This article is mentioned in:

Introduction

Pituitary adenomas, accounting for ~15% of all diagnosed intracranial tumors, are benign monoclonal adenomas that originate from cells of the anterior pituitary gland (1). Surgical resection, with or without adjuvant radiotherapy, is always the first line of treatment for the majority of pituitary adenomas, with the exception of prolactinomas (2). However, these treatments cannot usually control invasive pituitary adenomas due to the limited understanding of the underlying molecular mechanisms. Thereby, further research into the tumorigenesis will contribute to identifying novel therapeutic targets, which will be conductive to the development of novel therapeutic approaches for pituitary adenomas.

In past years, considerable progress has been made in identifying the key players in pituitary adenomas. A previous study has shown that the phosphoinositide 3-kinase/AKT signaling pathway is activated and enhanced in pituitary adenomas, which may be due to the mutation and amplification of an oncogene, PIK3CA (3). Mutation in another oncogene, GNAS, which encodes the guanine nucleotide-activating α subunit has also been suggested to be involved in pituitary hyperplasia (4). Meanwhile, a tumor suppressor aryl hydrocarbon receptor-interacting protein has been demonstrated to function in modulating cellular signaling and cAMP signaling pathways via regulation of the localization of the aryl hydrocarbon receptor (5). Also, the absence of expression of another two tumor suppressors, growth arrest and DNA-damage-inducible β (GADD45β) and γ (GADD45γ), has been observed in human pituitary adenomas (6,7). Aberrant methylation of a number of genes, such as DAPK (8) and FGFR2 (9) has been confirmed to have a momentous role in pituitary tumorigenesis. Additionally, certain cell cycle regulators, such as p16, p21, p27, cyclin D1 and cyclin E, have also been demonstrated to function in pituitary tumorigenesis (8,10). Recently, certain microRNAs (miRNA/miR) have been found to be crucial in pituitary adenomas. For instance, the expression levels of miR-431 and miR-770-5p have been found to be slightly higher in non-functioning pituitary adenomas compared to their levels in the normal pituitary gland (11). Recently, another study has shown that miRNA-dependent impairment of the HMGA/E2F1 pathway functions as pro-oncogene signaling in pituitary adenomas. Several miRNAs targeting HMGA2 (miR-326, miR-570 and miR-432) or E2F1 (miR-326 and miR-603) could inhibit the growth of pituitary cell lines (HP75 and GH3) (12).

Lee et al demonstrated that gonadotroph adenomas in MENX-affected rats closely resemble their human counterparts (13). The study further found that CYP11A1 and NUSAP1, two commonly dysregulated differentially-expressed genes (DEGs) in the gonadotroph adenomas of rats and humans, are upregulated in 77 and 95% of human gonadotroph adenomas, respectively. Using the microarray data deposited by Lee et al, the present study aimed to further identify genes that were differentially expressed between pituitary adenomas samples and normal controls. Following Gene Ontology (GO) functional and pathway enrichment analysis of the screened DEGs, Protein-Protein Interaction (PPI) networks were constructed for the up- and downregulated DEGs, respectively, in order to learn more about the interaction of proteins encoded by DEGs, which may aid in our understanding of the molecular mechanisms of pituitary adenomas. The results are expected to assist in elucidating the etiology of pituitary adenomas, and provide novel insights for the clinical diagnosis of this disease.

Materials and methods

Affymetrix microarray data

The gene expression profile data of GSE23207 (13) were acquired from Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/), which was based on the platform of the GPL6247 [RaGene-1_0-st] Affymetrix Rat Gene 1.0 ST Array. This dataset contains 16 samples of pituitary homozygous mutants (p27Kip1/Cdknb1) from MENX-associated rats, aged 7–8 months, with large tumors 1–2 mm in size, and 5 samples of normal pituitary tissues purchased from BioChain Inc. (Hayward, CA, USA).

