MiR-371a-5p Positively Associates with Hepatocellular Carcinoma Malignancy but Sensitizes Cancer Cells to Oxaliplatin by Suppressing BECN1-Dependent Autophagy

Oxaliplatin (OXA)-based chemotherapy demonstrates active efficacy in advanced hepatocellular carcinoma (HCC), while resistance development limits its clinical efficacy. Thus, identifying resistance-related molecules and underlying mechanisms contributes to improving the therapeutic efficacy of HCC patients. MicroRNA-371a-5p (MiR-371a-5p) fulfills an important function in tumor progression. However, little is known about the effect of miR-371a-5p on chemotherapy response. In this study, quantitative real-time polymerase chain reaction, Western blot and immunohistochemistry were used to determine the expression levels of miR-371a-5p, BECN1 and autophagy-related proteins in HCC cells, tissues and serum. The luciferase reporter assay was used to assess the directly suppressive effect of miR-371a-5p on BECN1 mRNA translation. Moreover, gain- and loss-of-function assays and rescue assays were used to evaluate the mediated effect of BECN1-dependent autophagy on the role of miR-371a-5p in the response of HCC cells to OXA. We found that miR-371a-5p was significantly up-regulated in HCC tissues and serum from patients, whereas BECN1 protein was down-regulated in HCC tissues compared to the corresponding controls. We also found that there was a negative correlation between the two molecules in HCC tissues. In addition, we found that miR-371a-5p expression was positively associated with malignant characteristics of HCC and BECN1 protein expression is negatively associated. Contrary to this, we found that miR-371a-5p enhances and BECN1 attenuates the response of HCC cells to OXA. Importantly, the enhanced effect of miR-371a-5p on the response of HCC cells to OXA could be reduced by re-expression of non-targetable BECN1, and then the reduced effect was restored following bafilomycin A treatment. Taken together, we identified a dual role of miR-371a-5p in HCC malignant characteristics and the response of HCC cells to oxaliplatin. Importantly, we reveal that miR-371a-5p enhances oxaliplatin response by target suppression of BECN1-dependent autophagy.


Introduction
Hepatocellular carcinoma (HCC) is the sixth frequent malignancy and the fourthleading cancer-related cause of death in the world [1,2], accounting for more than 8% of all cancer-related deaths [3], with an overall 5-year survival of patients less than 12% [4]. Most HCC patients are diagnosed late, when curative resection is impracticable. Therefore, new regimens are urgently needed to treat advanced HCC. The oxaliplatin-based chemotherapy regimen has been shown to be effective on advanced HCC [5,6], while resistance development often results in treatment failure. Therefore, uncovering resistance-related molecules and underlying mechanisms helps to improve the therapeutic efficacy of HCC patients.

