EGFR and SYNE2 are associated with p21 expression and SYNE2 variants predict post-operative clinical outcomes in HBV-related hepatocellular carcinoma

This study was to explore the association between gene variants and p21 expression and investigate the TP53-independent p21 regulation in hepatitis B virus (HBV) related hepatocellular carcinoma (HCC) patients from Guangxi by genome-wide association study. 426 HBV-related HCC patients were enrolled. Results showed that, after quality control, a total of 21,643 SNPs were identified in 107 p21 positive and 298 p21 negative patients. The variants of epidermal growth factor receptor (EGFR; rs2227983 and rs6950826) and spectrin repeat containing, nuclear envelope 2 (SYNE2; rs8010699, rs4027405 and rs1890908) were associated with p21 expression. Moreover the haplotype block (rs2227983 and rs6950826, r2 = 0.378) in EGFR and the haplotype block in SYNE2 (rs8010699 was in strong LD with rs4027405 and rs1890908 (r2 = 0.91 and 0.70, respectively)) were identified, and the haplotype A-G of EGFR and haplotype G-A-A of SYNE2 were significantly associated with p21 expression (P < 0.01). rs4027405 and rs1890908 were significantly associated with overall survival, and patients with AG/GG genotypes of SYNE2 gene had a worse overall survival (P = 0.001, P = 0.002). Our findings indicate that variants of EGFR and SYNE2 play an important role in p21 regulation and are associated with the clinical outcome of HBV-related HCC in a TP53-indenpdent manner.

Scientific RepoRts | 6:31237 | DOI: 10.1038/srep31237 Hepatocarcinogenesis is an extremely complex process, and there is no a specific molecular mechanism that can elucidate this process. Investigations have demonstrated that the dys-regulation of cell cycle plays an important role in the occurrence and progression of HCC 6,7 . p21 is a major cell cycle regulator. It is a cyclin-dependent kinase (CDK) inhibitor and a downstream factor of TP53. It can causes G1 growth arrest and thus it is also as known as a tumor suppressor 8,9 . p21 may bind to and inhibit the CDK activity, which mediates its various biological activities and may cause cell cycle arrest at a specific stage 10,11 . p21 has been found as one of most important and effective effector molecules of TP53. TP53 is able to bind to the promoter of p21 gene, leading to the direct activation of p21 expression 8,12 . Studies have shown that p21 can also be regulated in TP53-independent manners, such as by MYC, E2A and BRCA1 13,14 . In addition, it is controversial on whether the HBX gene of HBV affects the p21 regulation. Some studies report that HBX gene can suppress p21 expression 15,16 , but this is not found in other studies 17,18 .
Although p21 is a tumor suppressor, it may also promote oncogenesis in certain circumstances. p21-deficient mice may spontaneously develop tumors, supporting the tumor suppressor activity of p21 19 . Previous studies have reported that p21 mRNA or protein expression in HCC tissues is often lower than in adjacent normal tissues [20][21][22] . These results suggest that p21 might act as a tumor suppressor in HCC. Moreover, the spontaneous development of lymphomas is suppressed in p21-deficient mice with TP53-or ATM-deficiency 23,24 , indicating that p21 serves as an oncogene. In fact, p21 is over-expressed in a variety of cancers, such as breast cancer, prostate cancer, oesophageal squamous cell carcinoma and cervical cancer, and p21 up-regulation as a poor prognostic biomarker is closely related to the tumor grade, invasiveness and aggressiveness in many cases 11,[25][26][27] .
Kao et al. 28 and Zhang et al. 29 reported that p21 expression was a favorable survival factor in HCC, but this was not found in our prior study. This may be ascribed to the difference in patients selected. In their study, the HBV-related and non-HBV-related HCC patients were recruited, but only HBV-related HCC patients were enrolled into our study. Furthermore, different from patients in other regions, patients in Guangxi have a high level exposure to AFB1 in their food which is closely associated with hepatocarcinogenesis 5 . Study has reported that AFB1 exposure leads to a high mutation frequency at TP53 249Ser, which has been considered as a molecular fingerprint of HCC pathogenesis 30 . However, at present, whether AFB1 exposure is involved in the regulation of p21 is unclear.