Data preprocessing and screening of DEGs

CEL files and probe annotation files were downloaded, and the gene expression data of all the samples were preprocessed via the Robust Multichip Averaging background correction (14), quantile normalization and probe summarization methods using the Oligo package (15). The Linear Models for Microarray Data package (16) of R was used for the identification of genes that were significantly differentially expressed in pituitary adenomas samples. The raw P-value was adjusted by the Benjamin and Hochberg method (17), and only the genes meeting the cut-off criteria of a |log2fold change (FC)| of >1 and an adjusted P-value of <0.05 were selected as DEGs.

GO and pathway enrichment analysis

The Database for Annotation, Visualization and Integrated Discovery (DAVID) gene functional classification tool now provides a comprehensive set of novel and powerful tools for researchers to understand the biological meaning behind abundant genes (18). Pathway enrichment analysis was conducted to identify the significant metabolic pathways for the DEGs (19). P<0.05 and a count number of >2 were used as the cut-off criteria for GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses by DAVID.

PPI network construction

The Search Tool for the Retrieval of Interacting Genes database was used to analyze the PPIs for DEGs by calculating the combined scores (20), and a score >0.4 was chosen as the cut-off. Next, PPI networks for up- and downregulated DEGs were visualized using Cytoscape (http://cytoscape.org/) (21). The highly connected nodes (hub proteins) were detected by calculating the degree of each node protein based on the scale-free property of interaction networks (22).

Results

Identification of DEGs

Based on the cut-off criteria, a total of 629 DEGs were screened from the pituitary adenomas samples, including 391 upregulated and 238 downregulated DEGs.

Enrichment analysis of up- and downregulated DEGs

According to GO functional annotation, the upregulated DEGs were mainly enriched in GO terms associated with the cell cycle and cell division. For example, DEGs such as CCNA2, NUSAP1, CCNB1, CENPF, CDC20 and SPC25 were significantly involved in the cell cycle (P=1.08×10−12); NUSAP1, CENPF, SPC25, CDC20 and CCNB1 were involved in the M phase (P=3.09×10−11); and DEGs such as CCNB1, SPC25, TOP2A, CDC20 and CCNA2 were correlated with cell division (P=1.25×10−7). Notably, PTTG1 was found to be enriched in every GO term (Table I). The downregulated DEGs, such as KCND3, GABRA1, GABRA4 and GABRB1, were markedly associated with ion transport (P=6.05×10−7); DEGs such as KCNJ5, KCND3, KCNJ6 and KCNT2 were relevant to metal ion transport (P=3.87×10−6) and potassium ion transport (P=7.01×10−5); DEGs such as DRD2, POU1F1 and GHRHR were distinctly associated with the positive regulation of multicellular organism growth (P=2.47×10−4), pituitary gland development (P=5.70×10−4), adenohypophysis development (P=8.52×10−4), diencephalon development (P=1.24×10−3) and endocrine system development (P=6.09×10−3); and DEGs such as NOTCH2, ERBB4 and POU1F1 were involved in cell fate commitment (P=3.22×10−2) and the regulation of cell proliferation (P=4.55×10−2) (Table II).

Table I. Top two enriched GO biological process term clusters with the highest enrichment score for the upregulated differentially-expressed genes.

Table I.

Top two enriched GO biological process term clusters with the highest enrichment score for the upregulated differentially-expressed genes.