MiR-371a-5p Is Associated with HCC Malignant Characteristics and the Response of Patients to FOLFOX4 Therapy
MiR-371a-5p expression was examined by QPCR in 62 cases of fresh HCC tissues, patients' serum and their counterparts. We found that miR-371a-5p expression was up-regulated in HCC tissues compared to paired adjacent non-cancer liver tissues (PNCT) (Figure 1a). Similarly, miR-371a-5p expression was higher in serum from prechemotherapeutic HCC patients than in serum from health controls ( Figure 1b). As shown in Table 1, miR-371a-5p expression in HCC tissues, as well as in serum from prechemotherapeutic HCC patients, was associated with tumor size, tumor differentiation and patient age. Intriguingly, miR-371a-5p expression in HCC tissues from patients with serum AFP > 400 ug/L was higher than that from patients with serum AFP ≤ 400 ug/L, while there was no significant difference in miR-371a-5p expression between the serum of patients from the two group (Table 1). Moreover, we found that miR-371a-5p expression was significantly reduced in serum from postoperative patients compared to in serum from preoperative patients (Figure 1c). peutic HCC patients between the two groups ( Figure 1f). Moreover, we also found that in HCC patients underwent FOLFOX4 treatment, miR-371a-5p expression was decreased in postchemotherapeutic serum compared to in prechemotherapeutic serum (Figure 1g). The same result was determined in FOLFOX4-sensitive HCC patients (Figure 1h), while there was no difference in serum from FOLFOX4-resistant HCC patients (Figure 1i). These results indicated that HCC patients with high miR-371a-5p expression responded well to FOLFOX4 chemotherapy. Figure 1. MiR-371a-5p is associated with HCC malignant properties and the response of patients to FOLFOX4 treatment. (a) miR-371a-5p expression was detected by QPCR in HCC and the paired adjacent non-cancer tissues (PNCT) (n = 62). (b) miR-371a-5p expression in serum from 62 prechemotherapeutic HCC patients and 11 health controls was detected by QPCR. (c) miR-371a-5p expression in serum from preoperative and postoperative HCC patients was detected by QPCR (n = 62). (d−f) miR-371a-5p expression in HCC tissues (d), prechemotherapeutic serum (e) and postchemotherapeutic serum (f) between 11 FOLFOX4-resistant and 6 FOLFOX4-sensitive patients was detected by QPCR. (g−i) miR-371a-5p expression in serum from 17 patients underwent FOLFOX4 treatment patients (g), 6 FOLFOX4-sensitive HCC patients (h) and 11 FOLFOX4-resistant HCC patients (i) between pre-and post-chemotherapy was detected by QPCR. Figure 1. MiR-371a-5p is associated with HCC malignant properties and the response of patients to FOLFOX4 treatment. (a) miR-371a-5p expression was detected by QPCR in HCC and the paired adjacent non-cancer tissues (PNCT) (n = 62). (b) miR-371a-5p expression in serum from 62 prechemotherapeutic HCC patients and 11 health controls was detected by QPCR. (c) miR-371a-5p expression in serum from preoperative and postoperative HCC patients was detected by QPCR (n = 62). (d-f) miR-371a-5p expression in HCC tissues (d), prechemotherapeutic serum (e) and postchemotherapeutic serum (f) between 11 FOLFOX4-resistant and 6 FOLFOX4-sensitive patients was detected by QPCR. (g-i) miR-371a-5p expression in serum from 17 patients underwent FOLFOX4 treatment patients (g), 6 FOLFOX4-sensitive HCC patients (h) and 11 FOLFOX4-resistant HCC patients (i) between pre-and post-chemotherapy was detected by QPCR.
We then investigated the association of miR-371a-5p expression with the response of HCC patients to FOLFOX4 therapy. We found that miR-371a-5p expression was lower in tissue from FOLFOX4-resistant HCC patients than in tissue from FOLFOX4-sensitive HCC patients (Figure 1d). miR-371a-5p expression in serum from prechemotherapeutic HCC patients between the two groups showed the same result (Figure 1e), while there was no statistical significance in miR-371a-5p expression in serum from postchemotherapeutic HCC patients between the two groups ( Figure 1f). Moreover, we also found that in HCC patients underwent FOLFOX4 treatment, miR-371a-5p expression was decreased in postchemotherapeutic serum compared to in prechemotherapeutic serum (Figure 1g). The same result was determined in FOLFOX4-sensitive HCC patients (Figure 1h), while there was no difference in serum from FOLFOX4-resistant HCC patients (Figure 1i). These results indicated that HCC patients with high miR-371a-5p expression responded well to FOLFOX4 chemotherapy.

