We first examined the expression of three lncRNAs, C5orf66-AS1, NORAD, and TINCR, in different normal adult tissues by using RNA-seq data from the Genotype-Tissue Expression (GTEx) Project (http://www.gtexportal.org/home/). The data described in this study were retrieved from the GTEx Portal 11/17/2016 (GTEx Analysis Release V6p) and dbGaP (accession no. phs000424.v6.p1) 11/17/2016.

Patients who underwent endoscopic trans-sphenoidal surgery from March 2013 to November 2014 at Beijing Tiantan Hospital and who were diagnosed with pituitary null cell adenomas through immunostaining and electron microscopy were enrolled in the present study. Inclusion criteria were results of electron microscopy showing cells with relatively few, poorly developed organelles; absence of sufficiently distinctive structural characteristics to determine the cellular origin of the tumor; and negative results for the immunostaining of anterior pituitary hormones, which was consistent with the definition of pituitary null cell adenoma described above and according to the World Health Organization classification (2004 version) (2,3,15). Patients with a family history of pituitary adenoma or other tumors of the endocrine system were excluded. Each patient underwent a hormonal serum test and magnetic resonance imaging (MRI) before undergoing surgery. The excised tissues were examined by performing histopathological and immunohistochemical analysis for all anterior pituitary hormones and by electron microscopy. Tumors with Hardy-Wilson classification grade IV and/or Knosp classification grades III and IV (16–18) were defined as invasive. Invasiveness was determined by a group of neurosurgeons, radiologists, and pathologists. Maximum tumor diameter was measured using gadolinium (Gd)-enhanced T1WI MR images. Normal pituitary tissues were obtained from donors who died from diseases other than neurologic or endocrine diseases. All patients provided written informed consent, and the study was approved by the Ethics Committee of the Beijing Tiantan Hospital.

Pituitary adenoma and normal anterior pituitary tissues were stored in liquid nitrogen immediately after removal. Total RNA was extracted from the frozen tissue samples by using TRizol (Invitrogen, Carlsbad, CA, USA), according to the manufacturer's instructions. Integrity of the total RNA was determined by performing agarose gel electrophoresis. Only samples showing no RNA degradation (28S/18S ≥0.7) were used to generate labeled targets. RNA purity and concentration were determined using NAS-99 (Merinton, Beijing, China) and A260:A280 ratios yielded were consistently close to 2.0. Total RNA was reverse transcribed to cDNA by using HiFiScript gDNA Removal cDNA Synthesis kit (CWBio Co., Ltd., Beijing, China), according to the manufacturer's instructions.

qRT-PCR was performed using StepOnePlus™ Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) and Kapa SYBR FAST qPCR Master Mix ABI Prism™ (2X, KK4601; Kapa Biosystems, Inc., Wilmington, MA, USA), according to the manufacturer's instructions. SYBR Green assays were performed using 0.5 µl PCR forward primer (10 µM), 0.5 µl PCR reverse primer (10 µM), 2 µl cDNA, 5 µl 2X master mix, and 2 µl double distilled water in a total reaction volume of 10 µl. The PCR protocol was initiated at 95°C for 3 min, followed by 40 cycles of 95°C for 10 sec and 60°C for 60 sec. GAPDH was used as a reference gene. Results were obtained using three independent wells. Relative expression level was calculated using 2^−∆∆Ct method (19) after normalizing with GAPDH expression level. Sequences of primers used for performing qRT-PCR are presented in Table I.

Two algorithms were used to predict possible cis- and trans-acting target genes, as previously described (20). Cis-acting target genes were predicted using UCSC genome browser and genome annotation to locate the nearest known protein-coding genes from the lncRNAs (21). According to distance stratification, protein-coding genes located within a distance of 5 and 300 kb from the lncRNAs were captured as targets. Trans-acting genes were predicted using RNAplex software (http://www.bioinf.uni-leipzig.de/Software/RNAplex/). RNAplex can determine possible hybridization sites for a query RNA according to sequence complementarity and RNA duplex-binding energy prediction (22,23). First, BLAST (e<1E-5) was performed using the lncRNAs and known protein-coding genes; next, RNAplex (parameter set as -e −70) was used to screen trans-acting target genes.