Enrichment score	Term	Description	Count	P-value	Genes
7.432	GO:0007049	Cell cycle	31	1.08×10−12	SPC25, CCNA2, CENPF, NUSAP1, CDC20, NDC80, CCNB1, PTTG1, CCNB2, BUB1B…
	GO:0022403	Cell cycle phase	23	1.10×10−12	NUSAP1, CENPF, NDC80, CDC20, CCNB1, PTTG1, SPC25, CDKN1A, PLK1, BUB1B…
	GO:0022402	Cell cycle process	27	3.26×10−12	PTTG1, SPC25, CDKN2C, CENPF, NUSAP1, CDC20, NDC80, CCNB1, CDKN1C, BUB1B…
	GO:0000279	M phase	19	3.09×10−11	NUSAP1, CENPF, NDC80, CDC20, PTTG1, CCNB1, SPC25, PLK1, BUB1B, SKA3…
	GO:0051301	Cell division	14	1.25×10−7	NUSAP1, CDC20, PTTG1, CCNB1, SPC25, CCNB2, PLK1, BUB1B, TOP2A, CCNA2…
	GO:0000087	M phase of mitotic cell cycle	12	2.16×10−7	CCNB1, SPC25, PLK1, NUF2, NUSAP1, BUB1B, SKA3, CENPF, CDC20, PTTG1…
	GO:0000278	Mitotic cell cycle	16	5.40×10−7	NUSAP1, CENPF, NDC80, CDC20, PTTG1, CCNB1, SPC25, CDKN1A, CDKN2C, BUB1B…
	GO:0000280	Nuclear division	10	1.09×10−5	CCNB1, SPC25, KIF11, PLK1, NUF2, NUSAP1, BUB1B, SKA3, CDC20, PTTG1
	GO:0007067	Mitosis	10	1.09×10−5	CCNB1, SPC25, KIF11, PLK1, NUF2, NUSAP1, BUB1B, SKA3, CDC20, PTTG1
	GO:0048285	Organelle fission	10	1.74×10−5	CCNB1, SPC25, KIF11, PLK1, NUF2, NUSAP1, BUB1B, SKA3, CDC20, PTTG1
	GO:0051439	Regulation of ubiquitin-protein ligase activity during mitotic cell cycle	4	4.86×10−2	CCNB1, PLK1, BUB1B, CDC20
5.335	GO:0000279	M phase	19	3.09×10−11	NUSAP1, CENPF, NDC80, CDC20, PTTG1, CCNB1, SPC25, PLK1, BUB1B, SKA3…
	GO:0051327	M phase of meiotic cell cycle	7	2.32×10−4	ADCY3, KIF2C, MKI67, PLK1, SGOL2, PTTG1, RAD51
	GO:0007126	Meiosis	7	2.32×10−4	ADCY3, KIF2C, MKI67, PLK1, SGOL2, PTTG1, RAD51
	GO:0051321	Meiotic cell cycle	7	2.76×10−4	ADCY3, KIF2C, MKI67, PLK1, SGOL2, PTTG1, RAD51

[i] GO, gene ontology.

Table II. Top two enriched GO biological process term clusters with the highest enrichment score for the downregulated differentially-expressed genes.

Table II.

Top two enriched GO biological process term clusters with the highest enrichment score for the downregulated differentially-expressed genes.

Enrichment score	Term	Description	Count	P-value	Genes
4.727	GO:0006811	Ion transport	16	6.05×10−7	KCND3, GABRA1, GABRA4, GABRB2, GABRB1, CACNG6, ATP2B4, KCNT2, KCNH7, CACNA1A……
	GO:0030001	Metal ion transport	12	3.87×10−6	KCNJ5, KCND3, KCNJ6, ATP2B4, CACNG6, KCNH7, KCNH8, SCNN1G, KCNJ3, CACNA1A……
	GO:0006812	Cation transport	12	2.54×10−5	KCNJ5, KCND3, KCNJ6, ATP2B4, CACNG6, KCNH7, KCNH8, SCNN1G, KCNJ3, CACNA1A……
	GO:0006813	Potassium ion transport	7	7.01×10−5	KCNJ5, KCND3, KCNJ6, KCNT2, KCNH7, KCNH8, KCNJ3
	GO:0015672	Monovalent inorganic cation transport	8	5.52×10−4	KCNJ5, KCND3, KCNJ6, KCNT2, KCNH7, KCNH8, SCNN1G, KCNJ3
2.321	GO:0040018	Positive regulation of multicellular organism growth	4	2.47×10−4	GH1, DRD2, POU1F1, GHRHR
	GO:0021983	Pituitary gland development	4	5.70×10−4	DRD2, POU1F1, GHRHR, TBX19
	GO:0021984	Adenohypophysis development	3	8.52×10−4	DRD2, POU1F1, GHRHR
	GO:0021536	Diencephalon development	4	1.24×10−3	DRD2, POU1F1, GHRHR, TBX19
	GO:0043567	Regulation of insulin-like growth factor receptor signaling pathway	3	1.82×10−3	GH1, POU1F1, GHRHR
	GO:0040014	Regulation of multicellular organism growth	4	4.82×10−3	GH1, DRD2, POU1F1, GHRHR
	GO:0035270	Endocrine system development	4	6.09×10−3	DRD2, POU1F1, GHRHR, TBX19
	GO:0030900	Forebrain development	5	9.58×10−3	ERBB4, DRD2, POU1F1, GHRHR, TBX19
	GO:0045927	Positive regulation of growth	4	9.75×10−3	GH1, DRD2, POU1F1, GHRHR
	GO:0048732	Gland development	5	1.75×10−2	ERBB4, DRD2, POU1F1, GHRHR, TBX19
	GO:0045165	Cell fate commitment	4	3.22×10−2	NOTCH2, ERBB4, POU1F1, TBX19
	GO:0051240	Positive regulation of multicellular organismal process	5	3.49×10−2	GH1, ERBB4, DRD2, POU1F1, GHRHR
	GO:0042127	Regulation of cell proliferation	8	4.55×10−2	NOTCH2, ERBB4, DRD2, NR3C2, CDK6, POU1F1, GHRHR, TBX19

