Abstract

Previous studies have revealed that Epstein–Barr virus (EBV) was closely associated with nasopharyngeal carcinoma (NPC). This study aimed to characterize the global pathways affected in the EBV-associated NPC. Combined with microdissection, gene expression profiles in 22 NPCs and 10 non-tumor nasopharyngeal epithelial (NPE) tissue samples were analyzed. All NPC specimens served in the microarray analysis were positive for EBV, as judged by identification of the expression of EBV nuclear antigen 1 (EBNA1). Through gene set enrichment analysis (GSEA), we found that cell cycle pathway was the most disregulated pathway in NPC (P = 0.000, false discovery rate q-value = 0.007), which included some aberrant expressed components. We first found that overexpression of CDK4, cyclin D1, and Rb proteins, and loss of expression of proteins p16, p27, and p19 were statistically significant in NPC tissues compared with non-cancerous NPE (P < 0.05) by real-time RT–PCR and tissue microarray. EBV-encoded small RNA-1 (EBER-1) hybridization signals in the NPC showed significant associations with the overexpression of Rb (P = 0.000), cyclin D1 (P = 0.000), CDK4 (P = 0.000), and the loss of expression of p16 proteins (P = 0.039). In the final logistic regression analysis model, EBER-1 and abnormal expression of p16, Rb, cyclin D1, and E2F6 were independent contributions to nasopharyngeal carcinogenesis. Through survival analysis, only cyclin D1 could predict the prognosis of NPC patients. These results suggested that cell cycle pathway was the most disregulated pathway in the EBV-associated NPC, and EBER-1 was closely associated with p16, CDK4, cyclin D1, and Rb. cyclin D1 could be the prognosis biomarker for NPC.

Introduction

Nasopharyngeal carcinoma (NPC) is a squamous cell carcinoma that usually develops around the ostium of the Eustachian tube in the lateral wall of the nasopharynx. Although NPC is classified as a subtype of head and neck squamous cell carcinoma, its unique epidemiology, clinical characteristics, etiology, and histopathology warrant separate efforts for the study of its underlying molecular mechanism of carcinogenesis [1]. Epstein–Barr virus (EBV) infection, genetic susceptibility, and chemical carcinogens play important roles in the NPC pathogenesis [2–6]. A variety of proliferative signals are aberrantly activated in the NPC, such as the Akt pathway, mitogen-activated protein kinases, Wnt pathway, and epidermal growth factor receptor signaling pathway [3,7–9]. Although some evidence regarding the differential signaling pathway in NPC has been reported, the global pathways affected in the EBV associated NPC is not fully understood.

With the advance of the human genome project, a large amount of information is available for surveying the physiological and pathological processes of humans at global levels. Microarray is a powerful technology developed recently to identify and isolate differentially expressed genes [10–12]. In previous studies, the classical approach to NPC microarray analysis is to treat genes as independent agents, to apply some statistical test to per gene and follow that up with some form of P-value correction method, but not to analyze the pathway as a whole. Gene set enrichment analysis (GSEA) [13–15] is a microarray data analysis method that uses predefined gene sets and ranks of genes to identify significant biological changes in microarray data sets. Meanwhile, the method derives its power by focusing on gene sets, that is, groups of genes that share common biological function, chromosomal location, or regulation, and cover a wide variety of biological categories and pathways.

In the analysis of clinical tumor specimens in our former and other studies, a significant confounder is the cellular heterogeneity of normal and diseased tissues. To overcome this problem, microdissection was developed to analyze the clinical samples [16,17]. Therefore, in this study, we used microarray analysis combined with laser microdissection and GSEA to identify differential signaling pathways in the EBV-associated NPC.

Materials and Methods

Samples

For gene expression analysis, 10 non-cancerous nasopharyngeal epithelium (NPE) and 22 NPC biopsies were collected from Xiangya Hospital in 2003. Each biopsy sample was divided into two sections: one section was routinely stained with hematoxylin and eosin and the other was immersed into the RNALater reagent (Qiagen, Carlsbad, USA) at –80°C ready for microdissection. For preparation of the NPC tissue microarray (TMA), 448 patients with NPC and non-cancerous NPE with information such as name, sex, age, race, pathological diagnosis, tumor, nodal, and metastasis, as well as viral capsid antigen (VCA)-IgA, were collected between January, 2002 and Octobor, 2004 at Xiangya Hospital (Table 1). For real-time RT–PCR, five histologically normal NPEs and seven NPC biopsies were collected from Xiangya Hospital. All the individuals participating in this project signed the informed consent.

Table 1

Specimen for TMA construction

Specimen Number 
Histologically normal NPE 105 
Dysplastic NPE 69 
NPC 231 
Clinical stage 
 I 22 
 II 56 
 III 72 
 IV 81 
Histological type 
 WHO I 
 WHO II 215 
 WHO III 
Normal NPE of NPC after radiation 23 
Relapse of NPC after radiation 20 
Specimen Number 
Histologically normal NPE 105 
Dysplastic NPE 69 
NPC 231 
Clinical stage 
 I 22 
 II 56 
 III 72 
 IV 81 
Histological type 
 WHO I 
 WHO II 215 
 WHO III 
Normal NPE of NPC after radiation 23 
Relapse of NPC after radiation 20 

Microdissection, RNA extraction, RNA amplification, and hybridization

Microdissection of 22 NPCs and 10 non-cancerous NPE tissues was performed as described previously [18]. Total RNA was extracted from the microdissection samples with RNeasy® Mini Kit (Qiagen), according to manufacturer's instructions.

A 200 ng aliquot of total RNA from each sample was amplified to cDNA using Ambion's Illumina RNA amplification kit following the instructions (Ambion, Austin, USA). In vitro transcription reaction of cDNA to cRNA was performed overnight including biotin-11-dUTP for labeling the cRNA product. Each sample (1500 ng) labeled with cRNA was hybridized to the gene chip containing 8378 full-length or segmental novel and known human genes that were provided by Biostar Gene Technology Co. Ltd., Shanghai, China.

Pre-processing of gene chip data

The results of hybridization were scanned using the computer system (ScanArray 4000, General Scanning Inc., Bedford, USA). Primary data collection and analysis were carried out using Genepix Pro 3.0 (Axon Instruments, Foster City, USA). The area of the array with obvious blemish was manually flagged and excluded from subsequent analysis. To minimize artifacts arising from low expression values, only genes with raw intensity values for both Cy3 and Cy5 of >200 counts were chosen for differential analysis. Then overall intensities were normalized with a correction coefficient obtained using the ratios of these genes. Genes with <70% of measurements across all the samples were excluded from the following GSEA. Finally, of 8378 genes, 7007 passed this variation filter.

Pathway analysis by GSEA

Pathway analysis of the expression data was performed using GSEA implemented in javaGSEA application, version 1.0 [14]. Gene sets for GSEA were taken from Database C2 of MSigDB, version 2.0 (http://www.broad.mit.edu/cancer/software/gsea/wiki/index.php/Msigdb_v2_release_notes) of January 2007. This catalog includes 1687 gene sets in which several gene products are involved in specific metabolic and signaling pathways. Briefly, the GSEA algorithm ranks all array genes according to their expression under each experimental condition. Moreover, GSEA integrates the known knowledge of a gene's functional role with the expression data to detect concerted expression changes in a set of genes responsible for producing a phenotype.

RT–PCR analysis

Reverse transcription was carried out as described previously [19]. After the reverse transcriptase reaction, PCR was performed at 94°C for 5 min, then 38 cycles at 94°C for 45 s, 55°C for 45 s, and 72°C for 1 min followed by 72°C for 7 min. RT–PCR were repeated at least three times to avoid false-positive results. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an endogenous control for normalization. The primer sequences of the EBV nuclear antigen 1 (EBNA1) and GAPDH used for RT–PCR were as follows: EBNA1 forward, 5′-agtcgtctcccctttggaat-3′; reverse, 5′-tcctcaccctcatctccatc-3′; GAPDH forward, 5′-accacagtccat gccatcac-3′; reverse, 5′-tccaccaccctgttgctgta-3′.

