A Genomic Instability-Related Long Noncoding RNA Signature for Predicting Hepatocellular Carcinoma Prognosis

Abstract

Background: Long noncoding RNAs (lncRNAs) are found to be novel biomarkers for hepatocellular carcinoma (HCC) and play an important role in tumor progression. We established a genomic instability-related long noncoding RNA signature (GIlncSig) as an independent prognosis factor and also investigated its impact on prognosis significance.

Method: Somatic mutation profiles, clinical characteristics, and RNA sequencing data were obtained from The Cancer Genome Atlas (TCGA) database. Lasso regression was used to construct GIlncSig. KEGG was used to identify the possible biological pathways. ESTIMATE and CIBERSORT algorithms were used to calculate the immune microenvironment scores and proportion of immune cells in HCC patients. The expression of LINC00501 was conducted by qRT-PCR. Cell proliferation was measured by EdU, CCK-8, and colony formation assay, and cell migration and invasion ability were measured by wound healing and transwell assay.

Results: 135 genomic instability-related lncRNAs were identified, and GIlncSig was constructed using 13 independent lncRNAs with significant prognosis values. Based on the GIlncSig, high-risk group had worse clinical outcomes than low-risk group, while high-risk group also had higher UBQLN4, KRAS, ARID1A, and PIK3CA expression. Moreover, the efficiency of GIlncSig combining single-gene mutation was higher than single-gene mutation alone such as TP53. The results of CIBERSORT and ESTIMATE showed that GS group and GU group had significantly different immune infiltration. In addition, LINC00501 was identified as a potential biomarker in HCC with strong relationship with clinical characteristics. In vitro assays validated that LINC00501 promoted proliferation and migration of HCC cell lines.

Conclusion: Our results showed that GIlncSig serves as a potential independent prognosis factor to predict HCC patients' prognosis for exploring potential mechanism and therapy strategy. Besides, LINC00501 plays an important role in the progression of HCC, which may be a potential therapy target.

Figure 1

Identification of GI-related lncRNAs in HCC. (a) Volcano plot of differently expressed lncRNAs. The right orange labeled lncRNAs were significantly higher expressed in HM group, while the left blue labeled lncRNAs were significantly low expressed in HM group. (b) Heatmap of 353 HCC samples according to the expression of 135 GI-related lncRNAs. The right cluster was GS group and the left cluster was GU group. (c) Somatic mutation counts were significantly higher expressed in GU group, while the expression of UBQLN4, KRAS, PIK3CA, and ARID1A was significantly higher expressed in GS group. (d) Co-expression network of mRNAs and lncRNAs. The orange circles represent lncRNAs, and blue circles represent mRNAs. (e) KEGG analysis of lncRNAs co-expressed mRNAs.

Figure 2

Establishment of the GI-related lncRNAs signature. (a) Univariate cox regression analysis of 20 genomic instability-related lncRNAs. (b) Lasso regression model for 20 lncRNAs. (c) heatmap of the GIlncSig. The right cluster was low-risk group, and the left cluster was high-risk group. (d) Distribution of somatic mutation. (e) Distribution of UBQLN4, KRAS, PIK3CA, and ARID1A.

Figure 3

Evaluation of the GIlncSig efficacy. (a) K-M analysis showed that low-risk group had higher overall survival than high-risk group. (b) High-risk group has higher cumulative somatic mutation counts than low-risk group. ((c)-(f)) The expression of UBQLN4, KRAS, ARID1A, and PIK3CA was higher in high-risk group. (g) 1-, 3-, and 5-year ROC curve analysis of the GIlncSig.

Figure 4

Univariate and multivariate cox regression of clinical factors and risk score.

Figure 5

Evaluation of GIlncSig efficacy in clinical factors. (a) K-M analysis of os of patients with high- or low-risk scores with age < 65 or age ≥ 65. (b) K-M analysis of overall survival of patients with high- or low-risk scores with G1-2 or G3-4. (c) K-M analysis of os of patients with high- or low-risk scores with stage I-II or stage III-IV. (d) K-M analysis of os of patients with high- or low-risk scores with T1-2 or T3-4. (e) K-M analysis of os of patients with high- or low-risk scores with N0 or N1. (f) K-M analysis of os of patients with high- or low-risk scores with M0 or M1.

Figure 6

Overall survival of lncRNAs in GIlncSig. K-M analysis showed that 13 lncRNAs were correlated with os of HCC patients.

Figure 7

Clinical characteristics of lncRNAs in GIlncSig.

Figure 8

Comparison of GIlncSig with single-gene mutation. (a) Waterfall plot of the 20 most frequently mutated genes. (b) The proportion of top 6 genes mutation between high-risk and low-risk group. (c) K-M analysis of overall survival with different combinations of TP53 and GIlncSig.

Figure 9

Identification of immune microenvironment and genomic instability of HCC. (a) The expression of immune scores, ESTIMATE scores, and stromal scores in GS group was higher than GU group. (b) The proportion of 22 types of immune cells infiltrating in HCC samples between GU group and GS group. (c) The expression of immune checkpoints between GS group and GU group.

Figure 10

(a) Nomogram combing clinical factors and risk score. (b) Calibration curves illustrated the consistency between predicted and observed 3-year and 5-year survival rates depending on the prognostic nomogram.

Figure 11

The expression and function of LINC00501 in HCC. (a) LINC00501 was highly expressed in HCC tissues compared with the adjacent normal tissues. (b) The efficiency of siRNA targeting LINC00501. ((c), (d), (e)) CCK-8, EdU, and colony formation assay indicated that LINC00501 knockdown inhibited proliferation of Huh-7. ((f), (g)) Wound healing and transwell assay indicated that LINC00501 knockdown inhibited migration of Huh-7. All experiments were replicated three times.