In the present study, a genome-wide association study (GWAS) using Illumina HumanExome BeadChip-12-1_A was performed in Chinese HBV-related HCC patients from Guangxi, aiming to detect the gene variants associated with p21 expression and reveal the TP53-independent p21 regulation.

Results
GWAS. Baseline Characteristics. The clinicopathological characteristics are shown in Table 1. There were no significant differences in the age, gender, BMI, race, smoking status, drinking status, Barcelona Clinic Liver Cancer (BCLC) stage, serum alpha fetoprotein (AFP) level, transcatheter arterial chemoembolization (TACE) status, pathological grade and TP53 249Ser mutation between p21 negative group and p21 positive group (P > 0.05). TP53 expression status and hepatic cirrhosis status were significantly different between p21 negative group and p21 positive group (P = 0.005 and 0.001, respectively). A significantly positive relationship was observed between p21 expression status and TP53 expression status (r = 0.144, P < 0.01).
Quality Control. After quality control (QC), a total of 21,643 SNPs were identified in 107 p21 positive patients and 298 p21 negative patients. The total genotyping rate in remaining individuals was 98.34%. The principal component analysis (PCA) plot ( Fig. 1A) showed that no or mild population stratification in this study population, as indicated by the Quantile-Quantile (Q-Q) plot (Fig. 1B). The genomic control inflation factor (λ ) was 1.018.
Linkage disequilibrium (LD) and Haplotype Analysis. Analysis was performed for LD and haplotype about 2 Mb nearby EGFR and SYNE2 genes (Fig. 2) the haplotype block in SYNE2 (rs8010699 was in strong LD with rs4027405 and rs1890908 (r 2 = 0.91 and 0.7, respectively)) were identified, and the haplotype A-G of EGFR and haplotype G-A-A of SYNE2 were significantly associated with p21 expression (P = 2.43 × 10 −5 and 8.63 × 10 −4 , respectively; Table 4).
Pathway Analysis and Correlation analysis in mRNA. After above analysis, GeneMania Software 31 was employed (see URLs) to predict the interaction between genes shown in Table 2. Signaling pathway analysis (Fig. 3) showed   that p21, SYNE2 and EGFR were in a common pathway or co-expressed. Furthermore, the mRNA expression of p21, SYNE2 and EGFR was compared between HCC tissues and adjacent normal tissues. Data from Gene Expression Omnibus (GEO accession: GSE14520) showed, when compared with adjacent normal tissues, down-regulated expression of p21, SYNE2 and EGFR was observed in HCC tissues (Fig. 4A). Correlation analysis was used to analyze the correlation among p21, SYNE2 and EGFR ( Fig. 4B-D). Results showed that there were positive correlations among three genes (SYNE2 vs. p21, P < 0.001, r = 0.227; SYNE2 vs. EGFR, P < 0.001, r = 0.281; EGFR vs. p21, P < 0.001, r = 0.237).
Stratified analysis on the association of SYNE2 SNPs (rs4027405 and rs1890908) with survival. A stratified analysis was performed to evaluate the associations of rs4027405 and rs1890908 genotypes with HBV-related HCC survival by age, smoking status, drinking status, serum AFP level, radical resection, TP53 expression status, intrahepatic metastasis, vascular invasion, antiviral therapies, child-pugh class and BCLC stage (Fig. 6). Patients with rs4027405 AG/GG had a poorer OS when they were older than 45 years, or had AFP > 400 ng/ml, non-smoking status, non-drinking status, negative TP53 expression, child-pugh class B or BCLC stage C. Specifically, a significantly increased risk for death conferred by SNP rs4027405 genotypes (AG/GG) was observed in patients negative for TP53 expression (P = 0.001, HR = 2.82, 95%CI = 1.51-5.27) or child-pugh class B (P = 0.004, HR = 3.00, 95%CI = 1.41-6.38). Similar results were also observed in rs1890908.
Combined effect of SYNE2 SNPs (rs4027405 and rs1890908) and serum AFP on OS. The combined effect of SNPs and serum AFP on the OS of HBV-related HCC patients was further analyzed (   (Table 6).

