ROS-Related miRNAs Regulate Immune Response and Chemoradiotherapy Sensitivity in Hepatocellular Carcinoma by Comprehensive Analysis and Experiment

Reactive oxygen species (ROS) plays an essential role in the development of cancer. Here, we chose ROS-related miRNAs for consensus clustering analysis and ROS score construction. We find that ROS is extremely associated with prognosis, tumor immune microenvironment (TIME), gene mutations, N6-methyladenosine (m6A) methylation, and chemotherapy sensitivity in hepatocellular carcinoma (HCC). Mechanistically, ROS may affect the prognosis of HCC patients in numerous ways. Moreover, miR-210-3p and miR-106a-5p significantly increased the ROS level and stagnated cell cycle at G2/M in HCC; the results were more obvious in cells after ionizing radiation (IR). Finally, the two miRNAs suppressed cell proliferation, migration, and invasion and promoted apoptosis in huh7 and smmc7721 cells. It indicated that ROS might affect the prognosis of HCC patients through immune response and increase the sensitivity of HCC patients to radiotherapy and chemotherapy.


Introduction
Hepatocellular carcinoma (HCC) is the sixth most frequently diagnosed cancer and the third leading cause of cancer death in the world in 2020 [1,2]. The incidence of HCC is rising faster than other tumors, and the incidence of HCC in China is more than 10 times that of Europe and the United States, and its mortality rate ranks fifth among cancers [3]. There are many factors contributing to the high incidence of liver cancer, such as alcohol abuse, smoking, metabolic syndrome, and hepatitis virus. Especially in China, hepatitis B virus is the major risk factor for HCC [4]. Due to the insidiousness of the onset of HCC, patients are often diagnosed at an advanced stage. Although immunotherapy has developed rapidly in the past few years, improving the survival of patients with HCC, only a few patients with HCC could benefit from this treatment [5]. Therefore, novel molecular markers with more clinical util-ity are needed to improve diagnostic and prognostic prediction and guide the clinical treatment of HCC patients.
miRNAs are a group of approximately 21-25 nucleotides length and small endogenous single-stranded noncoding RNAs, which can negatively regulate gene expression via binding to the 3 ′ -untranslated region of messenger RNA (mRNA) [6]. miRNA profiling can be used to predict cancer diagnosis and prognosis, for the pattern of miRNA expression can be correlated with cancer type, stage, and other clinical variables [7].
Recently, reactive oxygen species (ROS) have been shown to promote metastasis in a variety of cancers and play important roles in tissue homeostasis, cellular signaling, differentiation, and survival [8][9][10][11][12][13]. ROS are pleiotropic molecules or free radicals produced by numerous complex mechanisms. Its excessive production, failure of clearance mechanisms, and even insufficient antioxidants may lead to the accumulation of ROS, ultimately leading to oxidative stress [14]. Moreover, increasing evidence has suggested that numerous miRNAs have been linked to processes associated with ROS. Wang and colleagues find that significant increases in miR-34a-5p and miR-495-3p are consistent with ROS levels, but mir-34a-5p inhibition is reduced only in intestinal injury [15]. In addition, the research indicates that miR-15b inhibits ROS production by regulating SIRT4 [16]. However, studies of ROS-related miRNAs mainly focus on cardiac diseases, and the mechanism of hypoxia-induced miRNAs in tumors remains not clear.
In this research, we screened 36 ROS-related miRNAs from three reviews [14,17,18]. Then, 9 ROS-related miR-NAs were selected and used for subsequent analysis. Through comprehensive analysis, ROS score was significantly corrected with prognosis, tumor immune microenvironment (TIME), gene mutations, m6A methylation, and chemotherapy sensitivity in HCC. In the experiments, miR-210-3p and miR-106a-5p extremely increased the ROS level and stagnated cell cycle at G2/M. Collectively, we confirmed the strong correlation between ROS and HCC through bioinformatics and experimental methods.

Materials and Methods
2.1. Dataset Source. The clinical data of HCC patients were downloaded from the University of California Santa Cruz (UCSC, https://xenabrowser.net/datapages/). The research included 374 tumor and 50 normal samples. The miRNA expression data of HCC patients were downloaded from the The Cancer Genome Atlas (TCGA) data portal by the "TCGAbiolinks" R package [19]. Furthermore, the mutation data of the TCGA-LIHC was downloaded from the websites (https://portal.gdc.cancer.gov/).

