quan J, Elhousiny M, Johnson NW, Gao J. Transforming growth factor- 21.TampourlouM,NtaliG,AhmedS,ArltW,AyukJ,ByrneJV,etal. beta1 treatment of oral cancer induces epithelial-mesenchymal transition Outcome of nonfunctioning pituitary adenomas that regrow after and promotes bone invasion via enhanced activity of osteoclasts. Clin Exp primary treatment: a study from two large UK centers. J Clin Endocrinol Metastasis 2013;30:659–70. Metab 2017;102:1889–97. 11. Nanes MS.Tumor necrosis factor-alpha: molecular and cellular mechan- 22. Boyle WJ, Simonet WS, Lacey DL. Osteoclast differentiation and activation. isms in skeletal pathology. Gene 2003;321:1–15. Nature 2003;423:337–42.

www.aacrjournals.org Clin Cancer Res; 2018 OF9

Zhu et al.

23. Horowitz MC, Lorenzo JA. Local regulators of bone. IL-1, TNF, lym- 35. Abu-Amer Y, Erdmann J, Alexopoulou L, Kollias G, Ross FP, Teitelbaum SL. photoxin, interferon-g, the LIF/IL-6 family, and additional cytokines. Tumor necrosis factor receptors types 1 and 2 differentially regulate In: Principles of Bone Biology; 2002. 961–77. osteoclastogenesis. J Biol Chem 2000;275:27307–10. 24. Wu Z, Wang L, Guo Z, Wang K, Zhang Y, Tian K, et al. Experimental study on 36. Kobayashi K, Takahashi N, Jimi E, Udagawa N, Takami M, Kotake S, et al. differences in clivus chordoma bone invasion: an iTRAQ-based quantita- Tumor necrosis factor alpha stimulates osteoclast differentiation by a tive proteomic analysis. PLoS One 2015;10:e0119523. mechanism independent of the ODF/RANKL-RANK interaction. J Exp 25. Lorenzo J, Horowitz M, Choi Y. Osteoimmunology: interactions of the Med 2000;191:275–86. bone and immune system. Endocr Rev 2008;29:403–40. 37. Noguchi T, Ebina K, Hirao M, Kawase R, Ohama T, Yamashita S, et al. 26. Hofbauer LC, Lacey DL, Dunstan CR, Spelsberg TC, Riggs BL, Khosla S. Progranulin plays crucial roles in preserving bone mass by inhibiting TNF- Interleukin-1beta and tumor necrosis factor-alpha, but not interleukin-6, alpha-induced osteoclastogenesis and promoting osteoblastic differentia- stimulate osteoprotegerin ligand gene expression in human osteoblastic tion in mice. Biochem Biophys Res Commun 2015;465:638–43. cells. Bone 1999;25:255–9. 38. Zhao B, Grimes SN, Li S, Hu X, Ivashkiv LB. TNF-induced osteoclastogen- 27. Palmqvist P, Persson E, Conaway HH, Lerner UH. IL-6, leukemia inhibitory esis and inflammatory bone resorption are inhibited by transcription factor factor, and oncostatin M stimulate bone resorption and regulate the RBP-J. J Exp Med 2012;209:319–34. expression of receptor activator of NF-kappa B ligand, osteoprotegerin, 39. Farh KK, Grimson A, Jan C, Lewis BP, Johnston WK, Lim LP, et al. The and receptor activator of NF-kappa B in mouse calvariae. J Immunol widespread impact of mammalian MicroRNAs on mRNA repression and 2002;169:3353–62. evolution. Science 2005;310:1817–21. 28. Campbell JJ, Butcher EC. Chemokines in tissue-specificandmicroen- 40. Alvarez-Garcia I, Miska EA. MicroRNA functions in animal development vironment-specific lymphocyte homing. Curr Opin Immunol 2000;12: and human disease. Development 2005;132:4653–62. 336–41. 41. Zhang L, Dong LY, Li YJ, Hong Z, Wei WS. The microRNA miR-181c 29. Thomson BM, Mundy GR, Chambers TJ. Tumor necrosis factors alpha and controls microglia-mediated neuronal apoptosis by suppressing tumor beta induce osteoblastic cells to stimulate osteoclastic bone resorption. J necrosis factor. J Neuroinflammation 2012;9:211–23. Immunol 1987;138:775–9. 42. Zhang L, Li YJ, Wu XY, Hong Z, Wei WS. MicroRNA-181c negatively 30. Konig A, Muhlbauer RC, Fleisch H. Tumor necrosis factor alpha and regulates the inflammatory response in oxygen-glucose-deprived microglia interleukin-1 stimulate bone resorption in vivo as measured by urinary by targeting Toll-like receptor 4. J Neurochem 2015;132:713–23. [3H]tetracycline excretion from prelabeled mice. J Bone Miner Res 43. Wu X, Ding N, Hu W, He J, Xu S, Pei H, et al. Down-regulation of BTG1 by 1988;3:621–7. miR-454–3p enhances cellular radiosensitivity in renal carcinoma cells. 31. Kim N, Kadono Y, Takami M, Lee J, Lee SH, Okada F, et al. Osteoclast Radiat Oncol 2014;9:179–90. differentiation independent of the TRANCE-RANK-TRAF6 axis. J Exp Med 44. Shao N, Wang L, Xue L, Wang R, Lan Q. Plasma miR-454–3p as a potential 2005;202:589–95. prognostic indicator in human glioma. Neurol Sci 2015;36:309–13. 32. Lu C, Gao S, Xu G. Geraniin inhibits TNF-alpha-induced impairments of 45. Harries LW. Long non-coding RNAs and human disease. Biochem Soc osteogenesis through NF-kappaB and p38 MAPK signalling pathways in Trans 2012;40:902–6. bone marrow stem cells. Stroke Vasc Neurol 2017;2:47–52. 46. Zhou Y, Zhang X, Klibanski A. MEG3 noncoding RNA: a tumor suppressor. 33. Kubota A, Hasegawa K, Suguro T, Koshihara Y. Tumor necrosis factor-alpha J Mol Endocrinol 2012;48:R45–53. promotes the expression of osteoprotegerin in rheumatoid synovial fibro- 47. Wang Q, Li Y, Zhang Y, Ma L, Lin L, Meng J, et al. LncRNA MEG3 inhibited blasts. J Rheumatol 2004;31:426–35. osteogenic differentiation of bone marrow mesenchymal stem cells from 34. Vandenabeele P, Declercq W, Beyaert R, Fiers W. Two tumour necrosis postmenopausal osteoporosis by targeting miR-133a-3p. Biomed Phar- factor receptors: structure and function. Trends Cell Biol 1995;5:392–9. macother 2017;89:1178–86.

OF10 Clin Cancer Res; 2018 Clinical Cancer Research

Functions and Mechanisms of Tumor Necrosis Factor-α and Noncoding RNAs in Bone-Invasive Pituitary Adenomas

Haibo Zhu, Jing Guo, Yutao Shen, et al.

Clin Cancer Res Published OnlineFirst July 6, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-18-0472

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/07/06/1078-0432.CCR-18-0472.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2018/09/06/1078-0432.CCR-18-0472. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.