Expression of predicted target genes was determined using microarray data of another cohort examined in our laboratory. Microarray analysis was performed using Whole Human Genome Oligo Microarray (4×44K) (Agilent Technologies, Santa Clara, CA, USA). Total RNA was extracted and was purified using mirVana™ miRNA Isolation kit (catalog no. AM1561; Ambion, Austin, TX, USA), according to the manufacturer's instructions. RIN was determined using Bioanalyzer 2100 (Agilent Technologies) to inspect RNA integration. None of the samples showed RNA degradation (RIN ≥7.0 and 28S/18S ≥0.7). Total RNA was amplified and was labeled using Low Input Quick Amp labeling kit, One-Color (catalog no. 5190-2305; Agilent Technologies). Labeled cRNA was purified using RNeasy mini kit (catalog no. 74106; Qiagen, GmBH, Hilden, Germany). Next, each slide was hybridized with Cy3-labeled cRNA by using Gene Expression Hybridization kit (catalog no. 5188-5242; Agilent Technologies), according to the manufacturer's instructions. After hybridization, the slides were scanned using a microarray scanner (catalog no. G2565CA; Agilent Technologies). Data were extracted using Feature Extraction software 10.7 (Agilent Technologies), and raw data were normalized using Quantile algorithm, limma packages in R.

We previously performed RNA sequencing of a cohort with various types of pituitary adenomas, which is under study in our laboratory. After the total RNA extraction and DNase I treatment, magnetic beads with Oligo (dT) were used to isolate mRNAs. mRNAs were fragmented into short fragments by mixing with fragmentation buffer. Then, cDNAs were synthesized using the mRNA fragments as templates. Fragments were purified and resolved with EB buffer for end reparation and single nucleotide adenine addition and were connected with adapters. After agarose gel electrophoresis, the suitable fragments were selected for PCR amplification as templates. During the quality control steps, the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and ABI StepOnePlus Real-time PCR System (Applied Biosystems) were used in quantification and qualification of the sample library. Finally, the library was sequenced using the Illumina HiSeq 2000 (Illumina, San Diego, CA, USA) at the BGI-Tech Bioinformatics Institute (Shenzhen, China).

The mouse pituitary cell line GT1-1 was purchased from the China Infrastructure of Cell Line Resources and cultured in phenol red-free Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (Gibco, Auckland, New Zealand) in a humidified incubator at 37°C with 5% CO₂. pUC57-C5orf66-AS1 was obtained from BGI (Shenzhen, China). Transfection was performed using Lipofectamine 3000 (Invitrogen) according to the manufacturer's instructions. After 24 h, C5orf66-AS1 was examined using qRT-PCR in triplicates.

Log-phase GT1-1 cells were harvested and confirmed to be >99% viable by trypan blue exclusion and adjusted to a density of 1×10⁵ cells/ml. A volume of 100 µl of cell suspension was plated into 96-well plates and cultured for 24 h. After transfection for 24, 48, and 72 h, 20 µl MTS solution was added to each well and further incubated for 4 h. Absorbance of each well at 490 nm was measured using an ELISA plate reader (Thermo Labsystems, Franklin, MA, USA).

Tumor cell invasion was measured by examining cell migration and invasion through fibronectin- and Matrigel-coated polycarbonate filters, respectively, using modified Transwell chambers (Corning Inc., Corning, NY, USA). GT1-1 cells (2×10⁴ cells) were added into the upper chambers. Migrated cells adhered to the lower membrane were fixed in 4% paraformaldehyde and stained with hematoxylin (Zhongshan Company, Beijing, China). The average number of migrated cells was quantified by counting five random high-power fields under a fluorescent microscope (Carl Zeiss, Jena, Germany). Experiments were performed in triplicates.

Statistical analysis of qRT-PCR results was performed using SPSS software (v23.0; IBM Corp., Armonk, NY, USA). Shapiro-Wilk test was used to test normality. Differences between the groups were determined using Mann-Whitney U test. Correlation between lncRNA expression level and maximum tumor diameter and between lncRNA expression level and patient age was determined using Spearman's rank correlation coefficient or Pearson's correlation coefficient on the basis of data distribution. Relationship between C5orf66-AS1 expression level and maximum tumor diameter in male patients was determined by performing simple linear regression analysis after calculating Pearson's correlation coefficient. Co-expression analysis was performed with R based on the Pearson correlation coefficients between C5orf66-AS1 and all mRNAs with unique mapping from RNA-seq data after ruling out types of pituitary adenomas other than the null cell type. P≤0.05 was considered statistically significant, except in the Shapiro-Wilk test, in which 0.1 was used as the level of significance to decrease the probability of false-negative errors.