Identification of Consensus
Clustering and Prognosis for ROS-Related miRNAs. According to the expression levels of 9 ROS-related miRNAs, HCC patients were clustered into different clusters by using R package "ConsensusCluster-Plus" (http://www.bioconductor.org/).
2.7. qRT-PCR. qRT-PCR was performed with Hairpin-it™ Real-Time PCR miRNAs (GenePharma). The specific primers, miR-106a-5p and miR-210-3p, were purchased from Suzhou GenePharma Co., Ltd. Thermal cycles were as follows: 95°C for 3 min and 40 cycles of 95°C for 15 s, 62°C for 40 s. Melting curve analysis was used to confirm the specificity of amplification. The relative expressions of miRNAs to SNORD44 were determined using the comparative 2 -ΔΔCt method.
2.8. Measurement of ROS Production. Flow cytometry was used to assay ROS levels in Huh7 and Smmc7721 using a ROS assay kit (Shanghai Biyuntian Biological Co., Ltd.). Briefly, the cells were seeded in 6-well plates at a density of 2:5 × 10 5 cells/well for 24 h. Next, the cells were incubated with 10 μM of 2 ′ ,7 ′ -dichlorodihydrofluorescein diacetate (DCFH-DA) for 15 min at 37°C in a dark room. A Beckman cytoFLEX flow cytometer was used to detect ROS levels. The data was analyzed by using the CytExpert 2.3 software.
2.9. Cell Cycle Analysis. Cell cycle analysis was conducted with a cell cycle staining kit (MultiScience Biotech Co., Ltd.). The transfected cells were seeded in 6-well plates at a density of 2:5 × 10 5 cells/well for 72 h. Then, the cells were collected and incubated with 1 mL of DNA Staining solution and 10 μM of permeabilization solution for 30 min in a dark room. A Beckman cytoFLEX flow cytometer was used to detect the cell cycle. The data was analyzed by using the CytExpert 2.3 software (Beckman Coulter, CA, USA).

Transwell Assay.
Transwell assay was conducted to analyze the migrated and invasive abilities of cells. For cell migration, transfected SMMC-7721 and Huh7 cells (1 × 10 4 cells/well) in 100 μL serum-free DMEM medium were placed in the upper chambers (Costar, Corning, NY, USA), and 500 μL DMEM medium containing 10% FBS was added to the lower chambers. Cells remaining on the top surface were removed with a cotton swab after the incubation for 24 h, while migrated cells through the membranes were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet (Sigma, St. Louis, MO, USA). The stained cells were counted from five random fields under a microscope. For cell invasion, the experiment was performed following the same approach using transwell chambers pretreated with Matrigel (BD, San Jose, CA, USA).           Oxidative Medicine and Cellular Longevity by the CCK8 reagent (Sigma-Aldrich) which was read from culture media at 450 nm (ELx800 Microplate Reader, BioTek Instruments, Winooski, VT, USA) at 0, 24, 48, and 72 h. Cell proliferation was expressed relative to the corresponding control.

Cell Proliferation
2.12. Wound Healing Assay. The transfected cells were seeded in 6-well plates at a density of 2:5 × 10 5 cells/well for 72 h, then draw a wound between the dense cells with a 200 μL gun head. The gap closure was monitored under the microscope and a digital camera (CK30-SLP; Olympus, Tokyo, Japan) at 0, 24, and 48 h. Images were analyzed using the ImageJ version 1.52a software (National Institute of Health, Bethesda, MD, USA).
2.13. Cell Apoptosis Analysis. Apoptosis was assayed using the annexin V-phycoerythrin (PE)/7-aminoactinomycin D (7-AAD) or annexin V-adenomatous polyposis coli (APC)/7-AAD kit (Becton-Dickinson). The transfected cells were rinsed with ice-cold PBS and resuspended in 100 μL of 1 × binding buffer. Then, the liquid was stained with 5 μL 7-AAD and 5 μL annexin V-APC/PE and incubated for 15 min in the dark. Then, another 400 μL binding buffer was added into the mixture before cell apoptosis was detected on a Beckman cytoFLEX flow cytometer. The analysis of the above data was carried out using the CytExpert 2.3 software (Beckman Coulter, CA, USA).

Correction between ROS Score and Clinical Features in HCC.
Firstly, 340 HCC patients were randomly divided into a validation dataset (170 patients) and training dataset (170 patients). Then, 5 candidate HBV-related ROS miRNAs were selected to calculate ROS score by using LASSO regression in the training dataset (Figure 2   Oxidative Medicine and Cellular Longevity   ExpressionÞ. Furthermore, patients were divided into highand low-risk groups, based on the median ROS score. The ROS score of high-risk group was higher, the low-risk group was instead. Interestedly, we found that high-risk groups were significantly associated with bad prognosis (Figure 2(b)). Moreover, the time-dependent receiver oper-ating characteristic (ROC) curve was constructed. The time-dependent ROC curve (AUC) of five candidate HBVrelated ROS miRNAs was 0.76, 0.81, and 0.78 at one year, three years, and five years (Figure 2(b)). With regard to the validation dataset, the AUC of the miRNAs was 0.58, 0.72, and 0.68 at one year, three years, and five years (Figure 2(c)).