Real-time quantitative RT–PCR analysis

Total RNA was extracted from five non-cancerous NPEs and seven NPC tissues using Trizol reagent (Gibco BRL, Grand Island, NY, USA) and 2 µg of total RNA was subjected to cDNA synthesis using Superscript First-Strand Synthesis Kit (Invitrogen, Carlsbad, CA, USA) according to instructions of the manufacturer. Primers for real-time RT–PCR of these genes are shown in Table 2, which are designed to span at least two exons to rule out gDNA contamination. The cDNA generated was used for real-time PCR amplification with SYBR Green I PCR Kit (TaKaRa, Shiga, Japan) as recommended by the manufacturer. The reaction was carried out for 40–45 cycles in a real-time PCR instrument (BIO-RAD, IQ5). Real-time PCR conditions were 95°C for 3 min, followed by 40 cycles of 94°C for 20 s, 51°C for 20 s, 72°C for 20 s, and a final extension at 72°C for 5 min. A series of diluted cDNA samples was used as template to generate the standard curves and melting curve analysis to verify the presence of a single amplicon. For each sample, the relative gene expression was calculated with GAPDH gene as an internal reference. Experiments were carried out in duplicate and independently performed on three separate occasions.

Table 2

Summary of primer sequences, annealing temperature and PCR product sizes for nine target genes

Gene Primer sequence (5′−3′) Annealing temperature (°C) Product sizes (bp) 
p27 Forward: ccggctaactctgaggacac 60 120 
 Reverse: agaagaatcgtcggttgcag 
p16 Forward: gacatccccgattgaaagaa 60 198 
 Reverse: tttacggtagtgggggaagg 
CDK4 Forward: ctgaccgggagatcaaggta 60 272 
 Reverse: gggtgtaagtgccatctggt 
cyclin D1 Forward: tcgggagaggaggactttg 61 196 
 Reverse: cagtgcggaggatgatgtg 
CDK8 Forward: ggctatgcaggacccctatt 60 176 
 Reverse: ggccagttccattagtgtgg 
p19 Forward: gctgcaggtcatgatgtttg 60 209 
 Reverse: ctgccagatggattggaagt 
DP2 Forward: aggcggatagaacggataaag 58 212 
 Forward: aacttgtcactggagatgctg 
Rb Forward: gtctgccaacaccaacaaaa 60 152 
 Reverse: agggttgcttccttcagcac 
GAPDH Forward: gtcagtggtggacctgacct 58 550 
 Reverse: aggggagattcagtgtggtg 
Gene Primer sequence (5′−3′) Annealing temperature (°C) Product sizes (bp) 
p27 Forward: ccggctaactctgaggacac 60 120 
 Reverse: agaagaatcgtcggttgcag 
p16 Forward: gacatccccgattgaaagaa 60 198 
 Reverse: tttacggtagtgggggaagg 
CDK4 Forward: ctgaccgggagatcaaggta 60 272 
 Reverse: gggtgtaagtgccatctggt 
cyclin D1 Forward: tcgggagaggaggactttg 61 196 
 Reverse: cagtgcggaggatgatgtg 
CDK8 Forward: ggctatgcaggacccctatt 60 176 
 Reverse: ggccagttccattagtgtgg 
p19 Forward: gctgcaggtcatgatgtttg 60 209 
 Reverse: ctgccagatggattggaagt 
DP2 Forward: aggcggatagaacggataaag 58 212 
 Forward: aacttgtcactggagatgctg 
Rb Forward: gtctgccaacaccaacaaaa 60 152 
 Reverse: agggttgcttccttcagcac 
GAPDH Forward: gtcagtggtggacctgacct 58 550 
 Reverse: aggggagattcagtgtggtg 

TMA construction

TMA was constructed as described in our previous study [2]. A total of 448 tissue cores representing 448 patients were placed in one TMA block. The tissue core numbers on each section were slightly different because of additional losses suffered from block trimming and staining procedures.

In situ hybridization and immunohistochemistry (IHC)

The DIG olignucleotide 3′-tailing labeling kit was purchased from Roche (Basel, Switzerland) and the in situ hybridization (ISH) method was the same as in our previous study [2]. Immunohistochemistry (IHC) studies were performed using the standard streptavidin/peroxidase staining method as described previously [2]. As many literatures have reported that NPC is closely associated with EBV, we took advantage of the high-throughput TMA to detect EBV-encoded small RNA-1 (EBER-1) using ISH. The EBER-1 probe was 5′-AGACACCGTCCTCACCACCCGGGACTTGTA-3′. These antibodies included monoclonal mouse antihuman p16, 1:100 (clone F-12, Santa Cruz Biotechnology, Santa Cruz, USA); p19INK4d, 1:100 (clone DSC-100, NeoMarkers, Fremont, USA); p27, 1:100 (clone DCS-72, NeoMarkers); Rb, 1:100 (clone CH18-0159, ZYMED, San Diego, USA); CDK4, 1:100 (clone DK4-CT, NeoMarkers); cyclin D1, 1:100 (Santa Cruz ), and polyclonal antibodies such as CDK8, cyclinC, DP2, E2F6, and phosphorylated Rb from Neomarkers. All known positive sections were taken as positive controls. Negative mouse serum and PBS were used instead of first antibody as negative control and blank control, respectively. A semiquantitative scoring criterion for ISH and IHC was used, in which both staining intensity and positive areas were recorded.

Survival analysis

There were follow-up data for 115 patients. Overall survival (OS) was defined as the time of diagnosis to date of death. Event-free survival (EFS) was defined as the time of diagnosis to date of first failure. The OS and EFS estimates over time were calculated using the Kaplan–Meier method and differences were compared using the log-rank test. Results of the analysis were considered significant in a log-rank test if P < 0.05.

Statistical analysis

Associations between clinicopathological parameters and the cell cycle regulators such as p16, CDK4, cyclin D1, Rb protein (pRb), and so on, were evaluated using the Pearson χ2-test or Fisher's exact test. Spearman correlation test was used to evaluate the pair-wise association of EBER-1 and cell cycle regulators in NPC. We used the standard logistic regression model to analyze EBER-1 and cell cycle regulators in NPC carcinogenesis. Real-time PCR (RT-PCR) results were evaluated using Student t-test. Calculations were performed using the SPSS 13.0 statistical software for Windows. P < 0.05 is considered statistically significant, and all statistical tests were two sided.

Results

Detection of EBNA1 expression of the samples served in the microarray analysis by RT–PCR and EBER-1, LMP1 expression using NPC-specific TMA

In this study, all NPC specimens served in the cDNA microarray analysis were positive for EBV, as judged by identification of the expression of EBNA1 (Fig. 1). Furthermore, we detected the association between EBV infection and VCA-IgA titer. In the 231 NPC patients, the positive rate of VCA-IgA was 86.1% (199/231), suggesting 86.1% NPC patients had the EBV infection. Nuclear EBER-1 expression was detected in 68.6% (127/185) NPC cases by ISH [Fig. 2(A,B) and 3]. The expression of LMP1 detected by IHC in NPC patients was over 78%. These results suggested a strong association between EBV infection and NPC.

Fig. 1

Detection of EBNA1 expression by RT–PCR in the same total RNA of NPC samples served in the cDNA microarray analysis  GAPDH served as an internal control for each reaction. All 22 NPC biopsies were positive for EBNA1 expression. M, DL2, 000 DNA marker. T1–T22 represent the 22 NPC tissues.

Fig. 1

Detection of EBNA1 expression by RT–PCR in the same total RNA of NPC samples served in the cDNA microarray analysis  GAPDH served as an internal control for each reaction. All 22 NPC biopsies were positive for EBNA1 expression. M, DL2, 000 DNA marker. T1–T22 represent the 22 NPC tissues.

Fig. 2

Representative images of EBER-1 and cell cycle regulators detected by ISH and ICH  Brown denote positive signal. Expression of EBER-1 mRNA in the columnar epithelia cells (A) and differentiated non-keratinizing NPC (B). Expression of p16 in chronic inflammation of nasopharyngeal mucosa (C) and differentiated non-keratinizing NPC (D). Expression of p27 in squamous epithelium (E) and differentiated non-keratinizing NPC (F). Expression of p19 (G) and cyclin D1 (H) in differentiated non-keratinizing NPC. Expression of Rb in chronic inflammation of nasopharyngeal mucosa (I) and differentiated non-keratinizing NPC (J). Expression of phosphorylated Rb in chronic inflammation of nasopharyngeal mucosa (K) and differentiated non-keratinizing NPC (L). Expression of CDK4 (M) and DP2 (N) in differentiated non-keratinizing NPC. Expression of E2F6 in the columnar epithelia cells (O) and differentiated non-keratinizing NPC (P). Bar = 200 µm.