MiR-371a-5p Enhances the Response of HCC Cells to Oxaliplatin In Vitro
To investigate the effect of miR-371a-5p on the treatment response, we evaluated the correlation between miR-371a-5p expression and the half inhibitory concentration (IC50) of oxaliplatin (OXA), a main anti-cancer component of the FOLFOX4 regimen in four HCC cells. We found that miR-371a-5p expression was lowest in HepG2 cell and highest in SMMC-7721 cell in the four HCC cells (Figure 2a), while the IC50 of the four cells to OXA showed an opposite trend (Figure 2b). An inverse correlation existed between the two, although the result did not show the statistical difference ( Figure 2c). Gain-and loss-offunction assays showed the IC50 of OXA in SMMC-7721 cell was increased in the miR-371a-5p silencing group with the miR-371a-5p inhibitor treatment for 72 h (72 h miR-371a-5p inhibitor) compared to the nonsense control (NC) group (Figure 2d Figure 2d,e. Noticeably, we found that compared to the 72 h miR-371a-5p inhibitor group, the OXA IC50 in SMMC-7721 cell was not dramatically altered in the miR-371a-5p inhibitor pretreatment group (Figure 2d). The same result was measured in the 72 h miR-371a-5p mimics group in HepG2 cell (Figure 2e). Similarly, flow cytometry (FCM) results of the OXA-induced cytotoxicity showed that the apoptosis proportion of SMMC-7721 cell in the 72 h miR-371a-5p inhibitor group, as well as in the miR-371a-5p inhibitor pretreatment group was reduced compared to the NC group (Figure 2f), and that of HepG2 cell in the 72 h miR-371a-5p mimics group and in the miR-371a-5p mimics pretreatment group showed an opposite result (Figure 2g). These results indicated that miR-371a-5p could enhance the sensitivity of HCC cells to OXA in vitro.
Life 2022, 12, x FOR PEER REVIEW 5 of 17 OXA IC50 between the NC group and the miR-371a-5p silencing/over-expression group with the miR-371a-5p inhibitor/mimics treatment for 48 h (48 h miR-371a-5p inhibitor/mimics), as shown in Figure 2d,e. Noticeably, we found that compared to the 72 h miR-371a-5p inhibitor group, the OXA IC50 in SMMC-7721 cell was not dramatically altered in the miR-371a-5p inhibitor pretreatment group (Figure 2d). The same result was measured in the 72h miR-371a-5p mimics group in HepG2 cell (Figure 2e). Similarly, flow cytometry (FCM) results of the OXA-induced cytotoxicity showed that the apoptosis proportion of SMMC-7721 cell in the 72 h miR-371a-5p inhibitor group, as well as in the miR-371a-5p inhibitor pretreatment group was reduced compared to the NC group (Figure 2f), and that of HepG2 cell in the 72h miR-371a-5p mimics group and in the miR-371a-5p mimics pretreatment group showed an opposite result (Figure 2g). These results indicated that miR-371a-5p could enhance the sensitivity of HCC cells to OXA in vitro.  Moreover, we examined the impact of miR-371a-5p on the apoptosis of HCC cells. We found that there was no difference in apoptosis between the miR-371a-5p silencing

MiR-371a-5p Represses HCC Cell Autophagy by Target Suppression of BECN1
Given that miRNAs come into play depending on their target genes, we analyzed potential functions of miR-371a-5p target genes by the on-line software OMICSBEAN (http://www.omicsbean.cn/, accessed on 5 March 2019). The results showed that the functions were implicated in the regulation of biological processes, cell components and molecular functions (Supplementary Figure S2a). Of them, approximately 10% targets were implicated in the regulation of autophagy and macro-autophagy ( Figure 3a). To verify the bioinformatic results, we investigated the effects of miR-371a-5p on autophagy in HCC cells. WB results showed that the miR-371a-5p silencing increased the ratio of LC3 II/I and reduced the SQSTM1/p62 expression in SMMC-7721 cell, while the miR-371a-5p over-expression exhibited an opposite result in HepG2 cell (Figure 3b). Similarly, the results of TEM showed that the miR-371a-5p silencing increased the frequencies of autophagic vacuoles in SMMC-7721 cell, while the miR-371a-5p over-expression decreased the frequencies of autophagic vacuoles in HepG2 cell ( Figure 3c).