ROS Was Associated with TIME in HCC.
Experimental evidences showed that the production of ROS in macrophages could affect both natural and acquired immunity and immune responses [21]. Therefore, the correction between ROS score and TIME in HCC was analyzed. We found that half of the immune cells (11/22) were significantly different in cluster1/ 2, especially "T cell CD4 memory resting," "T cell follicular helper," "Macrophages M0," and "Macrophages M2" (Figure 5(a)). Moreover, 9 of the 22 immune cells were

12
Oxidative Medicine and Cellular Longevity extremely different in the high-and low-risk groups, especially "T cells follicular helper," "Macrophages M0," "Mast cells resting," and "Eosinophils" (Figure 5(b)). We further analyzed the effects of ROS on immune checkpoint inhibitors; six wellknown immune markers were selected for analysis. We found that almost all markers (5/6) were distinct in cluster1/2 and high/low-risk groups (Figures 5(c) and 5(d)). All expression levels of six immune markers were higher in cluster1 and high-risk group, which were associated with worse prognosis in HCC patients. Furthermore, we analyzed the correction between immune cells/markers and ROS score. As showed in Figure 5(e), about one-third of the cells were significantly associated with ROS score. It indicated that ROS could affect the immune response and immunotherapy in HCC.

The Correction between ROS and Gene Muntion in HCC.
ROS contributes to the accumulation of DNA mutations [22]. We further analyzed the effects of ROS on gene mutations in HCC patients. We found that the mutation rate of HCC patients was higher in cluster1 and high-risk group than in cluster2 and low-risk group (Figures 6(a)-6(d)). The highest mutation rate was TP53 in cluster1 and highrisk group (Figures 6(a) and 6(b)). Moreover, the mutated genes were more in cluster1 and high-risk group than in cluster2 and low-risk group (Figures 6(e) and 6(f)). As showed in Figure 5(g), ROS score of HCC patients with TP53+ was higher than patients with TP53- (Figure 6(g)).
The results indicated that ROS could effect gene mutations, especially TP53 in HCC patients.

ROS Regulate m6A Methylation Level of HCC.
Alteration of N6-methyladenosine (m6A) levels participates in cancer pathogenesis and progression [23][24][25]. We further analyzed the correction between ROS and m6A levels. The results showed that most of the m6A RNA methylation regulators (14/19) were more highly expressed in the high-ROS score group than in the low-ROS score group (Figure 6(h)). In addition, ROS-related miRNAs were significantly correlated with m6A RNA methylation regulators (Figure 6(i)). We further found that ROS score was positively correlated with

15
Oxidative Medicine and Cellular Longevity the m6A score ( Figure 6(j)). The results indicated that ROS could promote m6A methylation level of HCC.

ROS Induced Chemotherapy Sensitivity in HCC.
Many chemotherapeutic agents act on cancer through ROS production [26][27][28]. We further explored whether ROS could also affect the chemotherapy sensitivity in HCC. The correction between chemotherapy drug and ROS score was analyzed by using R package "pRRophetic". As showed in Figure 7, "Sorafenib," "Gefitinib," "Rapamycin," and "Lapatinib" were more highly expressed in high-ROS score patients. It showed that ROS could increase the chemotherapy sensitivity in HCC.
3.7. miR-210-3p and miR-106a-5p Were Associated with ROS and Cell Cycle. We further analyzed the overall survival (OS) of the five ROS-related miRNAs in HCC, only the OS of miR-210-3p and miR-106a-5p have statistical significance ( Figure S3). We further find that miR-210-3p and miR-106a-5p increased ROS levels in huh7 and smcc7721 cells (Figure 8(a)). It is reported that ROS plays a role in radiation-induced cancer cell death [29]. Furthermore, we examined the effect of miRNAs on HCC cell cycle. Interestedly, miR-106a-5p and miR-210-3p stagnated the cell cycle at G2/M phases; the results were more obvious in cells after IR (Figures 8(b)-8(e)). It indicated that ROSrelated miRNAs might improve the radiotherapy sensitivity of HCC. Additionally, GO enrichment analysis was performed for the common target genes of miR-210-3p and miR-106a-5p; the results were similar to GSEA analysis (Figure 4(c)).