Fig. 2

Representative images of EBER-1 and cell cycle regulators detected by ISH and ICH  Brown denote positive signal. Expression of EBER-1 mRNA in the columnar epithelia cells (A) and differentiated non-keratinizing NPC (B). Expression of p16 in chronic inflammation of nasopharyngeal mucosa (C) and differentiated non-keratinizing NPC (D). Expression of p27 in squamous epithelium (E) and differentiated non-keratinizing NPC (F). Expression of p19 (G) and cyclin D1 (H) in differentiated non-keratinizing NPC. Expression of Rb in chronic inflammation of nasopharyngeal mucosa (I) and differentiated non-keratinizing NPC (J). Expression of phosphorylated Rb in chronic inflammation of nasopharyngeal mucosa (K) and differentiated non-keratinizing NPC (L). Expression of CDK4 (M) and DP2 (N) in differentiated non-keratinizing NPC. Expression of E2F6 in the columnar epithelia cells (O) and differentiated non-keratinizing NPC (P). Bar = 200 µm.

Fig. 3

Abnormal expression of EBER-1 hybridization signals and cell cycle regulators in histologically normal NPE, dysplastic NPE, and NPC  Comparison of abnormal expression of EBER-1 hybridization signals, p16, p27, p19, CDK4, cyclin D1, Rb, DP2, and E2F6 proteins between NPC and various types of non-cancerous NPE. Here P-value denotes significant differences between the groups statistically evaluated by χ2 test.

Fig. 3

Abnormal expression of EBER-1 hybridization signals and cell cycle regulators in histologically normal NPE, dysplastic NPE, and NPC  Comparison of abnormal expression of EBER-1 hybridization signals, p16, p27, p19, CDK4, cyclin D1, Rb, DP2, and E2F6 proteins between NPC and various types of non-cancerous NPE. Here P-value denotes significant differences between the groups statistically evaluated by χ2 test.

Cell cycle pathway: the most disregulated pathway in NPC tissues revealed by GSEA

In order to understand the globally affected pathways associated with NPC, we performed the cDNA microarray analysis in 10 non-cancerous NPEs and 22 NPC tissues by laser microdissection and GSEA [14]. Our results suggested that the cell cycle pathway was the most disregulated pathway in NPC according to the criteria (normalized enrichment score, NES = −2.22; NOM P, the uncorrected P = 0.000; false discovery rate (FDR) q-value =0.001) recommended by Subramanian et al. [14] (Table 3). The expression of the cell cycle pathway related genes correlated with phenotype non-cancerous NPE and NPC could be seen in Fig. 4. As the G1-S phase members of the cell cycle pathway, p16 and p27 were down-regulated in NPC tissues, whereas TP53, CDK4 and RB1, were up-regulated in NPC tissues in our gene chip data (Table 4).

Fig. 4

An expression data set sorted of cell cycle pathway by correlation with phenotype  N1–N10 represent the 10 non-cancerous NPE tissues, T1–T22 represent the 22 NPC tissues. Signal intensities are illustrated by varying shades of red (up-regulation) and blue (down-regulation).

Fig. 4

An expression data set sorted of cell cycle pathway by correlation with phenotype  N1–N10 represent the 10 non-cancerous NPE tissues, T1–T22 represent the 22 NPC tissues. Signal intensities are illustrated by varying shades of red (up-regulation) and blue (down-regulation).

Table 3

Affected pathways in NPC group as revealed by GSEA with FDR q-value cutoff 0.01

Name Size ES NES NOM P-value FDR q-value 
Cell cycle KEGG 52 −0.69 −2.22 0.000 0.001 
Cell cycle 51 −0.67 −2.16 0.000 0.002 
Tarte plasma blastic 224 −0.59 −2.13 0.000 0.003 
G1 to S cell cycle reactome 46 −0.60 −2.10 0.002 0.005 
Breastca two classes 86 −0.53 −2.05 0.000 0.008 
Cancer undifferentiated meta up 50 −0.67 −2.03 0.000 0.008 
Sana ifng endothelial dn 55 −0.63 −2.03 0.000 0.007 
Schumacher myc up 37 −0.65 −2.01 0.000 0.009 
Raccycd pathway 15 −0.74 −2.01 0.004 0.008 
Breastca three classes 30 −0.62 −2.01 0.000 0.008 
Name Size ES NES NOM P-value FDR q-value 
Cell cycle KEGG 52 −0.69 −2.22 0.000 0.001 
Cell cycle 51 −0.67 −2.16 0.000 0.002 
Tarte plasma blastic 224 −0.59 −2.13 0.000 0.003 
G1 to S cell cycle reactome 46 −0.60 −2.10 0.002 0.005 
Breastca two classes 86 −0.53 −2.05 0.000 0.008 
Cancer undifferentiated meta up 50 −0.67 −2.03 0.000 0.008 
Sana ifng endothelial dn 55 −0.63 −2.03 0.000 0.007 
Schumacher myc up 37 −0.65 −2.01 0.000 0.009 
Raccycd pathway 15 −0.74 −2.01 0.004 0.008 
Breastca three classes 30 −0.62 −2.01 0.000 0.008 

Size, number of genes in the gene set; ES, enrichment score; NES, normalized enrichment score; FDR q-value, false discovery rate and multiple testing corrections (q-value); NOM P-value, the uncorrected p-value.

Table 4

Partial list of differentially expressed genes in cell cycle pathway in NPC when compared with non-cancerous NPE

GenBank number Gene symbol Definition P-value Fold difference geometric means (T/N) 
NM_001211 BUB1B Budding uninhibited by benzimidazoles 1 (yeast homolog), β (BUB1B) 4.5 × 10−5 2.019 
NM_001237 CCNA2 Cyclin A2 (CCNA2) 6.3 × 10−7 2.721 
NM_031966 CCNB1 Cyclin B1 (CCNB1) 2.4 × 10−8 3.454 
NM_004701 CCNB2 Cyclin B2 (CCNB2) 5.8 × 10−8 3.591 
NM_001759 CCND2 Cyclin D2 (CCND2) 1.8 × 10−6 2.751 
NM_000077 CDKN2A Cyclin-dependent kinase inhibitor 2A (CDKN2A, p16) 1.5 × 10−6 0.285 
NM_004064 CDKN1B Cyclin-dependent kinase inhibitor 1B (CDKN1B, p27) 2.3 × 10−4 0.378 
NM_001786 CDC2 Cell division cycle 2,G1 to S and G2 to M (CDC2), transcript variant 1 3.4 × 10−6 1.433 
NM_000075 CDK4 Cyclin-dependent kinase 4 (CDK4), transcript variant 1 1.3 × 10−4 1.807 
NM_007111 DP-1 Transcription factor Dp-1 (TFDP1) 1.3 × 10−4 1.428 
NM_001951 E2F5 E2F transcription factor 5, p130-binding (E2F5) 2.1 × 10−5 1.837 
NM_002592 PCNA Proliferating cell nuclear antigen (PCNA) 2.3 × 10−7 2.377 
NM_000321 RB1 Retinoblastoma 1 (including osteosarcoma) (RB1) 1.4 × 10−8 1.985 
NM_000546 TP53 Tumor protein p53 (Li-Fraumeni syndrome) (TP53) 4.8 × 10−4 1.474 
GenBank number Gene symbol Definition P-value Fold difference geometric means (T/N) 
NM_001211 BUB1B Budding uninhibited by benzimidazoles 1 (yeast homolog), β (BUB1B) 4.5 × 10−5 2.019 
NM_001237 CCNA2 Cyclin A2 (CCNA2) 6.3 × 10−7 2.721 
NM_031966 CCNB1 Cyclin B1 (CCNB1) 2.4 × 10−8 3.454 
NM_004701 CCNB2 Cyclin B2 (CCNB2) 5.8 × 10−8 3.591 
NM_001759 CCND2 Cyclin D2 (CCND2) 1.8 × 10−6 2.751 
NM_000077 CDKN2A Cyclin-dependent kinase inhibitor 2A (CDKN2A, p16) 1.5 × 10−6 0.285 
NM_004064 CDKN1B Cyclin-dependent kinase inhibitor 1B (CDKN1B, p27) 2.3 × 10−4 0.378 
NM_001786 CDC2 Cell division cycle 2,G1 to S and G2 to M (CDC2), transcript variant 1 3.4 × 10−6 1.433 
NM_000075 CDK4 Cyclin-dependent kinase 4 (CDK4), transcript variant 1 1.3 × 10−4 1.807 
NM_007111 DP-1 Transcription factor Dp-1 (TFDP1) 1.3 × 10−4 1.428 
NM_001951 E2F5 E2F transcription factor 5, p130-binding (E2F5) 2.1 × 10−5 1.837 
NM_002592 PCNA Proliferating cell nuclear antigen (PCNA) 2.3 × 10−7 2.377 
NM_000321 RB1 Retinoblastoma 1 (including osteosarcoma) (RB1) 1.4 × 10−8 1.985 
NM_000546 TP53 Tumor protein p53 (Li-Fraumeni syndrome) (TP53) 4.8 × 10−4 1.474 

Quantitative expression level of differential expression genes validated by real-time RT–PCR

To determine the reliability of our gene chip data, we confirmed our microarray results by quantitative real-time RT–PCR analysis. A higher Ct value indicated less RNA copy in original samples, as more cycles were required to reach the same RNA density point on the graph. Compared with non-cancerous NPE tissues, CDK4, Rb, and cyclin D1 genes were found significantly up-regulated (P = 0.008, 0.006, and 0.035, respectively); p16, p27, and p19 genes were significantly down-regulated in NPC tissues (P = 0.006, 0.000, and 0.007, respectively). However, the expression of CDK8 and DP2 mRNA was not statistically significant (Table 5).