[i] GO, gene ontology.

According to the KEGG pathway enrichment analysis, the upregulated DEGs were mainly enriched in 10 pathways

For example, CDC20, CCNB1, CCNB2, BUB1, CDKN1A and MCM3 were enriched in the pathway of the cell cycle (P=5.01×10−7); CDC20, CCNB1, CCNB2, BUB1 and PLK1 were distinctly enriched in the pathway of oocyte meiosis (P=1.058×10−3); CCNB1, CCNB2, CCNA2, BUB1 and PLK1 were significantly enriched in the pathway of progesterone-mediated oocyte maturation (P=1.150×10−3); and CDKN1A, CCNB1, CCNB2 and CASP8 were markedly enriched in the p53 signaling pathway (P=3.4841×10−2) (Table III). Meanwhile, the downregulated DEGs were mainly enriched in 7 pathways. GH1, GABRA1, GABRA4 and GABRB1 were enriched in the pathway of neuroactive ligand-receptor interaction (P=4.35×10−4); MAOB, ALDH2 and ALDH1A7 were mainly enriched in the pathways of histidine metabolism (P=1.2476×10−2) and tryptophan metabolism (P=3.7487×10−2); ATP2B4, ERBB4 and PLCG2 were enriched in the calcium signaling pathway (P=3.9919×10−2); and GSTA4, MGST1, GSTM7 were significantly enriched in the pathways of drug metabolism (P=1.4397×10−2) and glutathione metabolism (P=4.9294×10−2) (Table III).

Table III. Results of pathway enrichment analysis of the up- and downregulated differentially-expressed genes.

Table III.

Results of pathway enrichment analysis of the up- and downregulated differentially-expressed genes.