First, we examined the expression of three lncRNAs, C5orf66-AS1, NORAD, and TINCR, in different normal tissues by using RNA-seq data from GTEx. We found that C5orf66-AS1 was overexpressed specifically in the pituitary gland, with its expression level being only lower than that in esophageal mucosa among the 53 tissue types examined (Fig. 1). TINCR was expressed at an average level, and no data are available on the expression of NORAD. These results suggest that C5orf66-AS1 performs special functions in the pituitary gland.

qRT-PCR analysis of TINCR, C5orf66-AS1, and NORAD was performed using normal pituitary tissues from four donors and tumor tissues from 11 patients, including six men and five women. Characteristics of the patients are shown in Table II. Ages of the patients ranged from 36 to 74 years, with an average age of 53.3 years. The main complaints of the patients were headache and visual disturbances; in some patients, pituitary null cell adenoma was diagnosed by performing incident MRI or CT scan. Time interval from the appearance of the first symptoms to surgery varied from 20 days to 9 years. For each patient, serum test yielded negative results for excessive pituitary hormone secretion, the sella turcica showed enlargement, and radiologic performance was consistent with the diagnosis of pituitary adenoma. Tumors of six patients were invasive; of these, three patients had Hardy-Wilson grade III and Knosp classification grade III tumors and the remaining three patients had Hardy-Wilson grade IV and Knosp classification grade IV tumors (Fig. 2). Maximum tumor diameters ranged from 23.0 to 57.2 mm. No differences were observed in maximum tumor diameters between male and female patients. All the tumors were diagnosed as pituitary null cell adenomas, with no major difficulty; some of these tumors showed oncocytic changes.

After predicting target genes, we examined some target genes by using microarray data obtained from pituitary null cell adenomas of 16 patients and normal pituitary tissues of six donors from another study cohort. Patients in this cohort were admitted during the same period and were diagnosed using the same criteria as those described above. Of the 16 patients, 12 were men and 4 were women. Moreover, six patients had invasive tumors, and nine patients had non-invasive tumors (invasiveness of the tumor of one patient could not be evaluated because of the absence of MRI data). Other clinical characteristics of these patients were similar to those of the 11 patients included in the present study. Then, we performed a co-expression analysis of the RNA-seq data of another cohort, including 12 pituitary null cell adenomas with other types of pituitary adenomas and five normal pituitaries. Of the 12 patients of pituitary null cell adenomas, six were men and six were women; seven patients had invasive tumors, and 5 patients had non-invasive tumors. Other clinical characteristics of these patients were similar to those of the 11 patients included in the present study.

C5orf66-AS1 expression level was significantly lower in pituitary null cell adenoma tissues than in normal pituitary tissues (Mann-Whitney U test, Z= −2.089; two-tailed, P=0.040; Table II, Fig. 3A). Moreover, C5orf66-AS1 expression level was significantly lower in invasive tumors than in noninvasive tumors (Mann-Whitney U test, Z= −2.739; two-tailed, P=0.004; Table II, Fig. 3B).

No correlation was observed between C5orf66-AS1 expression level and patient age. Analysis of the relationship between C5orf66-AS1 expression and pituitary tumor growth showed that C5orf66-AS1 lncRNA levels were negatively correlated with maximum tumor diameter (Spearman's rank correlation coefficient r_s= −0.601, P=0.05; Table II, Fig. 4A). Stratification of tumors based on sex showed a negative linear correlation between C5orf66-AS1 lncRNA levels and maximum tumor diameter in male patients (Pearson correlation: r= −0.884, P=0.019; simple linear regression analysis: R²=0.782, adjusted R²=0.728, F=14.360, P=0.019, Y=0.354–0.009X, t= −3.790, P=0.019; Table II). However, this linear correlation was not observed in female patients. Moreover, no significant difference was observed in C5orf66-AS1 expression between male and female patients. Stratification of tumors based on their invasive potential showed a significant positive linear correlation between C5orf66-AS1 expression and patient age in patients with non-invasive tumors (Pearson correlation, r=0.922, P=0.009; Table II). However, no correlation was observed between maximum tumor diameter and patient age; and this linear correlation was not observed in patients with invasive tumors.