Table 5

Statistical analysis of real-time RT–PCR

Gene Tissue Number Mean ± SD (Ct)2 P-value 
p16 Non-cancerous NPE 9.3000 ± 0.56635 0.006 
 NPC 11.8857 ± 1.55201 
Rb Non-cancerous NPE 12.6600 ± 1.04007 0.006 
 NPC 8.0343 ± 2.82198 
CDK4 Non-cancerous NPE 13.5360 ± 0.44200 0.008 
 NPC 10.7414 ± 0.38959 
p27 Non-cancerous NPE 5.3900 ± 0.82037 0.000 
 NPC 11.4743 ± 1.93016 
p19 Non-cancerous NPE 7.1160 ± 0.78325 0.007 
 NPC 12.0800 ± 3.16098 
cyclin D1 Non-cancerous NPE 11.9640 ± 0.31382 0.035 
 NPC 8.7657 ± 2.88830 
CDK8 Non-cancerous NPE 13.2240 ± 0.49303 0.095 
 NPC 11.1429 ± 2.45628 
DP2 Non-cancerous NPE 11.3460 ± 0.74527 0.697 
 NPC 10.7914 ± 2.98937 
Gene Tissue Number Mean ± SD (Ct)2 P-value 
p16 Non-cancerous NPE 9.3000 ± 0.56635 0.006 
 NPC 11.8857 ± 1.55201 
Rb Non-cancerous NPE 12.6600 ± 1.04007 0.006 
 NPC 8.0343 ± 2.82198 
CDK4 Non-cancerous NPE 13.5360 ± 0.44200 0.008 
 NPC 10.7414 ± 0.38959 
p27 Non-cancerous NPE 5.3900 ± 0.82037 0.000 
 NPC 11.4743 ± 1.93016 
p19 Non-cancerous NPE 7.1160 ± 0.78325 0.007 
 NPC 12.0800 ± 3.16098 
cyclin D1 Non-cancerous NPE 11.9640 ± 0.31382 0.035 
 NPC 8.7657 ± 2.88830 
CDK8 Non-cancerous NPE 13.2240 ± 0.49303 0.095 
 NPC 11.1429 ± 2.45628 
DP2 Non-cancerous NPE 11.3460 ± 0.74527 0.697 
 NPC 10.7914 ± 2.98937 

Protein expression of cell cycle regulators using NPC-specific TMA

The overexpression of cyclin C, CDK8, and E2F6 protein could be seen in NPC, histologically normal NPE, and dysplastic NPE, but no differences were seen between them. In different clinical stages, cervix lymph node metastasis, and NPC pathogenic subtypes, no statistical difference of p16, p27, p19, CDK4, cyclin D1, Rb, and DP2 expression was found (data not shown). Some of the most striking findings were the loss of p16, p27, and p19 expression in NPC and the overexpression of cyclin D1, CDK4, and Rb in NPC. Representative images could be seen in Fig. 2.

Among the down-regulated proteins, the expression rate of p16 in NPC samples, in dysplastic NPE samples, and in histologically normal NPE samples was 47.4%, 64.3%, and 82.3%, respectively (P = 0.025, 0.013, and 0.000); Moreover, for p27 and p19, the expression rates were 50.5%, 56.9%, 74.2% (P = 0.389, 0.029, and 0.000); and 49.8%, 52.6%, 72.1% (P = 0.701, 0.017, and 0.001), respectively. Among the up-regulated proteins, the expression rate of cyclin D1 in NPC samples, in dysplastic NPE samples, and in histologically normal NPE samples was 91.9%, 81.4%, and 71.7%, respectively (P = 0.019, 0.180, and 0.000). The expression rate for Rb, CDK4, and DP2 was 90.6%, 70.9%, 56.3% (P = 0.000, 0.084, and 0.000); 62.7%, 51.9%, 46.2% (P = 0.144, 0.511, and 0.007); and 94.7%, 87.9%, 85.3% (P = 0.080, 0.642, and 0.006), respectively (Fig. 3).

Simultaneously, we compared the abnormal expression of p16, CDK4, cyclin D1, Rb, DP2, E2F6, p19, and p27 proteins between NPC and the matched adjacent epithelia of NPC (Table 6). Significant differences were observed in p16, p27, p19, CDK4, cyclin D1, and Rb proteins (P = 0.005, 0.007, 0.030, 0.028, 0.004, and 0.000, respectively), but no statistical difference was seen in E2F6 and DP2 proteins.

Table 6

Abnormal expression of cell cycle regulators between NPC and the matched adjacent epithelia of NPC

Variable NPC (%) The matched adjacent epithelia of NPC (%) P-value 
Rb 183/202 (90.6) 46/67 (70.1) 0.000 
cyclin D1 193/210 (91.9) 58/73 (79.5) 0.004 
E2F6 213/219 (97.3) 67/72 (93.1) 0.147 
p19 101/203 (49.8) 42/65 (64.6) 0.030 
DP2 197/208 (94.7) 63/71 (88.7) 0.101 
p27 101/204 (50.5) 47/69 (68.1) 0.007 
CDK4 133/212 (62.7) 34/71 (47.9) 0.028 
p16 100/211 (47.4) 47/68 (69.1) 0.005 
Variable NPC (%) The matched adjacent epithelia of NPC (%) P-value 
Rb 183/202 (90.6) 46/67 (70.1) 0.000 
cyclin D1 193/210 (91.9) 58/73 (79.5) 0.004 
E2F6 213/219 (97.3) 67/72 (93.1) 0.147 
p19 101/203 (49.8) 42/65 (64.6) 0.030 
DP2 197/208 (94.7) 63/71 (88.7) 0.101 
p27 101/204 (50.5) 47/69 (68.1) 0.007 
CDK4 133/212 (62.7) 34/71 (47.9) 0.028 
p16 100/211 (47.4) 47/68 (69.1) 0.005 

Pair-wise association between abnormal expression of EBER-1 hybridization signals and cell cycle regulators

Correlations between the immunohistochemical expressions of cell cycle regulators and EBER-1 hybridization signals are summarized in Table 7. In NPC, the expressions of Rb, cyclin D1, and CDK4 were strongly and positively correlated with EBER-1 hybridization signals (P = 0.000 for Rb, P = 0.000 for cyclin D1, and P = 0.000 for CDK4), whereas the expression of p16 was negatively correlated with EBER-1 hybridization signals (P = 0.039).

Table 7

Pair-wise association between abnormal expression of EBER-1 hybridization signals, p27, p16, CDK4, DP2, cyclin D1, E2F6, p19, and Rb proteins in NPC (Spearman's correlation)