We sought to explore the mediator responsible for miR-371a-5p-related autophagy in HCC cells. We screened the autophagy-related targets of miR-371a-5p in TargetScan Human 8.0 (https://www.targetscan.org/vert_80/, accessed on 12 October 2021), the most common on-line software predicting the relationship between miRNAs and targets. We identified three genes that met our selection criteria, beclin1, autophagy related (BECN1), autophagy related 14 (ATG14) and autophagy related 13 (AT13) in 409 potential targets (Supplementary Table S1). To further narrow down target candidates, we constructed wildtype (Wt) and mutant-type (Mut) luciferase report systems of BECN1, ATG14 and AT13 mRNA 3 UTR in pmirGLO vector (Supplementary Figure S2b). The reporter assays showed that the ectopic miR-371a-5p expression dramatically suppressed the luciferase activity of Wt BECN1 mRNA 3 UTR in 293T cell and SMMC-7721 cell, while the suppressive effect was abolished in Mut BECN1 mRNA 3 UTR report systems ( Figure 3d). The phenomenon was not observed in the ATG13 and ATG14 mRNA 3 UTR report systems (Supplementary Figure S2c,d). Consistent with this, WB results showed that the ectopic miR-371a-5p reduced BECN1 protein expression in HepG2 cell and vice versa in SMMC-7721 cell, while the protein expressions of ATG13 and ATG14 were not altered ( Figure 3e). Moreover, we found that the increased effects of the miR-371a-5p silencing on the autophagy-related phenotypes in SMMC-7721 cell could be reversed by silencing BECN-1 (Figure 3b,c and Supplementary Figure S2e,f). We also found that the decreased effects of the miR-371a-5p over-expression on the autophagy-related phenotypes in HepG2 cell could be rescued by the re-expression of BECN1 without 3 UTR (Figure 3b,c and Supplementary Figure S2e,f).
These results indicated that miR-371a-5p regulated HCC cell autophagy by suppressing BECN1, not ATG13 or ATG14 expression.

Inhibition of BECN1-Dependent Autophagy Is Essential for the miR-371a-5p-Caused Sensitivity of HCC Cells to Oxaliplatin In Vitro
To determine whether BECN1 is indispensable to the miR-371a-5p function, we firstly evaluated the effect of BECN1 on the response of HCC cells to OXA. We found that HepG2 cell silencing BECN1 showed a decreased IC50 to OXA and an elevated OXA-induced apoptosis compared to the control (Supplementary Figure S3a We then performed rescue assays to investigate the mediated effect of BECN1-dependent autophagy on the role of miR-371a-5p in the OXA response in vitro. We found that the miR-371a-5p silencing increased the OXA IC50 compared to the control, while the increased effect could be attenuated by the BECN1 silencing in SMCC-7721 cell (Figure 4a,b). We also found that the miR-371a-5p over-expression reduced the OXA IC50 compared to the control, while the reduced effect could be rescued by the re-expression of BECN1 without 3 UTR in HepG2 cell (Figure 4c,d). Importantly, we found the rescued effects of re-expression of BECN1 without 3 UTR on the OXA IC50 reduction by the miR-371a-5p over-expression were abolished following bafilomycin A1 (BafA1) treatment (Figure 4c,d). Consistent with these results, the miR-371a-5p silencing reduced OXA-induced apoptosis compared to the control, while the reduced effect could be rescued by the BECN1 silencing in SMMC-7721 cell ( Figure 4e). Inversely, the miR-371a-5p over-expression increased OXA-induced apoptosis compared to the control, while the increased effect could be abolished by the re-expression of BECN1 without 3 UTR in HepG2 cell (Figure 4f). Importantly, the abolished effect of the BECN1 re-expression on the increased apoptosis by miR-371a-5p over-expression was reversed following BafA1 treatment (Figure 4f). These results indicated that miR-371a-5p enhanced the response of HCC cells to OXA by inhibiting BECN1-dependent autophagy.   To determine whether BECN1 is indispensable to the miR-371a-5p function, we firstly evaluated the effect of BECN1 on the response of HCC cells to OXA. We found that BECN1 silencing in SMMC-7721 cell (Figure 4e). Inversely, the miR-371a-5p over-expres-sion increased OXA-induced apoptosis compared to the control, while the increased effect could be abolished by the re-expression of BECN1 without 3′UTR in HepG2 cell ( Figure  4f). Importantly, the abolished effect of the BECN1 re-expression on the increased apoptosis by miR-371a-5p over-expression was reversed following BafA1 treatment (Figure 4f). These results indicated that miR-371a-5p enhanced the response of HCC cells to OXA by inhibiting BECN1-dependent autophagy.