Discussion
In this study, GWAS was performed to investigate the p21 regulation in Chinese HBV-related HCC patients in Guangxi. Results showed several SNPs in EGFR gene (rs6950826, rs2227983) and SYNE2 gene (rs4027405,     rs1890908 and rs8010699) were associated with p21 expression, and the mRNA expression of p21, EGFR and SYNE2 had positive correlations between each other. Furthermore, the p21 expression was significantly   associated with the cirrhosis status and TP53 expression, which was consistent with previous results 11,32,33 , but not related to TP53 Ser249 mutation (AFB1 exposure). According to the NCBI database (NG_007726), EGFR gene is mapped to chromosome 7p12 spanning 188.3 kb and contains 30 exons. EGFR is also known as ErbB1 or HER-1 and is a transmembrane receptor belonging to the family of receptor tyrosine kinases (RTK) including ErbB2/HER2, ErbB3/HER3 and ErbB4/HER4 34 . Structurally, these ligands can bind to EGFR system which contains EGF-like domains, including epidermal growth factor (EGF), transforming growth factor α (TGF-α ), amphiregulin (AR), epiregulin (EREG), betacellulin (BTC), heparin-binding EGF (HB-EGF) and epigen (EPGN) 35 . The EGFR system is not only essential for the cell proliferation, survival and migration, but closely related to the occurrence, growth and outcome of HCC [36][37][38] . In addition, in transgenic animals, results showed the over-expression of different EGFR ligands lead to different incidence of HCC, such as TGF-α or EGF over-expression leads to a higher incidence of HCC [39][40][41] . There is evidence showing that EGFR expression is related to p21 expression 42 , which was confirmed in the present study that there was correlation between EGFR and p21 mRNA expression (Fig. 4). In addition, our results showed rs2227983 (G > A, Arg 521 Lys) and rs6950826 (G > A) of EGFR gene were associated with p21 expression. rs2227983 (G > A, R521K) is located in the exon 13 of EGFR gene. Previous study showed, in metastatic colorectal cancer patients treated with EGFR inhibitor, patients harboring at least one minor allele of rs2227983 G > A were more likely to show a higher tumor response as compared to those with homozygous wild-type allele 43 . In our study, the genetic model of rs2227983 was additive. Patients with allele G had a higher p21 expression. This may be explained as that rs2227983 of EGFR gene may result in an Arg (R) to Lys(K) substitution, leading to the decrease in EGFR activity 44 . Besides, several studies have shown that rs2227983 is also associated with the interstitial lung disease 45 and gastric cancer 46 . rs6950826 is located in the intron 7 of EGFR gene. To date, no study has been conducted to investigate the association of rs6950826 with the susceptibility to disease or its influence on the clinical outcome of a specific disease. In our study, rs2227983 and rs6950826 were found to be not associated with OS of HBV-related HCC patients. However, how these two loci regulate the expression of EGFR and p21 is still unclear. According to the available findings, we speculate that both loci are likely to affect the function or structure of EGFR gene to regulate the p21 expression, but it is required to be further studied.

a Others include haplotypes G-G and A-A. b Others include haplotypes A-G-A and G-G-G.