 p27 p16 p19 CDK4 cyclin D1 Rb E2F6 DP2 
EBER-1 
 Spearman correlation coefficient 0.087 −0.119 0.058 0.307** 0.390** 0.415** 0.064 0.102 
 Significance (two-tailed) 0.775 0.039 0.335 0.000 0.000 0.000 0.279 0.089 
p27 
 Spearman correlation coefficient  0.112* 0.275** 0.227** 0.114* 0.115* 0.297** 0.189** 
 Significance (two-tailed)  0.036 0.000 0.000 0.035 0.038 0.000 0.000 
p16 
 Spearman correlation coefficient   0.167** 0.146** 0.091 0.002 −0.017 0.019 
 Significance (two-tailed)   0.002 0.006 0.087 0.970 0.744 0.719 
p19 
 Spearman correlation coefficient    0.207** 0.121* 0.126* 0.180** 0.216** 
 Significance (two-tailed)    0.000 0.026 0.024 0.001 0.000 
CDK4 
 Spearman correlation coefficient     0.534** 0.283** 0.175** 0.262** 
 Significance (two-tailed)     0.000 0.000 0.001 0.000 
cyclin D1 
 Spearman correlation coefficient      0.347** 0.328** 0.299** 
 Significance (two-tailed)      0.000 0.000 0.000 
Rb 
 Spearman correlation coefficient       0.203** 0.182** 
 Significance (two-tailed)       0.000 0.001 
E2F6 
 Spearman correlation coefficient        0.526** 
 Significance (two-tailed)        0.000 
 p27 p16 p19 CDK4 cyclin D1 Rb E2F6 DP2 
EBER-1 
 Spearman correlation coefficient 0.087 −0.119 0.058 0.307** 0.390** 0.415** 0.064 0.102 
 Significance (two-tailed) 0.775 0.039 0.335 0.000 0.000 0.000 0.279 0.089 
p27 
 Spearman correlation coefficient  0.112* 0.275** 0.227** 0.114* 0.115* 0.297** 0.189** 
 Significance (two-tailed)  0.036 0.000 0.000 0.035 0.038 0.000 0.000 
p16 
 Spearman correlation coefficient   0.167** 0.146** 0.091 0.002 −0.017 0.019 
 Significance (two-tailed)   0.002 0.006 0.087 0.970 0.744 0.719 
p19 
 Spearman correlation coefficient    0.207** 0.121* 0.126* 0.180** 0.216** 
 Significance (two-tailed)    0.000 0.026 0.024 0.001 0.000 
CDK4 
 Spearman correlation coefficient     0.534** 0.283** 0.175** 0.262** 
 Significance (two-tailed)     0.000 0.000 0.001 0.000 
cyclin D1 
 Spearman correlation coefficient      0.347** 0.328** 0.299** 
 Significance (two-tailed)      0.000 0.000 0.000 
Rb 
 Spearman correlation coefficient       0.203** 0.182** 
 Significance (two-tailed)       0.000 0.001 
E2F6 
 Spearman correlation coefficient        0.526** 
 Significance (two-tailed)        0.000 

*Correlation is significant at the 0.05 level (two-tailed); **correlation is significant at the 0.01 level (two-tailed).

Logistic regression analysis of abnormal expression of EBER-1 hybridization signals and p27, p16, CDK4, DP2, cyclin D1, E2F6, p19, and Rb proteins related to nasopharyngeal carcinogenesis

To determine whether differential expressions of EBER-1, p27, p16, CDK4, DP2, cyclin D1, E2F6, p19, and Rb might serve as new predictive factors for nasopharyngeal carcinogenesis, we comprehensively evaluated and compared the statistical data. In the preliminary univariate analysis giving rise to the final logistic regression analysis to evaluate and compare independent contributions of the molecules, the final model is shown in Table 8. The positive hybridization signal of EBER-1 (the hazard ratio (HR), given by the expression Exp (B) is 4.413; 95% confidence interval [CI] 2.359–8.254; P < 0.001) was the most independent predictor of nasopharyngeal carcinogenesis. The overexpression of Rb (HR = 2.870, 95% CI 1.452–5.670, P = 0.002), cyclin D1 (HR = 2.083, 95% CI 1.096–3.961, P = 0.025), E2F6 (HR = 1.592, 95% CI 1.005–2.523, P = 0.048), and the loss expression of p16 protein (HR = 0.572, 95% CI 0.369–0.886, P = 0.012) was the independent predictors of nasopharyngeal carcinogenesis.

Table 8

Multivariate logistic regression analysis of abnormal expression of EBER-1 hybridization signals and cell cycle regulators related to nasopharyngeal carcinogenesis

Factor B SE Wald Df Significance Exp (B95% CI for Exp (B)
 
       Lower Upper 
EBER-1 1.484 0.319 21.589 0.000 4.413 2.359 8.254 
Rb 1.054 0.347 9.207 0.002 2.870 1.452 5.670 
cyclin D1 0.734 0.328 5.017 0.025 2.083 1.096 3.961 
E2F6 0.465 0.235 3.919 0.048 1.592 1.005 2.523 
p19 0.233 0.448 0.272 0.602 1.263 0.525 3.036 
DP2 0.205 0.320 0.410 0.522 1.227 0.655 2.298 
p27 −0.152 0.371 0.167 0.683 0.859 0.415 1.780 
CDK4 −0.170 0.431 0.155 0.694 0.844 0.363 1.962 
p16 −0.559 0.224 6.243 0.012 0.572 0.369 0.886 
Factor B SE Wald Df Significance Exp (B95% CI for Exp (B)
 
       Lower Upper 
EBER-1 1.484 0.319 21.589 0.000 4.413 2.359 8.254 
Rb 1.054 0.347 9.207 0.002 2.870 1.452 5.670 
cyclin D1 0.734 0.328 5.017 0.025 2.083 1.096 3.961 
E2F6 0.465 0.235 3.919 0.048 1.592 1.005 2.523 
p19 0.233 0.448 0.272 0.602 1.263 0.525 3.036 
DP2 0.205 0.320 0.410 0.522 1.227 0.655 2.298 
p27 −0.152 0.371 0.167 0.683 0.859 0.415 1.780 
CDK4 −0.170 0.431 0.155 0.694 0.844 0.363 1.962 
p16 −0.559 0.224 6.243 0.012 0.572 0.369 0.886 

B, the coefficient for the constant; SE, standard error; Wald, the Wald chi-square test; df, degree of freedom; significance, the P-value; Exp (B), the exponentiation of the B coefficient (odds ratio).

Cluster analysis and display

We used an unsupervised hierarchical clustering algorithm [20] to analyze and visualize the data generated in this pilot study using NPC-specific TMA. The results are presented graphically in Fig. 5. Using the two-dimensional clustering algorithm for both protein expression and tumors, we could cluster the protein biomarkers according to the similarity of their expression patterns in tumors and the tumors based on the similarity of expression of different biomarkers. The 11 protein biomarkers were clustered into two groups. The first was a ‘proliferation-related’ cluster that included CDK8, cyclin D1, phosphorylated Rb, DP2, E2F6, Rb, cyclin C, and CDK4. The second was cyclin-dependent kinases inhibitors (CDKIs) family that included p27, p19, and p16. Also the nasopharyngeal specimens were clustered into two main groups. The first cluster was mainly composed of NPC with the loss of expression of p27, p19, and p16 proteins, and the overexpression of cyclin D1, Rb, CDK8, and DP2. The second cluster was histologically normal NPE and atypical hyperplastic NPE with high expression of p27, p19, and p16, whereas the low expression of cyclin D1, CDK4, and CDK8, respectively.

Fig. 5

Unsupervised hierarchical clustering algorithm was used to analyze and visualize the data for the amount of information generated in this pilot study by NPC-specific TMA  A row corresponds to a single tumor, and each column corresponds to a single protein biomarker. Red means strongly positive, brown means weakly positive, green means negative, and black means data not present. For each tumor, the protein expression score is represented by the color of corresponding cell in the matrix. The length and pattern of the branches of the vertical and horizontal dendrograms reflect the similarity between the tumors and the similarity between the markers, respectively. The shorter the arm of the branches, the greater the similarity of the cases and markers.

Fig. 5

Unsupervised hierarchical clustering algorithm was used to analyze and visualize the data for the amount of information generated in this pilot study by NPC-specific TMA  A row corresponds to a single tumor, and each column corresponds to a single protein biomarker. Red means strongly positive, brown means weakly positive, green means negative, and black means data not present. For each tumor, the protein expression score is represented by the color of corresponding cell in the matrix. The length and pattern of the branches of the vertical and horizontal dendrograms reflect the similarity between the tumors and the similarity between the markers, respectively. The shorter the arm of the branches, the greater the similarity of the cases and markers.

Survival analysis of cell cycle regulators in NPC patients

Among these cell cycle regulators, only cyclin D1 could be the prognosis biomarker for NPC patients. EFS of NPC patients with cyclin D1 positive expression was significantly higher than those with cyclin D1 negative expression (60.6 vs. 30.0%, P = 0.049). However, the OS was not significantly different between the cyclin D1 positive expression and the cyclin D1 negative expression group (69.1% vs. 60.0%) (Fig. 6).

Fig. 6

Kaplan–Meier curve of the recurrence depending on the expression of cyclin D1  (A) EFS of the two groups of cyclin D1 expression is distinguished: positive expression (grey) and negative expression (black), P = 0.049, log-rank test. (B) OS of the two groups of cyclin D1 expression is not significant: positive expression (grey) and negative expression (black), P = 0.499, log-rank test.