BECN1 Is Down-Regulated in HCC Tissues and Associated with HCC Malignant Characteristics
Having identified the negative regulatory effect of miR-371a-5p on BECN1 in the response of HCC cells to OXA, we speculated that the BECN1 expression was downregulated in HCC. Unexpectedly, the data from the CPTAC database, a web resource for analyzing cancer OMICS data, showed that the proteins of BECN1 were dramatically up-regulated in HCC tissue samples compared to in normal tissues (Figure 5a). To verify this, we detected the BECN1 protein expression in the 62 paraffin-embedded tissue samples. Oppositely, the IHC result showed that the BECN1 protein expression was significantly lower in HCC tissues than in adjacent non-cancer liver tissues (Figure 5b,e). We found that the BECN1 expression was higher in tissues from FOLFOX4-resistant HCC patients than in tissue from FOLFOX4-sensitive HCC patients (Figure 5c).
Moreover, Pearson's correlation analysis showed there was an inverse correlation between the BECN1 protein expression and miR-371a-5p expression in HCC tissues ( Figure  5d). We then evaluated the contributions of the BECN1 protein expression to HCC malignant characteristics based on their IHC staining extent. As shown in Figure 5c and Table  2, the BECN1 protein expression was negatively associated with tumor size and differentiation. These results indicated that the BECN1 protein was down-regulated in HCC and negatively associated with HCC malignant characteristics. Moreover, Pearson's correlation analysis showed there was an inverse correlation between the BECN1 protein expression and miR-371a-5p expression in HCC tissues (Figure 5d). We then evaluated the contributions of the BECN1 protein expression to HCC malignant characteristics based on their IHC staining extent. As shown in Figure 5c and Table 2, the BECN1 protein expression was negatively associated with tumor size and differentiation. These results indicated that the BECN1 protein was down-regulated in HCC and negatively associated with HCC malignant characteristics.

Discussion
Most HCC was diagnosed in advanced stages, when patients had few chemotherapeutic options. The OXA-based FOLFOX4 chemotherapy regimen has recently exhibited active efficacy and excellent safety and been approved for advanced HCC in China [6,22]. However, the development and neglect of resistance to the regimen reduces its clinical benefits. Therefore, it is urgently needed to identify reliable biomarkers associated with the response of patients to the regimen.
In this study, we found that miR-371a-5p expression was higher in HCC tissue and prechemotherapeutic serum from FOLFOX4-sensitive patients than in these from FOLFOX4resistant patients. Interestingly, there was no significant difference in postchemotherapeutic serum between patients from the two group. One possible explanation for this result is that the FOLFOX4 regimen could suppress miR-371a-5p expression, which was supported by the result that the miR-371a-5p serum level was decreased after patients underwent FOLFOX4 treatment. In fact, we detected the decreased miR-371a-5p expression following OXA treating HCC cells in an ongoing study (data not shown). In addition, we found that the miR-371a-5p serum level was decreased in FOLFOX4-sensitive patients after the chemotherapy, whereas the phenomenon was not observed in FOLFOX4-resistant patients. This finding appears to indicate the possibility that the miR-371a-5p serum level could be a potential biomarker for monitoring the response of HCC patients to FOLFOX4 treatment.