In our study, results showed that SNPs (rs4027405, rs1890908 and rs8010699) of SYNE2 gene were associated with p21 expression, and these three loci had strong linkage disequilibrium. Haplotype analysis for the association of SNPs (rs4027405, rs1890908, rs8010699) with p21 expression indicated that patients with haplotype A-G-G had a lower p21 expression in HBV-related HCC. SYNE2, also known as EDMD5, NUANCE and Nesprin-2, is mapped to chromosome 14p23.2 and consists of 116 exons. SYNE2 belongs to the family of giant spectrin-repeat (nesprins), which plays an important role in linking the nucleus to the cytoskeleton and is a key component of the linker of the nucleoskeleton and cytoskeleton (LINC) complex 47,48 . Study has reported that SYNE2 depletion may reduce the endothelial cell migration and angiogenic loop formation by regulating the architecture between nucleus and cytoplasm 49 . Recently, nesprins were identified as candidate cancer genes by high-throughput genome sequencing [50][51][52] . Strikingly, SYNE2 alterations have also been revealed in many cancers, including breast cancer, head and neck cancer and colorectal cancer 50,53,54 . Meanwhile, our study showed that the mRNA expression of SYNE2 was significantly down-regulated in HCC (Fig. 4A). These findings imply the significant roles of SYNE2 in cancer biology. Currently, how nesprins regulate and control the cancer development and progression is still poorly understood. Derek et al. 55 found that siRNA induced SYNE2 depletion augmented extracellular signal-regulated MAPK1 and 2 (ERK1/2) activity, leading to the increased cell proliferation. p21 has been implicated in cell cycle progression and proliferation via activating ERK1/2 pathway in HCC 56 . Interestingly, the mRNA expression of SYNE2 and p21 is positively correlated in HBV-related HCC (Fig. 4C) and both are involved in the cell cycle progression and proliferation through activating ERK1/2 pathway. On the basis of above findings and our result that SYNE2 variants were associated with p21 expression, we speculate that there is a close relationship between SYNE2 and p21, even SYNE2 is able to regulate p21 expression, but this is needed to be confirmed in future studies.
Kao et al. 28 reported that p21 expression was an independent prognostic factor of favorable survival. Survival analysis in our study showed, although p21 expression was not associated with OS of HBV-related HCC patients, patients positive for p21 expression had higher HR than those negative for p21 expression (HR = 1.13, 95%CI = 0.83-1.54, P = 0.443, Table 3). This was not conflicting with above findings. Remarkably, SNPs (rs4027405 and rs1890908) of SYNE2 gene were not only significantly associated with OS, but with p21 expression in HBV-related HCC. The G allele is a risk allele in rs4027405, and G allele carriers have a higher p21 expression and a worse prognosis. Similar results were observed in rs1890908, largely thanking to the strong linkage disequilibrium between it and rs4027405. Sebastian et al. 57 revealed that SYNE2 expression was associated with shorter disease-free survival of gastrointestinal stromal cancer patients. Hence, rs4027405 and rs1890908 are likely to influence the prognosis through affecting the expression or function of SYNE2 and p21. Moreover, univariate and multivariate Cox regression analyses indicated that rs4027405 and rs1890908 were independent predictors of poor OS of HBV-related HCC patients, and it combined with serum AFP can suggest that patients with AFP > 400 ng/ml and higher p21 expression allele have significantly higher risk for death than those with AFP ≤ 400 ng/ml and lower p21 expression allele.
However, our study has several limitations. This was a single center study, and a small fraction of clinical data was missing. The SNPs were not functionally characterized to reveal their effects on the p21 expression and clinical outcome of HBV-related HCC. Therefore, additional studies with larger sample size are needed to further validate the association of these SNPs with p21 expression in HBV-related HCC and with the clinical outcome of HBV-related HCC patients, and in vitro studies are required to clarify the underlying mechanisms.
In conclusion, our results demonstrate that SNPs of EGFR (rs6950826 and rs2227983) and SYNE2 (rs4027405, rs1890908 and rs8010699) are associated with p21 expression in HBV-related HCC. rs4027405 and rs1890908 are potential biomarkers for clinical outcome of HBV-related HCC Chinese patients after hepatectomy in Guangxi. In GWAS, TP53 expression as a covariate, and hence, our findings indicate that variants of EGFR and SYNE2 genes may play important roles in the regulation of p21 expression and affect the outcome of HBV-related HCC via TP53-indenpdent manner. Immunohistochemistry. Paraffin embedded HCC tissues were used in immunohistochemistry for p21 and TP53 with general two-step method. Briefly, the antigen retrieval was done with EDTA Tris at a high temperature for 2.5 min. Then, sections were treated with 3% hydrogen peroxide for 10 min to inactivate endogenous peroxidase. After washing thrice with PBS, the mouse against human p21 (1:150) (ZSGB-BIO ORIGENE, Beijing, China) and mouse against human TP53 (1:150) (ZSGB-BIO ORIGENE, Beijing, China) antibodies were independently added, followed by incubation overnight at 4 °C. The sections were incubated with secondary antibody (Dako Cytomation, Glostrup, Denmark) at 37 °C for 30 min. Finally, these sections were visualized with DAB (Dako Cytomation) and counterstained with hematoxylin. Positive and negative controls were included in each immunohistochemical reaction. All the sections were reviewed and scored by two pathologists independently who were blind to the clinical characteristics of patients. At least 10 fields were randomly selected at a high-power (x400 magnifications) at the regions distant from necrotic areas, and the proportion of positive cells was calculated with the following formula: (number of positive cells/total number of the cells × 100%). Positive cells had brown granules in the nucleus. According to previous criteria, positive p21 expression was defined as the presence of ≥ 5% positive cancer cells 28,62 , and positive staining of TP53 as ≥ 10% positive cancer cells 28,63,64 (Fig. 7).