Category	Term	Description	Count	P-value	Genes
Upregulated	rno04110	Cell cycle	14	5.01×10−7	CDC20, MCM3, CDKN1C, CCNB1, CDKN1A, CCNB2, CDKN2C, BUB1, BUB1B, CCNA2……
	rno04114	Oocyte meiosis	9	1.06×10−3	CCNB1, ADCY3, AR, CCNB2, MAPK12, PLK1, BUB1, CDC20, PTTG1
	rno00601	Glycosphingolipid biosynthesis	5	1.10×10−3	B4GALT1, B3GALT2, B3GALT5, FUT4, B4GALT4
	rno04914	Progesterone-mediated oocyte maturation	8	1.15×10−3	CCNB1, ADCY3, CCNB2, KRAS, MAPK12, PLK1, BUB1, CCNA2
	rno04062	Chemokine signaling pathway	9	1.46×10−2	ADCY3, KRAS, LYN, ARRB1, PREX1, GRK5, PRKCD, CCL6, SHC4
	rno00330	Arginine and proline metabolism	5	1.70×10−2	ARG1, GOT1, NOS1, ASS1, GAMT
	rno05219	Bladder cancer	4	2.94×10−2	CDKN1A, KRAS, PGF, VEGFA
	rno04115	p53 signaling pathway	5	3.48×10−2	CCNB1, CDKN1A, CCNB2, CASP8, IGFBP3
	rno04610	Complement and coagulation cascades	5	4.19×10−2	C1QA, A2M, C1S, C1QC, F2R
	rno00510	N-glycan biosynthesis	4	4.89×10−2	B4GALT1, MAN2A1, ALG5, MAN1A1
Downregulated	rno04080	Neuroactive ligand-receptor interaction	9	4.35×10−4	GH1, GABRA1, GABRA4, GABRB2, DRD2, GABRB1, TSHB, LHB, GHRHR
	rno00410	β-alanine metabolism	3	1.05×10−2	ALDH2, ALDH1A7, DPYD
	rno00340	Histidine metabolism	3	1.25×10−2	MAOB, ALDH2, ALDH1A7
	rno00982	Drug metabolism	4	1.44×10−2	GSTA4, MAOB, MGST1, GSTM7
	rno00380	Tryptophan metabolism	3	3.75×10−2	MAOB, ALDH2, ALDH1A7
	rno04020	Calcium signaling pathway	5	3.99×10−2	ATP2B4, ERBB4, PLCG2, PLCB1, CACNA1A
	rno00480	Glutathione metabolism	3	4.93×10−2	GSTA4, MGST1, GSTM7

Analysis of PPI network

The PPI networks constructed with the up- and downregulated DEGs consisted of 1,044 and 69 PPI pairs, respectively. In the former, PTTG1, along with CCNB1, CCNA2, SPC25, CENPF, NUSAP1, CDC20, TOP2A and BUB1, were observed to interact with each other (Fig. 1). Within the PPI network built with downregulated DEGs, GABRA1, GABRA4, GABRB1 and GABRB1 were observed to interact with each other; GSTA3, GSTA4, GSTM7 and MGST1 were also found to interact with each other, and POU1F1 was observed to interact with POMC (Fig. 2). The connection degrees of the top 15% highly-connected upregulated DEGs were each >30, and those of CDK1, CCNB1, CCNA2 and BUB1 were 51, 47, 46 and 44, respectively (Table IV). The top 20% highly-connected downregulated DEGs all had connection degrees of at least 3, and the degrees of POMC, GSTA4, POU1F1, ERBB4, KCND3 and NOTCH2 were 6, 5, 4, 4, 3 and 3, respectively (Table IV).

Figure 1. Protein-protein interaction network constructed with the upregulated differentially-expressed genes. Different shades of nodes colors represent the degree of up- or downregulation.

Figure 2. Protein-protein interaction network constructed with the downregulated differentially-expressed genes. Different shades of nodes colors represent the degree of up- or downregulation.

Table IV. Upregulated DEGs with connection degrees of >30 and the downregulated DEGs with connection degrees of at least 3 in the protein-protein interaction networks.

Table IV.

Upregulated DEGs with connection degrees of >30 and the downregulated DEGs with connection degrees of at least 3 in the protein-protein interaction networks.

Category	Degree
Upregulated DEGs
CDK1	51
CCNB1	47
CCNA2	46
BUB1	44
ECT2	43
TPX2	42
NDC80	42
PRC1	42
NUSAP1	41
TOP2A	41
CCNB2	41
PBK	41
RACGAP1	41
TTK	41
PIK1	40
SPC25	40
BUB1B	40
CENPF	40
CDKN3	39
NUF2	38
CDC20	38
KIF11	38
DLGAP5	38
SGOL2	37
DTL	36
KIF20A	36
CDCA2	36
KIF20B	35
ESCO2	35
RAD51	34
ARHGAP11A	34
CKAP2	33
HMMR	33
KIF2C	32
DEPDC1	32
Downregulated DEGs
POMC	6
GSTA4	5
GSTA1	5
POU1F1	4
ERBB4	4
GABRA1	3
MGST1	3
GABRA4	3
GSTM7	3
ALDH2	3
GH1	3
NR4A2	3
MAOB	3
KCND3	3
NOTCH2	3
GABRB1	3

[i] DEGs, differentially-expressed genes.

