Circular RNAs hsa_circ_0001438 and hsa_circ_0000417 are downregulated and upregulated, respectively, in hepatocellular carcinoma

Abstract

Hepatocellular carcinoma (HCC) is the most predominant type of liver cancer and is frequently fatal. Alpha‐fetoprotein, alpha‐fetoprotein‐L3, and protein induced by vitamin K absence or antagonist‐II are used as biomarkers to diagnose HCC. However, these biomarkers are not highly specific, especially for early‐stage HCC diagnosis; therefore, more specific biomarkers are needed. Recently, circular RNA (circRNA) biomarkers have been used to diagnose several intractable diseases. In this study, we sought to identify circRNA biomarkers for the specific diagnosis of HCC. To this end, we compared the expression levels of circRNAs in primary HCC and normal tissues using publicly available RNA‐seq data. Our analysis revealed that the expression levels of eight circRNAs were altered in primary HCC tissues compared with normal tissues. To confirm our findings, we examined the expression levels of selected circRNAs in HCC cell lines and normal hepatocytes. The expression level of hsa_circ_0001438, a circRNA that was downregulated in primary HCC, was lower in poorly and well‐differentiated HCC cell lines than in normal hepatocytes. By contrast, the expression level of hsa_circ_0000417, which was increased in primary HCC, was strongly upregulated in a well‐differentiated HCC cell line compared with normal hepatocytes. Thus, hsa_circ_0001438 and hsa_circ_0000417 might be potential biomarkers for the specific diagnosis of HCC. The experimental strategy described here, using publicly available RNA‐seq data, is a useful and cost‐effective method of identifying circRNA biomarkers.

1. INTRODUCTION

GLOBOCAN, a database provided by the International Agency for Research on Cancer, ¹ recorded 18.1 million new cancer patients and 9.6 million new cancer deaths worldwide in 2018. Among these were 0.84 million new liver cancer patients and 0.78 million new deaths due to liver cancer. An analysis of 38 specific cancers detailed in GLOBOCAN 2018 revealed that liver cancer is the third most fatal cancer, ¹ and hepatocellular carcinoma (HCC) is the main type of liver cancer. ² Viral HCC is caused by chronic infection with the hepatitis B or hepatitis C virus ² and was predominant in Western countries and Japan until the late 1990s. The number of nonviral HCC cases caused by non‐alcoholic fatty liver disease or non‐alcoholic steatohepatitis has been increasing since the early 2000s, ³ , ⁴ although viral HCC is still the predominant form in developing countries.

Traditionally, clinicians use alpha‐fetoprotein (AFP) or cancer antigen 19‐9 (CA19‐9) as biomarkers to diagnose liver cancer, ⁵ although AFP‐L3 and protein induced by vitamin K absence or antagonist‐II (PIVKA‐II) have also been used more recently. However, these biomarkers are not sensitive enough to detect early‐stage HCC. ⁵ In view of advances in diagnostic imaging sensitivity, the American Association for the Study of Liver Diseases stopped recommending the use of AFP as a liver cancer biomarker in 2018. ⁶ Thus, more sensitive biomarkers are needed to diagnose liver cancer, especially early‐stage HCC.

Circular RNAs (circRNAs) are noncoding RNAs produced by back‐splicing of precursor mRNAs (pre‐mRNAs). Specifically, the 5′ and 3′ ends of back‐spliced pre‐mRNA segments covalently connect to form circRNAs, ⁷ which maintain cellular homeostasis and can either promote or inhibit tumorigenesis. ⁸ CircRNAs are also used as cancer biomarkers. For example, circ_ZKSCAN1 is downregulated in liver cancer tissues and is a potential HCC biomarker. ⁹ CircRNAs can be detected by liquid biopsy because they are secreted into blood, saliva and urine. ¹⁰ In this study, we sought to discover new circRNA biomarkers for the diagnosis of HCC and identified hsa_circ_0001438 and hsa_circ_0000417 as potential candidates.

2. MATERIALS AND METHODS

2.1. Screening of circRNAs

Fragments per million mapped fragments (FPM) data from GSE77509, ¹¹ a GEO RNA‐seq dataset of matched adjacent normal, primary tumour and portal vein tumour thrombosis samples from HCC patients (deposited at NCBI: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE77509), were downloaded from CIRCpedia v2 (http://yang‐laboratory.com/circpedia/). The FPM data were analysed to identify circRNAs with expression levels that differed between HCC and normal tissues. Details of the analysis are shown in Figure S1. The identification numbers of the circRNAs were confirmed using circBase (http://www.circbase.org/). Statistical analyses of circRNA expression levels in normal and HCC tissues were performed via a Wilcoxon signed‐rank test in R version 4.0.3 (https://www.r‐project.org/), and p < .05 was considered statistically significant.