Genotyping. All samples were genotyped using the Illumina HumanExome BeadChip-12-1_A, which includes 242,901 markers on the protein-altering variants. These markers include non-synonymous single nucleotide polymorphisms (SNPs), SNPs in splice sites, stop variants, SNPs in promoter regions, SNPs in extended MHC region, GWAS tag markers, HLA tags, etc. Details about SNPs can be found at the exome array design webpage. Samples were processed according to the manufacturer's instructions. Genotype calling was carried out using the Genotyping Module v1.0 in GenomeStudio version 2011.1, and average call rate was 99.84%. A total of 50 samples (> 10%) were randomly selected and sequenced for the candidate loci using the ABI Prism 3100 (Applied Biosystems, Shanghai Sangon Biological Engineering Technology & Services Co, Ltd, Shanghai, China). The candidate loci and primers used for sequencing are shown in Supplementary  Statistical analysis. Quality Control. A quality control (QC) procedure was used before association analysis as follows: 1. The sample was removed if they (i) had an overall genotyping rate of < 95%; (ii) had ambiguous gender; (iii) had genome-wide identity-by-descent (IBD) > 0.1875; (iv) were outliers in principal components analysis (PCA) for ancestry and population stratification. 2. The SNP was excluded if they had (i) a call rate of < 95%; (ii) a P value in Hardy-Weinberg equilibrium (HWE) < 1 × 10 −6 ; (iii) a Minor allele frequencies (MAF) < 0.05. 3. PCA: Guangxi in Southern China is a multi-ethnic region. An analysis of population stratification by PCA 65 was performed to eliminate the multi-ethnic interference using the EIGENSOFT package. Genomic inflation factors (GIF) 66 was used to investigate the residual population stratification, which was calculated with MATLAB 7.0 (see URLs). The procedure was performed with the Plink version 1.07, R 3.0.1 and EIGENSOFT package 67 .
Scientific RepoRts | 6:31237 | DOI: 10.1038/srep31237 Association Analysis. The association analysis was evaluated using Single Variant Test 68 (chose Logistic Score Test ) with EPACTS package version 3.2.6 (see URLs), after adjustment for age, gender, BCLC stage, TP53 expression status, TACE status before hepatectomy, pathological grade and hepatic cirrhosis. The Chi-squared test and logistic regression model were independently used to analyze the association of genetic model of SYNE2 and EGFR with p21 expression. Local linkage disequilibrium (LD) and recombination patterns nearby SYNE2 and EGFR were analyzed using the LocusZoom 69 (see URLs).
Survival Analysis. Overall survival (OS) was defined as the time from hepatetomy to death or the last follow-up. MST was calculated for all variables, univariate and multivariate survival analysis were done using the Cox proportional hazard regression model. Hazard ratios (HR) and their 95% confidence interval (CI) were calculated from the Cox proportional hazards regression model with adjustment for age, gender, race, body mass index (BMI), smoking status, drinking status, TACE status, serum AFP level, BCLC stage, child-pugh class, radical resection, pathological grade, hepatic cirrhosis and PVTT. A value of P < 0.05 was considered statistically significant. Statistical analysis was performed with SPSS version 18.0 (SPSS, Inc., Chicago, IL, US).