Fig. 6

Kaplan–Meier curve of the recurrence depending on the expression of cyclin D1  (A) EFS of the two groups of cyclin D1 expression is distinguished: positive expression (grey) and negative expression (black), P = 0.049, log-rank test. (B) OS of the two groups of cyclin D1 expression is not significant: positive expression (grey) and negative expression (black), P = 0.499, log-rank test.

Discussion

This study revealed activation of globally affected differential signaling pathway by cDNA microarray and GSEA, and found that the cell cycle pathway was the most disregulated signaling pathway in EBV-associated NPC.

The classical approach to DNA microarray analysis has been to treat genes as independent agents, to apply some statistical tests to per gene and follow that up with some forms of the P-value correction method. GSEA [13–15] is supplementary to single-gene approaches, and provides a framework with which we may examine changes at a higher level of biological organization. Gene-set analysis evaluates the expression of biological pathways or previously defined gene sets, rather than that of individual genes, in association with a binary phenotype, and is of great biologic interest in many DNA microarray studies. GSEA has been applied widely as a tool for gene-set analysis. In this study, we used GSEA to analyze the microarray data in the gene set of Database C2 of the MSigDB version 2.0 in which most pathways are concerned specific metabolic and signaling pathways. The results showed that cell cycle pathway was the most disregulated pathway with a nominal P-value of 0.000, enrichment score (ES) of −0.69, NES of −2.22, FDR q-value of 0.001, suggesting that the cell cycle pathway was closely associated with NPC development.

The mechanism of cell cycle regulation is the reciprocity of cyclins, cyclin-dependent kinases (CDKs), and CDKIs [21]. Alterations of any component of the pathway, such as deletion or mutation of the p16 gene, amplification or the overexpression of CDKs of cyclin D, and mutations to CDKs that affect p16 binding, will lead to Rb phosphorylation and subsequent progression of G1 into S phase transition [22,23]. These alterations have been found in many human tumors, suggesting that inactivation of the cell cycle pathway may play an important role in their pathogenesis, but detailed and comprehensive explanation about the aberrant activation of cell cycle pathway in NPC have not yet been reported.

In this study, some members of the cell cycle pathway had been screened by analyzing the gene expression profiles of NPC, using GSEA software and confirmed by the quantitative real-time RT–PCR and TMA at the mRNA and protein levels of p16, p27, p19, Rb, CDK4, cyclin D1, CDK8, cyclin C, E2F6, phosphorylated Rb, and DP2. In our work, loss of p16 expression was observed in NPC specimens compared with histologically normal NPE, dysplastic NPE, and the matched adjacent epithelia of NPC. Loss of p16 expression in dysplastic NPE was also lower than that in histologically normal NPE. These findings accorded with other previously reported results [2, 24–28], suggesting that inactivation of p16 gene might be the early molecular event of nasopharyngeal carcinogenesis and might significantly correlate with the development of NPC. However, in our study there existed the overexpression of EBV LMP1. Zhao et al. [29] had reported that LMP1 could inhibit p16 expression, induce pRb phosphorylation, up-regulate E2F1 transactivity and expression, and promote the progression of G1/S phase. Song et al. [30] reported that the decrease of p16 cooperated with cyclin D1 and caused deregulation of G1/S checkpoint, leading to abnormal cell proliferation in NPC. However, the precise mechanism of p16 inactivation in NPC is not clear. It may be inactivated through homozygous deletion, promoter hypermethylation, and point mutation [31]. Our current data of multivariate analysis showed that p16 was an independent predictor of nasopharyngeal carcinogenesis. The present study suggested that p27 protein expression was reduced in NPC specimens and dysplastic NPE compared with histologically normal NPE, which is in accordance with previous reports [2]. Loss of p27 protein expression may result in tumor development and/or progression; however, this loss of expression does not appear to be resulted from gene mutations according to some studies [32,33]. Oh et al. [34] reported an association between p27 and cyclin D1 expression, which indicated that the balance between the two opposing regulators was important for the end result of cell cycle progression. cyclin D1 is an important member of cell cycle pathway, so we decided to look at its expression though there was not any clone representing the cyclin D1 gene in this cDNA microarray. The positive expression rate of cyclin D1 protein in NPC tissues was as high as 91.9%, consistent with the previous report [35]. Multivariate analysis showed that cyclin D1 was an independent predictor of nasopharyngeal carcinogenesis. Despite advances in imaging technology and treatment techniques, about 50% of patients had NPC recurrence after receiving adequate treatment for the nasopharynx. This highlights the importance of cyclin D1 in early prediction of recurrence of NPC [24]. Some studies have provided in vivo evidence demonstrating that low levels of cyclin D1 are associated with ipsilateral breast cancer and early-stage laryngeal cancer recurrence following radiation therapy [36,37]. Our data suggested that patients with positive expression of cyclin D1 protein had better prognosis than those with negative expression of cyclin D1, indicating that cyclin D1 protein expression may be useful as a predictor of local tumor control in NPC. Zhao et al. [38] reported that LMP1 could up-regulate the transcription of cyclin D1 through NF-κB signaling pathway, thus promoting the transition of the cell cycle from G1 to S phase, and leading to premature S phase entry. Hui et al. [28] revealed that both inactivation of p16 and activation of cyclin D1 might be important in the alteration of cell cycle controls and development of NPC. Some papers [39,40] reported that EBV LMP1 could increase CDK4 expression, promote its nuclear translocation, and induce cyclin D1 expression. The expression of CDK4 was much higher in NPC than in histologically normal NPE. The activity of CDK4 kinase is restricted to the G1-S phase, which is controlled by the regulatory subunits D-type cyclins and CDK inhibitor p16 (INK4a). This kinase was shown to be responsible for the phosphorylation of Rb protein. The increases of cyclin D1 and CDK4 expression were found to be associated with tumorigenesis of a variety of cancers [41–44]. In addition to this, our data suggested that 90.6% of the NPC specimens expressed Rb, which was much higher than the matched adjacent epithelia of NPC, histologically normal NPE, and dysplastic NPE. However, immunohistochemical stains are blind to the phosphorylation status of Rb, so identification of Rb in tissue sections does not exclude functional inactivation of this protein. Chen et al. [45] had suggested that NPC transformation depended on timely regulation of Rb and several transcriptional cascades, interconnected by E2F, AP-1, NF-κB, and STAT3 among others during latent and lytic cycles. Multivariate analysis showed that Rb was an independent predictor of nasopharyngeal carcinogenesis. Furthermore, unsupervised hierarchical clustering in NPC-specific TMA revealed that one group belonged to the CDKI family, which included p27, p19, and p16; whereas the other group was ‘proliferation-related’ proteins that included CDK8, cyclin D1, DP2, E2F6, Rb, cyclin C, phosphorylated Rb, and CDK4. If the expression of these proteins is viewed together, most tumors contained abnormal expression of at least one of p16, cyclin D1, CDK4, and Rb. This revealed that cell cycle progression is regulated not by a single cell cycle regulator, but by a balance of negative and positive regulators. This observation confirms the fundamental role of the p16-cyclin D1/CDK4-Rb pathway in NPC tumorigenesis.

As many studies [39,46–50] and our previous studies [2,3] had reported that NPC was closely associated with EBV infection, and all NPC specimens served in the cDNA microarray analysis were positive for EBV. It was reported [3,51] that EBNA1 was consistently expressed in NPC tumors and was a useful marker in the clinical treatment of NPC. The expression of EBNA1 in this study confirmed that EBV infection was consistently observed in all NPC samples. Furthermore, we detected the expression of EBER-1 and LMP1, together with those abnormally expressed genes, which were included in the cell cycle pathway using NPC-specific TMA. Our result suggested that the expression of EBER-1 and LMP1 existed in most of NPC patients. As for the relationship between abnormally regulated cell cycle pathway and EBV infection, we found that in NPC, EBER-1 was strongly and positively correlated with CDK4, cyclin D1, and Rb expression, yet negatively correlated with the p16 expression, suggesting that the abnormally regulated cell cycle pathway might be associated with EBV infection. Through multivariate analysis, we found that EBER-1 was the most significant and independent predictor of nasopharyngeal carcinogenesis from the non-cancerous NPE to NPC. These findings suggested a strong association between EBV infection and abnormally regulated cell cycle pathway in NPC, but the exact mechanism needs further study.

In summary, we combined the cDNA microarray with microdissection to comprehensively analyze the gene expression profiles between NPC and non-cancerous NPE samples. Using GSEA algorithm, we found that the cell cycle pathway was the most disregulated pathway in NPC samples. Abnormally activated cell cycle pathway was identified to be potentially biologically meaningful in NPC by NPC-specific TMA.

