HNRNPA2B1, as a m6A Reader, Promotes Tumorigenesis and Metastasis of Oral Squamous Cell Carcinoma

Abstract

N6-methyladenosine (m⁶A) modification is the most prevalent modification on eukaryotic RNA, and the m⁶A modification regulators were involved in the progression of various cancers. However, the functions of m⁶A regulators in oral squamous cell carcinoma (OSCC) remain poorly understood. In this study, we demonstrated that 13 of 19 m⁶A-related genes in OSCC tissues are dysregulated, and HNRNPA2B1 was the most prognostically important locus of the 19 m⁶A regulatory genes in OSCC. Moreover, HNRNPA2B1 expression is elevated in OSCC, and a high level of HNRNPA2B1 is significantly associated with poor overall survival in OSCC patients. Functional studies, combined with further analysis of the correlation between the expression of HNRNPA2B1 and the EMT-related markers from the TCGA database, reveal that silencing HNRNPA2B1 suppresses the proliferation, migration, and invasion of OSCC via EMT. Collectively, our work shows that HNRNPA2B1 may have the potential to promote carcinogenesis of OSCC by targeting EMT via the LINE-1/TGF-β1/Smad2/Slug signaling pathway and provide insight into the critical roles of HNRNPA2B1 in OSCC.

Keywords: EMT; HNRNPA2B1; Metastasis; N6-methyladenosine; Oral squamous cell carcinoma.

Figure 1

The expression of 19 m⁶A-related genes in OSCC. (A) Boxplot visualizing the expression of m⁶A-related genes in 330 OSCC samples and 32 normal samples. (B) m⁶A level was detected in six pairs of OSCC tissues and adjacent normal tissues (p < 0.05). (C) The expression levels of 19 m⁶A-related genes in OSCC were displayed via a heatmap. The ascending normalized expression level in the heatmaps is colored from blue to red. *, **, *** and **** indicated p<0.05, p<0.01, p<0.001 and p<0.0001, respectively. “ns” indicates no significance.

Figure 2

Cluster analysis based on m⁶A-related genes. (A) Consensus matrix for the two groups from the TCGA database. (B) Principal component analysis (PCA) score plot of the two clusters has few overlaps. (C) Overall survival analysis between cluster 1 and cluster 2. (D) Heatmap of the 19 m⁶A-related genes, showing distinct expression profiles for correlation of cluster analysis and clinical characteristics. * and ** indicated p<0.05 and p<0.01, respectively.

Figure 3

Construction of the Cox regression model and validation of the prognostic signature. (A, B) Univariate and multivariate cox regression based on 19 m⁶A-related genes. (C) LASSO coefficient profiles of 19 m⁶A-related genes associated with the overall survival of OSCC; each curve represents a gene. (D) Partial likelihood deviance is shown against log (lambda). The vertical dotted line indicated the lambda value with the minimum error and the largest lambda value, where the deviance is within one SE of the minimum. (E) The prognostic signature model showed good prediction efficiency with the area under the ROC curve (AUC = 0.632). (F) Survival analysis based on risk score. (G, H) Univariate cox regression and multivariate cox regression according to risk score and clinical characteristics. (I) The heatmap of risk score level and clinical characteristics. * and ** indicated p<0.05 and p<0.01, respectively.

Figure 4

The correlation between HNRNPA2B1 and clinicopathologic parameters of OSCC. (A–C) Rank chart for the 19 m⁶A regulatory genes by neural network, random forest (rf), and gradient boosting machine (gbm). (D) HNRNPA2B1 expression is higher in tumor tissues compared with normal tissues. High HNRNPA2B1 expression was significantly associated with (E) tumor stage, (F) T categories, and (G) metastasis. (H) HNRNPA2B1 expression was not associated with histological grade (p > 0.05). (I) The mRNA expression of HNRNPA2B1 in patients with or without lymph node metastasis was detected by RT-PCR. (J) Representative images of immunohistochemical staining and (K) quantification of HNRNPA2B1 in human normal mucosa, OSCC tissue, and metastasis LN. Metastasis means at least one lymph node metastasis, local or distant metastasis. (L) LINE-1 expression is positively correlated with HNRNPA2B1 expression according to the analysis results of the IHC score. * and **** indicated p<0.05 and p<0.0001, respectively. “ns” indicates no significance.

Figure 5

HNRNPA2B1 is associated with prognosis in patients with OSCC. Kaplan–Meier survival of (A) 330 cases from the TCGA database and (B) 38 patients from our department.

Figure 6

The expression of HNRNPA2B1 significantly affected the proliferation, migration, and invasion in OSCC cell lines. (A) Western blot was used to detect the protein expression after stable knockdown or overexpress HNRNPA2B1 in CAL27 and SCC4 cell lines. (B, C) Knockdown of HNRNPA2B1 inhibited cell proliferation as indicated by CCK-8 assay in CAL27 and SCC4 cells. On the contrary, overexpression of HNRNPA2B1 could enhance its proliferation. (D, E) Wound healing assay showed that downregulation of HNRNPA2B1 inhibits cell migration while HNRNPA2B1 overexpression could increase cell migration in CAL27 and SCC4. (F) Trans-well assay showed that the invasion abilities of CAL27 and SCC4 cells were impaired after knocking down HNRNPA2B1 and were increased after overexpression of HNRNPA2B1 compared with those of the control group. The left panel shows the representative images. The right panel shows the statistic data of the left panel. The data are presented as the mean of three independent experiments. *** and **** indicated p<0.001 and p<0.0001, respectively.

Figure 7

HNRNPA2B1 is correlated with the EMT process in OSCC. (A) EMT scores in normal and OSCC tissues. (B, C) Gene set enrichment analyses show the upregulated genes for HNRNA2B1 high-expression patients compared to low-expression patients. (D) GSE data (GSE138206) and TCGA data show that HNRNPA2B1 is correlated with multiple EMT markers. (E–H) HNRNPA2B1 expression is negatively correlated with E-cadherin (R = -0.35, p < 0.05) expression and positively correlated with N-cadherin (R = 0.53, p < 0.05) and vimentin (R = 0.56, p < 0.05), but was not correlated with the LINE-1 mRNA expression (R = 0.08, p = 0.15). (I) LINE-1 mRNA expression was not correlated with the HNRNPA2B1 expression in OSCC cell lines. (J) Western blot analysis of EMT markers related with LINE-1/TGF-β1/Snail signaling pathway. Protein levels of N-cadherin, E-cadherin, TGF-β1, Snail, and LINE-1 ORF1 of CAL27 and SCC4 cells were detected by Western blot after HNRNPA2B1 was knocked down or overexpressed. (K) Relative enrichment of LINE-1 mRNA associated with HNRNPA2B1 protein was identified by RIP assays using anti-IgG and anti-HNRNPA2B1 antibodies. The IgG group was a negative control to preclude nonspecific binding. The Y-axis represents the percent of input for each IP sample according to the formula: %Input =1/10*2^{Ct [IP]-Ct [input]}. *, **, *** and **** indicated p<0.05, p<0.01, p<0.001 and p<0.0001, respectively. “ns” indicates no significance.