NORAD and TINCR expression did not show normal distribution in the tumor samples of the complete cohort examined in the study. No difference was observed in NORAD and TINCR expression levels between tumor and normal pituitary tissues and between invasive and non-invasive tumors. Moreover, no correlation was observed between NORAD and TINCR expression levels and maximum tumor diameter or patient age. However, in non-invasive pituitary null cell tumors, NORAD lncRNA level was significantly negatively correlated with maximum tumor diameter (Spearman's rank correlation, r_s= −0.943, P=0.005; Table II). Furthermore, NORAD lncRNA level was significantly negatively correlated with maximum tumor diameter (Spearman's rank correlation, r_s= −0.886, P=0.019; Table II, Fig. 4B) but significantly positively correlated with age (Spearman's rank correlation, r_s=0.943, P=0.005; Table II, Fig. 4C) in male patients. However, no correlation was observed between maximum tumor diameter and age in male patients.

Nine cis-acting target genes of C5orf66-AS1 were detected in a distance of <300 kb from C5orf66-AS1 (Table III). Of these, the top two genes, namely, C5orf66 and PITX1, were within a distance of 5 kb from C5orf66-AS1. Location of C5orf66-AS1 is in chromosome 5: 135,038,831-135,040,047 (reverse strand), while that of C5orf66 is in chromosome 5: 135,033,280-135,344,680 (forward strand) and PITX1 is chromosome 5: 135,027,734-135,034,274 (reverse strand) (in assembly of GRCh38.p7/GCF_000001405.33, Fig. 5). Furthermore, we predicted 347 trans-acting target genes of C5orf66-AS1. The top 20 trans-acting genes with the best interaction energy are listed in Table III.

Next, we examined the predicted target genes (nine cis-acting genes and top 20 trans-acting genes) in microarray data of another study cohort (16 patients with pituitary null cell adenomas and six donors with normal pituitary gland) examined in our laboratory. By using a stringency of P≤0.01 and fold change of >1.5, we found that 3 of the 9 predicted cis-acting genes and 8 of the 20 predicted trans-acting genes were differentially expressed between tumor tissues and normal pituitary tissues (Table III). Moreover, by using a stringency of P≤0.05 and fold change of >1.5, we found that only SCGB3A1 was significantly differentially expressed in invasive and non-invasive pituitary null cell adenomas (P=0.046, fold change=6.817). Then, we performed qRT-PCR to validate expression of SCGB3A1 in pituitary null cell adenomas and in normal pituitary tissues. The result showed that SCGB3A1 was significantly underexpressed in pituitary adenomas compared with normal pituitaries (P=0.030) and was significantly underexpressed in invasive tumors compared with non-invasive ones (P=0.041).

Co-expression analysis of C5orf66-AS1 in the RNA-seq data showed that among the total 19,378 mRNAs which were uniquely mapped to, there were 32 mRNAs with r>0.90 or r< −0.90 (P<0.001) and 3,002 mRNAs with r>0.70 or <-0.70 (P<0.001). The most correlated gene was PAQR7 (r=0.938, P<0.001). In the top 20 trans-acting genes we predicted in silico, there were eight genes with r>0.70 (P<0.001). They were NME6 (r=0.851), RASSF1 (r=0.822), EVC2 (r=0.821), COL7A1 (r=0.746), POLR1E (r=0.730), SCGB3A1 (r=0.730), LGI3 (r=0.728), and GTF2IRD2 (r=0.723). Out of these eight genes, seven genes (NME6, RASSF1, EVC2, COL7A1, SCGB3A1, LGI3, and GTF2IRD2) were differentially expressed between tumor tissues and normal pituitary tissues.

To confirm the tumor suppressor role of C5orf66-AS1, we first overexpressed it in mouse pituitary cell line GT1-1 by transfection. We performed qRT-PCR to examine expression of C5orf66-AS1 before transfection and at 24 h after transfection. The results showed that C5orf66-AS1 was significantly overexpressed after transfection. A cell viability assay was performed at 24, 48, and 72 h after transfection. The result showed that after C5orf66-AS1 transfection, the viability of tumor cells was significantly inhibited compared with that of the control group in a time-dependent manner. After that, we performed the Transwell invasion assay. The result showed that overexpression of C5orf66-AS1 markedly reduced the effect of invasion.