Funding

This work was supported by grants from the National Key Project of Scientific Research Program (2006CB910502, 2006CB910504), the National Natural Science Foundation of China (30700469, 30770825, 30871282, 30871356), the New Century Excellent Talents in University (NECT 04-0761), the Foundation for the Author of National Excellent Doctoral Dissertation of the PR China (200559), the 111 project (111-2-12), the National ‘863’ High Technology Program of China (2007AA02Z170), and the Hunan Provincial Natural Science Foundation of China (06JJ20013).

References

1
Xiong
W
Zeng
ZY
Xia
JH
Xia
K
Shen
SR
Li
XL
Hu
DX
, et al.  . 
A susceptibility locus at chromosome 3p21 linked to familial nasopharyngeal carcinoma
Cancer Res
 , 
2004
, vol. 
64
 (pg. 
1972
-
1974
)
2
Fan
SQ
Ma
J
Zhou
J
Xiong
W
Xiao
BY
Zhang
WL
Tan
C
, et al.  . 
Defferential expression of Epstein–Barr virus-encoded RNA and several tumor-related genes in various types of nasopharyngeal epithelial lesions and nasopharyngeal carcinoma using tissue microarray analysis
Hum Pathol
 , 
2006
, vol. 
37
 (pg. 
593
-
605
)
3
Zeng
ZY
Zhou
YH
Zhang
WL
Xiong
W
Fan
SQ
Li
XL
Luo
XM
, et al.  . 
Gene expression profiling of nasopharyngeal carcinoma reveals the abnormally regulated Wnt signaling pathway
Hum Pathol
 , 
2007
, vol. 
38
 (pg. 
120
-
133
)
4
Morrison
JA
Gulley
ML
Pathmanathan
R
Raab-Traub
N
Differential signaling pathways are activated in the Epstein–Barr virus-associated malignancies nasopharyngeal carcinoma and Hodgkin lymphoma
Cancer Res
 , 
2004
, vol. 
64
 (pg. 
5151
-
5260
)
5
Hildesheim
A
Levine
PH
Etiology of nasopharyngeal carcinoma: a review
Epidemiol Rev
 , 
1993
, vol. 
15
 (pg. 
466
-
485
)
6
Chou
J
Lin
YC
Kim
J
You
L
Xu
Z
He
B
Jablons
DM
Nasopharyngeal carcinoma-review of the molecular mechanisms of tumorigenesis
Head Neck
 , 
2008
, vol. 
30
 (pg. 
946
-
963
)
7
Song
G
Ouyang
G
Bao
S
The activation of Akt/PKB signaling pathway and cell survival
J Cell Mol Med
 , 
2005
, vol. 
9
 (pg. 
59
-
71
)
8
English
J
Pearson
G
Wilsbacher
J
Swantek
J
Karandikar
M
Xu
S
Cobb
MH
New insights into the control of MAP kinase pathways
Exp Cell Res
 , 
1999
, vol. 
253
 (pg. 
255
-
270
)
9
Soo
R
Putti
T
Tao
Q
Goh
BC
Lee
KH
Kwok-Seng
L
Tan
L
, et al.  . 
Overexpression of cyclooxygenase-2 in nasopharyngeal carcinoma and association with epidermal growth factor receptor expression
Arch Otolaryngol Head Neck Surg
 , 
2005
, vol. 
131
 (pg. 
147
-
152
)
10
Carter
KL
Cahir-McFarland
E
Kieff
E
Epstein–Barr virus induced changes in B-lymphocyte gene expression
J Virol
 , 
2002
, vol. 
76
 (pg. 
10427
-
10436
)
11
Jones
JO
Arvin
AM
Microarray analysis of host cell gene transcription in response to varicella-zoster virus infection of human T cells and fibroblasts in vitro and SCIDhu skin xenografts in vivo
J Virol
 , 
2003
, vol. 
77
 (pg. 
1268
-
1280
)
12
Kawamata
H
Furihata
T
Omotehara
F
Sakai
T
Horiuchi
H
Shinagawa
Y
Imura
J
, et al.  . 
Identification of genes differentially expressed in a newly isolated human metastasizing esophageal cancer cell line, T.Tn-AT1, by cDNA microarray
Cancer Sci
 , 
2003
, vol. 
94
 (pg. 
699
-
706
)
13
Mootha
VK
Lindgren
CM
Eriksson
KF
Subramanian
A
Sihag
S
Lehar
J
Puigserver
P
, et al.  . 
PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes
Nat Genet
 , 
2003
, vol. 
34
 (pg. 
267
-
273
)
14
Subramanian
A
Tamayo
P
Mootha
VK
Mukherjee
S
Ebert
BL
Gillette
MA
Paulovich
A
, et al.  . 
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
Proc Natl Acad Sci USA
 , 
2005
, vol. 
102
 (pg. 
15545
-
15550
)
15
Sweet-Cordero
A
Mukherjee
S
Subramanian
A
You
H
Roix
JJ
Ladd-Acosta
C
Mesirov
J
, et al.  . 
An oncogenic KRAS2 expression signature identified by cross-species gene-expression analysis
Nat Genet
 , 
2005
, vol. 
37
 (pg. 
48
-
55
)
16
Luo
L
Salunga
RC
Guo
H
Bittner
A
Joy
KC
Galindo
JE
Xiao
H
, et al.  . 
Gene expression profiles of laser-captured adjacent neuronal subtypes
Nat Med
 , 
1999
, vol. 
5
 (pg. 
117
-
122
)
17
Crnogorac-Jurcevic
T
Nielsen
TO
Lemoine
NR
RT–PCR from laser-capture microdissected samples
Methods Mol Biol
 , 
2002
, vol. 
193
 (pg. 
197
-
204
)
18
Zhou
YH
Zeng
ZY
Xiong
W
Luo
XM
Li
XL
Fan
SQ
Zhang
WL
Microdissection and RNA liner-amplification of nasopharyngeal carcinoma tissue
Prog Biochem Biophys
 , 
2005
, vol. 
32
 (pg. 
463
-
467
)
19
Zhang
B
Nie
X
Xiao
B
Xiang
J
Shen
S
Gong
J
Zhou
M
, et al.  . 
Identification of tissue-specific genes in nasopharyngeal epithelial tissue and differentially expressed genes in nasopharyngeal carcinoma by suppression subtractive hybridization and cDNA microarray
Gene Chromosome Cancer
 , 
2003
, vol. 
38
 (pg. 
80
-
90
)
20
Liu
CL
Prapong
W
Natkunam
Y
Alizadeh
A
Montgomery
K
Gilks
CB
van de Rijn
M
Software tools for highthroughput analysis and archiving of immunohistochemistry staining data obtained with tissue microarrays
Am J Pathol
 , 
2002
, vol. 
161
 (pg. 
1557
-
1565
)
21
Hirama
T
Koeffler
HP
Role of the cyclin-dependent kinase inhibitors in the development of cancer
Blood
 , 
1995
, vol. 
86
 (pg. 
841
-
854
)
22
Zuo
L
Weger
J
Yang
Q
Goldstein
AM
Tucker
MA
Walker
GJ
Hayward
N
, et al.  . 
Germline mutations in the p16INK4a binding domain of CDK4 in familial melanoma
Nat Genet
 , 
1996
, vol. 
12
 (pg. 
97
-
99
)
23
Paggi
MG
Baldi
A
Bonetto
F
Giordano
A
Retinoblastoma protein family in cell cycle and cancer: a review
J Cell Biochem
 , 
1996
, vol. 
62
 (pg. 
418
-
430
)
24
Hwang
CF
Cho
CL
Huang
CC
Wang
JS
Shih
YL
Su
CY
Chang
HW
Loss of cyclin D1 and p16 expression correlates with local recurrence in nasopharyngeal carcinoma following radiotherapy
Ann Oncol
 , 
2002
, vol. 
13
 (pg. 
1246
-
1251
)
25
Mäkitie
AA
MacMillan
C
Ho
J
Shi
W
Lee
A
O'sullivan
B
Payne
D
, et al.  . 
Loss of p16 expression has prognostic significance in human nasopharyngeal carcinoma
Clin Cancer Res
 , 
2003
, vol. 
9
 (pg. 
2177
-
2184
)
26
Baba
Y
Tsukuda
M
Mochimatsu
I
Furukawa
S
Kagata
H
Satake
K
Koshika
S
, et al.  . 
Reduced expression of p16 and p27 proteins in nasopharyngeal carcinoma
Cancer Detect Prev
 , 
2001
, vol. 
25
 (pg. 
414
-
419
)
27
Huang
GW
Mo
WN
Kuang
GQ
Nong
HT
Wei
MY
Sunagawa
M
Kosugi
T
, et al.  . 
Expression of p16, nm23-H1, E-cadherin, and CD44 gene products and their significance in nasopharyngeal carcinoma
Laryngoscope
 , 
2001
, vol. 
111
 (pg. 
1465
-
1471
)
28
Hui
AB
Or
YY
Takano
H
Tsang
RK
To
KF
Guan
XY
Sham
JS
, et al.  . 
Array-based comparative genomic hybridization analysis identified cyclin D1 as a target oncogene at 11q13.3 in nasopharyngeal carcinoma
Cancer Res
 , 
2005
, vol. 
65
 (pg. 
8125
-
8133
)
29
Zhao
XR
Deng
L
Weng
XX
The effects of exogenous p16 expression on CDK4, Cyclin D1 and pRb in nasopharyngeal carcinoma cell lines
Hunan Yi Ke Da Xue Xue Bao
 , 
2000
, vol. 
25
 (pg. 
428
-
430
)
30
Song
X
Tao
YG
Zeng
L
Deng
XY
Lee
LM
Gong
JP
Wu
Q
, et al.  . 
Latent membrane protein 1 encoded by Epstein–Barr virus modulates directly and synchronously cyclin D1 and p16 by newly forming a c-Jun/Jun B heterodimer in nasopharyngeal carcinoma cell line
Virus Res
 , 
2005
, vol. 
113
 (pg. 
89
-
99
)
31
Lo
KW
Huang
DP
Genetic and epigenetic changes in nasopharyngeal carcinoma
Semin Cancer Biol
 , 
2002
, vol. 
12
 (pg. 
451
-
462
)
32
Spirin
KS
Simpson
JF
Takeuchi
S
Kawamata
N
Miller
CW
Koeffler
HP
p27Kip1 mutation found in breast cancer
Cancer Res
 , 
1996
, vol. 
56
 (pg. 
2400
-
2404
)
33
Croix
B St
Flørenes
VA
Rak
JW
Flanagan
M
Bhattacharya
N
Slingerland
JM
Kerbel
RS
Impact of the cyclin-dependent kinase inhibitor p27Kip1 on resistance of tumor cells to anti-cancer agents
Nat Med
 , 
1996
, vol. 
2
 (pg. 
1204
-
1210
)
34
Oh
YL
Choi
JS
Song
SY
Ko
YH
Han
BK
Nam
SJ
Yang
JH
Expression of p21waf1, p27kip1 and cyclin D1 proteins in breast ductal carcinoma in situ: relation with clinicopathologic characteristics and with p53 expression and estrogen receptor status
Pathol Int
 , 
2001
, vol. 
51
 (pg. 
94
-
99
)
35
Sriuranpong
V
Mutirangura
A
Gillespie
JW
Patel
V
Amornphimoltham
P
Molinolo
AA
Kerekhanjanarong
V
, et al.  . 
Global gene expression profile of nasopharyngeal carcinoma by laser capture microdissection and complementary DNA microarrays
Clin Cancer Res
 , 
2004
, vol. 
10
 (pg. 
4944
-
4958
)
36
Turner
BC
Gumbs
AA
Carter
D
Glazer
PM
Haffty
BG
Cyclin D1 expression and early breast cancer recurrence following lumpectomy and radiation
Int J Radiat Oncol Biol Phys
 , 
2000
, vol. 
47
 (pg. 
1169
-
1176
)
37
Yoo
SS
Carter
D
Turner
BC
Sasaki
CT
Son
YH
Wilson
LD
Glazer
PM
, et al.  . 
Prognostic significance of cyclin D1 protein levels in early-stage larynx cancer treated with primary radiation
Int J Cancer
 , 
2000
, vol. 
90
 (pg. 
22
-
28
)
38
Zhao
XR
Gu
HH
Weng
XX
Yi
W
Deng
XY
Cao
Y
The primary study on expression and function of D-type cyclins in nasopharyngeal carcinoma cell lines
Sheng Wu Hua Xue Yu Sheng Wu Wu Li Xue Bao
 , 
2000
, vol. 
32
 (pg. 
192
-
196
)
39
Ai
MD
Li
LL
Zhao
XR
Wu
Y
Gong
JP
Cao
Y
Regulation of survivin and CDK4 by Epstein–Barr virus encoded latent membrane protein 1 in nasopharyngeal carcinoma cell lines
Cell Res
 , 
2005
, vol. 
15
 (pg. 
777
-
784
)
40
Song
X
Zhao
XR
Wang
Y
Zhou
JH
Guan
XY
Cao
Y
Tissue micro-array analysis of cyclin D1 gene overexpression in the multistage carcinogenesis of nasopharyngeal carcinoma
Prog Biochem Biophys
 , 
2002
, vol. 
29
 (pg. 
385
-
389
)
41
Masaki
T
Shiratori
Y
Rengifo
W
Igarashi
K
Yamagata
M
Kurokohchi
K
Uchida
N
, et al.  . 
Cyclins and cyclin-dependent kinases: comparative study of hepatocellular carcinoma versus cirrhosis
Hepatology
 , 
2003
, vol. 
37
 (pg. 
534
-
543
)
42
Skomedal
H
Kristensen
GB
Lie
AK
Holm
R
Aberrant expression of the cell cycle associated proteins TP53, MDM2, p21, p27, CDK4, Cyclin D1, RB, and EGFR in cervical carcinomas
Gynecol Oncol
 , 
1999
, vol. 
73
 (pg. 
223
-
228
)
43
Chen
Q
Luo
G
Li
B
Samaranayake
LP
Expression of p16 and CDK4 in oral premalignant lesions and oral squamous cell carcinomas: a semi-quantitative immunohistochemical study
J Oral Pathol Med
 , 
1999
, vol. 
28
 (pg. 
158
-
164
)
44
Wunder
JS
Eppert
K
Burrow
SR
Gokgoz
N
Bell
RS
Andrulis
IL
Co-amplification and overexpression of CDK4, SAS and MDM2 occurs frequently in human parosteal osteosarcomas
Oncogene
 , 
1999
, vol. 
18
 (pg. 
783
-
788
)
45
Chen
X
Liang
S
Zheng
WL
Liao
ZJ
Shang
T
Ma
WL
Meta-analysis of nasopharyngeal carcinoma microarray data explores mechanism of EBV-regulated neoplastic transformation
BMC Genomics
 , 
2008
, vol. 
9
 pg. 
322
 