Discussion

In the present study, 391 DEGs were identified to be significantly upregulated and 238 were significantly downregulated in the pituitary adenomas samples. According to the constructed PPI network with the upregulated DEGs, PTTG1 interacted with other DEGs with higher connection degrees, such as CCNB1, CCNA2, SPC25, CENPF, NUSAP1, CDC20, TOP2A and BUB1.

PTTG1, a tumorigenic gene in vivo (23), is abundantly expressed in pituitary tumors (24). As a securin protein, PTTG1 is correlated with the mitotic checkpoint that prevents abnormal chromosome segregation (25), and peaks at the G2/M phase (26). The overexpression of PTTG1 results in cell transformation and induces aneuploidy (27), and this exists in >90% of pituitary tumors (28). PTTG1, together with CCNB1, CCNA2, BUB1, SPC25, CENPF, NUSAP1, TOP2A and CDC20, were all found to be enriched in GO terms associated with the cell cycle or cell division, which are indispensable for tumor growth. It has been reported that CDC20 is involved in the degradation of PTTG1-encoding products (29). Meanwhile, previous studies have also reported the abnormal expression of CCNB1 (30), CCNA2 (31), and BUB1 (32) in pituitary adenomas. Furthermore, CCNB1 was enriched in the p53 signaling pathway. PTTG1-encoding protein can cooperate with p53 to take part in cell apoptosis and DNA damage/repair (33,34). Altered p53 expression has been reported in pituitary carcinomas (35). Also, PTTG1 can activate β-fibroblast growth factor, cyclin D3 and c-myc to promote cell proliferation (36,37). Therefore, PTTG1 may play a crucial role in the occurrence of pituitary adenomas via interaction with CCNB1, CCNA2, CENPF, NUSAP1, CDC20, TOP2A, BUB1 and p53.

Within the PPI network constructed with downregulated DEGs, GABRA1, GABRA4 and GABRB1 had higher degrees of connection to other genes. These genes were enriched in the pathway of neuroactive ligand-receptor interaction. GABRA1, GABRA4 and GABRB1 encode γ-aminobutyric acid (GABA) receptors. GABA is the major inhibitory neurotransmitter in the mammalian brain and may act as a paracrine or autocrine regulating factor in the human pituitary gland and human pituitary growth hormone adenomas (38). It has been reported that GABA has a specific effect on the electrical activity of a tumoral line of rat pituitary cells, and that it inhibits prolactin secretion directly at the pituitary level (39). Additionally, POU1F1 was also observed to have a higher connection degree in the PPI network. This gene encoding a member of the POU family of transcription factors (40), was correlated with the development of the pituitary gland, adenohypophysis and endocrine system. In humans, POU1F1 mutation has been shown to be associated with combined pituitary hormone deficiency (41). POU1F1 is also implicated in cell growth and prevents cell apoptosis (42). In the present study, it was observed to interact with POMC, which encodes a polypeptide hormone precursor. The encoded polypeptide hormone precursor is synthesized mainly in corticotrophin cells of the anterior pituitary (43). A previous study has shown that in silent pituitary adenomas, POMC mRNA has a diffuse low level or is absent (44). Thus, GABRA1, GABRA4, GABRB1, POU1F1 and POMC may also have critical roles in pituitary adenoma occurrence via close interaction.

In conclusion, upregulated DEGs, such as those associated with the cell cycle or cell division (e.g., CCNB1, CCNA2, BUB1, CENPF, NUSAP1, CDC20, TOP2A and particularly PTTG1) and downregulated DEGs, such as those relevant to neuroactive ligand-receptor interaction (e.g., GABRA1, GABRA4 and GABRB1), as well as those correlated with the development of the pituitary gland, adenohypophysis and endocrine system (e.g., POU1F1) may have essential roles in the pathogenesis of pituitary adenomas. The present study provides novel information for the clinical diagnosis of this disease.

Acknowledgements

This study was supported by grants from the National Natural Science Foundation of China (no. 81273732/H2708) and the Shanghai Educational Commission Funding Project (no. 2012JW68).

References