Although miR-371a-5p has been identified to function as an onco-miRNA in HCC [11,12,23], little is known about the role of miR-371a-5p in chemotherapy response [13]. Based on the differential expression of miR-371a-5p between FOLFOX4-sensitive and resistant patients in clinical settings, we speculated that miR-371a-5p may be involved in the regulation of the therapy response. To test it, we explored the effect of miR-371a-5p on the response of HCC cells to OXA, the main component of FOLFOX4 regimen, which exhibited better efficacy and safety than other parallel chemotherapeutic agents and was preferred as a postoperative maintenance therapy in advanced HCC patients [22]. Here, we found that there was a potentially inverse correlation between miR-371a-5p expression and the IC50 of HCC cells to OXA, suggesting that miR-371a-5p exerted a positive role in coordinating the response of HCC cells to OXA. As expected, gain-and loss-of-function assays showed that the miR-371a-5p silencing lowered the response of SMMC-7721 cell to OXA, and the miR-371a-5p over-expression showed an opposite effect in HepG2 cell. These results were consistent with the finding that miR-371a-5p promotes the response of colorectal cancer cells to cetuximab [13]. What puzzles us was the effect of miR-371a-5p on the OXA response was observed in HCC cells with the inhibitor or mimics and OXA co-treatment for 72 h, but not for 48 h. We presumed that it may be required for the pretreatment of HCC cells with the miR-371a-5p inhibitor or mimics in the reduced or enhanced OXA-induced cytotoxicity. To verify this conjecture, we designed miR-371a-5p inhibitor and mimics pretreatment assays (details were shown in the material section). We found that there was no difference in the response of HCC cells to OXA between 72 h miR-371a-5p inhibitor treatment and the miR-371a-5p inhibitor pretreatment, as well as between 72h miR-371a-5p mimics treatment and the miR-371a-5p mimics pretreatment. Hence, we have reason to believe that miR-371a-5p could play an important role in reducing OXA side effects.
To explore the underlying mechanism by which miR-371a-5p enhances the OXA response, we executed bioinformatics analysis for functions of miR-37a-5p target genes. The enrichment of them in the autophagy pathway caught our eyes. Gain-and loss-of-miR-371a-5p assays showed that the miR-371a-5p over-expression repressed autophagy-related phenotypes and vice versa, which was consistent with a most recent study [21]. Nevertheless, whether autophagy is responsible for the effect of miR-371a-5p on the response of HCC cells to OXA still needs to be identified. Bioinformatics analysis screened three candidate mediators related to autophagy. Mechanically, we found that miR-371a-5p suppressed the BECN1 expression by binding to the 3 UTR of BECN1 mRNA. Functionally, rescue assays showed that the BECN1 silencing restored the reduced effect of miR-371a-5p inhibitor on the OXA response, and the non-target BECN1 re-expression reversed the increased effect of miR-371a-5p mimics on the OXA response. Importantly, the autophagy blockage with BafA1 eliminated the effect of the non-target BECN1 re-expression. These results demonstrate that miR-371a-5p enhances the response of HCC cells to OXA by target suppression of BECN1-dependent autophagy.
In sharp contrast with the contribution of miR-371a-5p to the response of HCC cells to OXA, we found that miR-371a-5p expression in HCC tissue, as well as in serum from prechemotherapeutic HCC patients was positively associated with the size and grading of HCC. Additionally, previous studies have demonstrated that miR-371a-5p promotes proliferation of HCC cells by inducing G1/S transition [11]. These findings indicate that miR-371a-5p fulfills a dual function in cancer development and therapy. The duality seems paradoxical, while it is a common phenomenon. It has been reported that the activity of cancer-promoting gene Thymidine Phosphorylase (TP) is necessary for the activation of various chemotherapeutic agents [24]. Similarly, inactivation of tumor suppressor gene p53 strengthens the sensitivity to multiple chemotherapeutic drugs [25]. Moreover, the tumor dormancy theory indicates that conventional chemotherapy preferentially eradicates actively proliferating cancer cells, whereas the relatively quiescent or slowly proliferating ones escape from the therapy and survive by inducing autophagy [26,27], which may provide a decent explanation for the duality.