2.2. Heatmaps

Heatmaps were prepared using R version 4.0.3 (https://www.r‐project.org/). Briefly, the heatmap.2 function from the gplots package was used, and clustering was conducted using the average mode.

2.3. Receiver operating characteristic (ROC) curves

ROC curves were prepared using R version 4.0.3 (https://www.r‐project.org/) with the default parameter settings.

2.4. Cell lines

The HepG2 hepatoblastoma cell line ¹² was obtained from the RIKEN BRC through the National Bio‐Resource Project of the MEXT/AMED, Japan. The HCC cell lines JHH‐4 and JHH‐5 ¹³ , ¹⁴ , ¹⁵ were obtained from the JCRB Cell Bank. HepG2 cells were cultured in Dulbecco's modified Eagle's medium (Wako) containing 10% (v/v) foetal bovine serum (FBS) (Nichirei Biosciences, Inc.) and penicillin–streptomycin (Sigma‐Aldrich). JHH‐4 cells were cultured in Eagle's minimal essential medium (Wako) containing 10% (v/v) FBS and penicillin–streptomycin. JHH‐5 cells were cultured in Williams' Medium E medium (Thermo Fisher Scientific) containing 10% (v/v) heat‐inactivated FBS and penicillin–streptomycin. All cell lines were cultured in a humidified incubator with 5% CO₂ at 37°C. Human hepatocytes (pooled from ten mixed‐gender donors) were purchased from Lonza (HUCS10P), and total RNA was extracted without culturing the cells.

2.5. Preparation of RNA

Each HCC cell line was cultured independently in three cell culture dishes and collected in three tubes to represent three biological replicates. Total RNA was extracted using the RNeasy Mini Kit (QIAGEN). A 5 μg sample of total RNA was incubated with or without RNase R (Lucigen) for 30 min at 37°C and then purified using the RNeasy Mini Kit (30 μl eluate).

2.6. Quantitative reverse transcription PCR (RT‐qPCR)

Purified RNA (6 μl) was reverse transcribed using ReverTra Ace qPCR RT Master Mix with gDNA Remover (Toyobo). Template cDNA was amplified using GoTaq qPCR Master Mix (Promega) and the following thermal cycling conditions: one cycle of 95°C for 2 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. To calculate the number of target circRNA‐derived cDNAs, DNA fragments corresponding to target cDNAs were used as references. If necessary, amplicons were analysed by sequencing (Eurofins). The primer sequences used in these experiments are shown in Table S1.

3. RESULTS

3.1. Screening for circRNAs as potential biomarkers of liver cancer

To identify new biomarkers of liver cancer, we compared the expression levels of circRNAs in primary HCC and adjacent normal tissues by analysing publicly available RNA‐seq data from 16 HCC patients (GSE77509; clinical features were summarized in Table S2). A total of 271 circRNAs were commonly expressed in these patients and were included in our analysis. On the basis of the circRNA expression patterns, we classified the normal and HCC samples into three groups: Group I, normal tissue samples only; Group III, HCC samples only; and Group II: mainly HCC samples but also some normal tissue samples (N3, N6, N8 and N10) (Figure S2 and Table S3). Eight circRNAs were significantly (p < .05) differentially expressed by at least threefold in HCC versus normal tissues (Figure 1). Three and five of these circRNAs were upregulated and downregulated in tumour tissues respectively.

FIGURE 1.

Heatmap of circRNA expression in HCC and adjacent normal tissues. The expression levels of eight selected circRNAs commonly expressed in 16 HCC samples were determined by analysis of RNA‐seq data. The chromosomal location and circBase identification number of each circRNA gene are shown. Circ_ZKSCAN1 is also known as hsa_circ_0001727

Our subsequent analyses focused on circ_ZKSCAN1, hsa_circ_0001438 and hsa_circ_0000417. Circ_ZKSCAN1 has been reported as a potential biomarker of liver cancer ⁹ and was downregulated in the HCC tissues. Hsa_circ_0001438, which was also downregulated in the HCC tissues, is derived from the La‐related protein 1 B (LARP1B) gene, a paralog of the LARP1A gene, both of which encode proteins carrying a conserved 90 amino acid signature La motif. Although the potential involvement of LARP1B in various cancers remains unclear, LARP1A is an oncogene that is often upregulated in HCC and cervical, liver, breast and prostate cancers. ¹⁶ , ¹⁷ Hsa_circ_0000417, which was upregulated in HCC tissues, is derived from the cleavage and polyadenylation specific factor 6 (CPSF6) gene, the protein product of which plays a critical role in the progression of HCC via alternative polyadenylation, ¹⁸ and promotes glycolysis to suppress apoptosis of HCC cells. ¹⁹