46
Tsao
SW
Tramoutanis
G
Dawson
CW
Lo
AK
Huang
DP
The significance of LMP1 expression in nasopharyngeal carcinoma
Semin Cancer Biol
 , 
2002
, vol. 
12
 (pg. 
473
-
487
)
47
Nicholls
J
Hahn
P
Kremmer
E
Fröhlich
T
Arnold
GJ
Sham
J
Kwong
D
, et al.  . 
Detection of wild type and deleted latent membrane protein 1(LMP1) of Epstein–Barr virus in clinical biopsy material
J Virol Methods
 , 
2004
, vol. 
116
 (pg. 
79
-
88
)
48
Yip
KW
Shi
W
Pintilie
M
Martin
JD
Mocanu
JD
Wong
D
MacMillan
C
, et al.  . 
Prognostic significance of the Epstein–Barr virus, p53, Bcl-2, and survivin in nasopharyngeal cancer
Clin Cancer Res
 , 
2006
, vol. 
12
 (pg. 
5726
-
5732
)
49
Huang
DP
Ho
HC
Henle
W
Henle
G
Saw
D
Lui
M
Presence of EBNA in nasopharyngeal carcinoma and control patient tissues related to EBV serology
Int J Cancer
 , 
1978
, vol. 
22
 (pg. 
266
-
274
)
50
Sam
CK
Brooks
LA
Niedobitek
G
Young
LS
Prasad
U
Rickinson
AB
Analysis of Epstein–Barr virus infection in nasopharyngeal biopsies from a group at high risk of nasopharyngeal carcinoma
Int J Cancer
 , 
1993
, vol. 
53
 (pg. 
957
-
962
)
51
Young
LS
Dawson
CW
Clark
D
Rupani
H
Busson
P
Tursz
T
Johnson
A
, et al.  . 
Epstein–Barr virus gene expression in nasopharyngeal carcinoma
J Gen Virol
 , 
1988
, vol. 
69
 (pg. 
1051
-
1065
)

Author notes

These authors contributed equally to this work.