Discussion
The mechanisms of ROS in cancer are complex for numerous reasons [29]. First, ROS play a key role in the development of cancer [30][31][32][33]. Second, ROS induced G2/M arrest leads to affect the cell-cycle progression [34][35][36][37]. Third, chronic inflammation is regulated by ROS, which promote tumorigenesis [38,39]. Forth, the expression levels of various tumor-related genes are regulated by ROS, including p53 [40][41][42]. Fifth, ROS are associated with cell apoptosis [43,44]. Sixth, lots of chemotherapeutic and radiotherapeutic agents kill cancer cells by increasing ROS levels [45,46]. Although ROS act on tumors through numerous pathways, the specific mechanisms remain unclear. ROS have a dual effect on tumors [29]. For example, ROS can promote tumor cell proliferation, such as bladder, liver, breast, lung, and ovarian cancer cells [47][48][49][50]. However, ROS also inhibits tumor cell proliferation, including liver, prostate, and breast cancer cells [51][52][53]. According to these previous reports, the role of ROS in tumors is contradictory and complex. Therefore, the mechanism of ROS in tumors needs to be further investigated. Currently, the effect of ROS in HCC needs further analyzed.
Furthermore, 340 HCC patients were randomly divided into a validation dataset (170 patients) and training dataset (170 patients). Five candidate HBV-related ROS miRNAs were selected to calculate ROS score by using LASSO regression in the training dataset, including let-7a-5p, miR-106a-5p, miR-144-5p, miR-20b-5p, and miR-210-3p. ROS score was calculated by using the formula. Patients were divided into high-risk (high ROS score) and low-risk groups (low ROS score), based on the median ROS score. The high-risk groups were significantly associated with bad prognosis in the training dataset. Also, the similar result occurred in the validation dataset. Moreover, the AUC of five candidate HBV-related ROS miRNAs was 0.76, 0.81, and 0.78 at one year, three years, and five years in training dataset. In regard to the validation dataset, the AUC of the miRNAs was 0.58, 0.72, and 0.68 at one year, three years, and five years. AUC > 0:5 indicated prognostic ability, while AUC > 0:7 indicated a strong prognostic ability. The results showed that five candidate HBV-related ROS miRNAs had a strong prognostic ability.

20
Oxidative Medicine and Cellular Longevity natural and acquired immunity and the immune response [21]. Thus, previous reports confirm our results. In regard to gene mutation, the mutation rates of HCC patients with high ROS score were higher, and TP53 mutation rates were the highest among all genes in HCC patients with high ROS score. The results have also been verified in cluster1. Due to the lack of antioxidant damage repair system, mitochondria DNA (mtDNA) is more susceptible to ROS damage and mutation [61]. In turn, mtDNA mutations can increase ROS production, further aggravating the mutation effect and accumulating mutations, thus increasing the risk of tumor mutation [62]. What is more, ROS promote HCC cell survival by regulating TP53 degradation [63]. These researches confirm our results.
Regarding to m6A methylation, Zhuang and his colleagues find that m6A demethylase FTO could induce oxidative stress and ROS production and show impaired tumor growth [64]. Also, Yu and his colleagues find that ROS significantly induces global mRNA N6-methyladenosine (m6A) levels by modulating ALKBH5, to induce various biological processes quickly and effectively including DNA damage repair [65]. In this research, it is the first time to report that ROS is significantly associated with m6A in HCC. However, there are few reports about ROS and m6A; the mechanism of ROS and m6A needs further exploration.
Many chemotherapeutic agents kill tumor cells through ROS production [29]. For example, procarbazine is one of the first drugs to be developed based on its ROS-producing properties [45]. Until to now, the drug is approved to treat Hodgkin's lymphoma, non-Hodgkin's lymphoma, and primary brain tumors [66,67]. In our research, we analyzed the difference of chemotherapy drugs between high-ROS score patients and low-ROS score patients. The results showed that "Sorafenib," "Gefitinib," "Rapamycin," and "Lapatinib" were more highly expressed in high-ROS score patients, which indicated that ROS could increase the chemotherapy sensitivity in HCC.
Finally, our experiments showed that ROS-related miR-NAs miR-210-3p and miR-106a-5p could increase the ROS level of HCC. Moreover, these two miRNAs could stagnate HCC cell cycle at G2/M; the results were more obvious in cells after IR. The relative radiosensitivity of cells is determined by the cell cycle stage. Cells are most sensitive to radiation in G2/M phase, less sensitive in G1 phase, and least sensitive in late S phase [68].
Therefore, ROS-related miRNAs miR-210-3p and miR-106a-5p can increase the radiotherapy sensitivity of HCC cells. ROS play a role in radiation-induced cancer cell death, including lung cancer, prostate cancer, and breast cancer [69][70][71]. Our experimental results provide data support for    23 Oxidative Medicine and Cellular Longevity the role of ROS in HCC radiotherapy. Moreover, the experimental results showed that miR-210-3p and miR-106a-5p suppressed HCC cell proliferation, migration, and invasion. And two miRNAs promoted apoptosis of huh7 and smmc721 cells. It indicated that miR-210-3p and miR-106a-5p played a key role in the biological function of HCC.