To evaluate the specificities and sensitivities of hsa_circ_0001438, circ_ZKSCAN1 and hsa_circ_0000417 as potential biomarkers of HCC, we prepared ROC curves (Figure 2). The area under the curve (AUC) values for these three circRNAs were 0.941, 0.965 and 0.738 respectively. Because diagnostic tests with AUC > 0.9 are considered highly accurate, and those with 0.7 < AUC <0.9 are considered moderately accurate, ²⁰ these findings suggest that, in addition to the previously reported circ_ZKSCAN1, ⁹ hsa_circ_0001438 and hsa_circ_0000417 might also be effective diagnostic biomarkers of HCC. Overall, the downregulated circRNAs were more sensitive and specific HCC biomarkers (AUC ≥ 0.9) than the upregulated circRNAs (AUC < 0.9) (Figures 2 and S3).

FIGURE 2.

Receiver operating characteristic (ROC) curves for HCC circRNA biomarkers. ROC curves were prepared for hsa_circ_0001438 (A), circ_ZKSCAN1 (B) and hsa_circ_0000417 (C). Cut‐off values are shown under each graph

3.2. Expression of hsa_circ_0001438 and hsa_circ_0000417 in liver cancer cells

To verify their potential use as biomarkers of HCC, we examined the expression levels of hsa_circ_0001438 and hsa_circ_0000417 in liver cancer cells via RT‐PCR using divergent primers. Total RNA was extracted from hepatoblastoma HepG2 cells ¹² , ²¹ and treated with or without RNase R prior to RT‐PCR amplification. Hsa_circ_0001438 was stably amplified using the divergent primers (Figure S4A,B), and specific amplification of the intended target was confirmed by sequencing the amplicon (Figure S4B–D). Specific amplification of hsa_circ_0000417 using divergent primers was also confirmed by sequencing (Figure S5). Although we also designed some divergent primers to detect other circRNAs, only those designed against circ_ZKSCAN1 successfully amplified the intended target (Figure S4A), as reported previously. ⁹

Next, we used RT‐qPCR to compare the expression level of hsa_circ_0001438 in commercially available normal hepatocytes (mixed hepatocytes from ten donors) with those in the nonviral HCC cell lines JHH‐5 (Edmondson–Steiner grade I) and JHH‐4 (grade III). ¹³ , ¹⁴ , ¹⁵ Giemsa staining revealed that the normal hepatocytes had a typical phenotype (e.g. well‐stained, uniformly sized, round, with one or two round nuclei; Figure S6), although the average body mass index of the donors (27.2) was slightly high (Table S4). ²² Therefore, we concluded that these control cells were nonpathological and did not comprise many fatty hepatocytes. An absolute quantification analysis (without normalization) showed that the expression level of hsa_circ_0001438 was lower in JHH‐4 and JHH‐5 cells than in normal hepatocytes (Figure S7A). This trend was retained after normalization of hsa_circ_0001438 expression to that of the glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) mRNA ⁹ as an internal control (relative quantification) (Figures 3A and S8). Consistent with a previous report, ⁹ similar results were obtained for circ_ZKSCAN1 (Figures S7B and 3B). On the contrary, compared with that in the normal hepatocytes, the absolute and relative expression levels of hsa_circ_0000417 were strongly and slightly upregulated in JHH‐5 and JHH‐4 cells, respectively (Figures S7C and 3C).

FIGURE 3.

Expression levels of hsa_circ_0001438, circ_ZKSCAN1 and hsa_circ_0000417 in HCC cell lines. The normalized expression levels of hsa_circ_0001438 (A), circ_ZKSCAN1 (B) and hsa_circ_0000417 (C) in normal hepatocytes (NH) and two HCC cell lines (JHH‐4 and JHH‐5). For each cell line, three RNA samples were prepared independently as biological replicates (see Section 2). The glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) mRNA was used for normalization

4. DISCUSSION

In this study, we found that hsa_circ_0001438 and hsa_circ_0000417 are downregulated and upregulated, respectively, in HCC and may be useful biomarkers for the specific diagnosis of HCC. JHH‐4 and JHH‐5 are both nonviral HCC cell lines, and JHH‐5 is an early‐stage, Edmondson–Steiner grade I HCC cell line; therefore, as hsa_circ_0001438 was downregulated in both of these cell lines, it may be a useful biomarker to diagnose early‐stage or nonviral HCC. A ROC analysis revealed that the AUC value for hsa_circ_0001438 (0.941; Figure 2A) was comparable to those of hsa_circ_0005075 (0.94) ²³ and cSMARCA5 [hsa_circ_0001445] (0.938), ²⁴ suggesting that it may perform as well as these previously reported biomarkers. In fact, hsa_circ_0001438 may even be diagnostically superior because its expression level was downregulated by threefold in HCC, whereas that of cSMARCA5 was only downregulated by twofold (Table S3), and because hsa_circ_0005075 expression was undetectable in our dataset (Table S3), contrary to a previous report. ²³

Hsa_circ_0000417 was upregulated strongly in the Edmondson–Steiner grade I HCC cell line, JHH‐5, but just slightly in the Edmondson–Steiner grade III HCC cell line, JHH‐4 (Figure 3B). These results suggest that hsa_circ_0000417 is expressed at an early stage of carcinogenesis and, in combination with hsa_circ_0001438, might be useful to distinguish between undifferentiated and highly differentiated HCC.

Absolute quantification of hsa_circ_0001438 and hsa_circ_0000417 (i.e. quantitative comparison without normalization) requires only amplification primers for each target. By contrast, relative quantification requires additional primers to amplify an internal control (such as the GAPDH mRNA). Relative quantification of circRNA biomarkers may diagnose HCC more accurately than absolute quantification because normalization addresses sample‐to‐sample variation due to differences in RNA quality or technical error. Absolute and relative quantification of has_circ_0001438 and hsa_circ_0000417 demonstrated that these circRNAs were downregulated and upregulated, respectively, in the HCC cell lines (Figures 3 and S7 and S8).

The expression patterns of circRNAs in some normal tissue samples (N3, N6, N8 and N10) were similar to those in some HCC samples; these normal and similar HCC samples were classified together (Group II; Figures 1 and S2). The normal tissue samples in Group II might have contained some malignant or premalignant cells. We expect that the detection of abnormal cells in regions adjacent to HCCs may help to develop treatment strategies or predict patient outcome. Therefore, it may be useful to measure the expression levels of hsa_circ_0001438 and hsa_circ_0000417 in normal tissues taken from regions adjacent to HCCs during surgical treatment of liver cancer. We expect that these normal tissues collected during biopsy do in fact contain abnormal cells. Further experiments are required to measure hsa_circ_0001438 and hsa_circ_0000417 levels in normal tissue samples and liquid biopsies.

Overall, this study identified two new potentially useful biomarkers for HCC diagnosis. Our experimental strategy using a public database was simpler and more cost‐effective than performing RNA‐seq and a bioinformatical analysis of the data. The strategy described here is useful and reliable, as demonstrated by the fact that we identified circRNAs that have been reported as HCC biomarkers previously (such as circZKSCAN). We expect that our strategy could be used to inform and aid efforts to identify circRNA biomarkers for other diseases. Nonetheless, our study had some limitations. First, we used a restricted set of normal hepatocytes and HCC cell lines to confirm the expression of circRNAs in HCC. Second, we were unable to perform some statistical analyses because the commercially obtained normal cells were a mixture of hepatocytes from ten different donors. Thus, further studies using unmixed normal hepatocytes (from distinct, healthy donors) would be useful to evaluate the expression of the identified circRNAs more accurately.

AUTHOR CONTRIBUTIONS

S.I. conceived the project, designed and performed the experiments, analysed data and wrote the manuscript. S.N. performed the experiments. T.F. and H.F. supervised the project, designed the experiments and wrote the manuscript. All authors approved the final manuscript.

Funding information

This work was supported by Grant‐in‐Aid for Encouragement of Scientists from JSPS KAKENHI (#21H04261 to S.N.).

CONFLICT OF INTEREST

The authors declare no competing interests.

Supporting information

Table S1. Primers used in this study.

Table S2. Clinical features of the 16 HCC patients.

Table S3. FPM data from GSE77509.

Table S4. Characteristics of the HUCS10P donors.

Figure S1. Experimental scheme to select circRNAs.

Figure S2. Heatmap of circRNA expression in HCC and adjacent normal tissues.

Figure S3. ROC curves for differentially expressed circRNAs.

Figure S4. Amplification of hsa_circ_0001438 using divergent primers.

Figure S5. Amplification of hsa_circ_0000417 using divergent primers.

Figure S6. Giemsa staining of HUCS10P.

Figure S7. Expression levels of hsa_circ_0001438, circ_ZKSCAN1 and hsa_circ_0000417 in HCC cell lines.

Figure S8. Experimental scheme.

Imanishi S, Nagata S, Fujita T, Fujii H. Circular RNAshsa_circ_0001438 and hsa_circ_0000417 are downregulated and upregulated, respectively, in hepatocellular carcinoma. Int J Exp Path. 2022;103:245‐251. doi: 10.1111/iep.12457