To our surprise, miR-371a-5p expression data from HCC patients in the two age groups with a cut-off age of 50 reached statistical significance. On the one hand, the difference may be due to the effect of age on the expression of miRNAs. This viewpoint is confirmed by numerous studies [28][29][30], and an earlier study that the age-dependent 1a, 25 dihydroxy vitamin D3 and testosterone decline results in miR-371a-5p down-regulation reinforces it [31]. On the other hand, the impact of age on HCC malignant behaviors could contribute to the difference [32,33]. Additionally, the difference in the association of miR-371a-5p expression with serum AFP level did not display the consistency in tissue level and serum level. In our opinion, it may be due to the limitation of the small sample size, large variety of a clinical setting and the impacts of the FOLFOX4 regimen on miR-371a-5p expression. Further studies with a large sample size and investigating the effects of the component of the FOLFOX4 regimen on the expression of miR-371a-5p are necessary to clarify the issue. It is worth noting that the serum miR-371a-5p level of patients was substantially decreased after HCC resection, which may provide a potential biomarker for assessing the degree of tumor resection. It also seems like a possibility that dynamically monitoring HCC recurrence by detection of miR-371a-5p.
According to the negative regulatory effect of miR-371a-5p on BECN1, the BECN1 expression should be down-regulated in HCC tissue. Unexpectedly, proteomic results from the CPTAC database indicate that the BECN1 protein expression was up-regulated in HCC tissues compared to in normal tissues, which was paralleled by transcriptomics from the TCGA database at the mRNA level (http://ualcan.path.uab.edu/cgi-bin/ualcan-res.pl, accessed on 12 May 2022). In the present study, however, we detected the diminished BECN1 protein expression in HCC tissues compared to in PNCT by IHC. Consistent with our results, multiple studies demonstrate that both protein and mRNA levels of BECN1 are lower in HCC tissues and cells than in the corresponding normal controls [34][35][36]. To our knowledge, the discrepant expression pattern of BECN1 in HCC tissues may be owing to the differences in testing means and study object, including the tumor histological type, stage, number of patients included and patients' race [24]. A change of perspective, the discrepancies are just according with the duality of autophagy in HCC [17]. Contrary to the contribution of miR-371a-5p to HCC, BECN1 was negatively associated with the size and grading of HCC but hampered the response of HCC cells to OXA. The negative relationship between BECN1 and miR-371a-5p was confirmed by Pearson's correlation analysis in HCC tissue.
The lack of in vivo experiments is a major deficiency of the present study. Most researchers currently establish the tumor model with subcutaneous injection of cancer cells in nude mice, and then agents are administrated intraperitoneally to inspect their effects on the tumor growth [4,37]. It is undeniable that the model mimics metabolism and circulation of agents in the body to some extent, while in our opinion, the effects of the specific tumor micro-environment on agents and cancer cells are neglected. A spontaneous tumor model from HCC would be required to explore the effect of miR-371a-5p on the response of HCC cells to OXA. This is currently impossible but could be done in future research. Moreover, insufficient follow-up time prevented us from inspecting the effect of miR-371a-5p on the prognosis of HCC patients, so longitudinal records of patients' survival condition will be a key point for us in next work.

Clinical Samples
A total of 62 fresh tissue samples consisting of HCC tissue and the paired adjacent non-cancer liver tissue (PNCT) were obtained from patients undergoing liver resection at Nanchong Central Hospital between January 2019 and January 2021. Among 62 patients, 19 patients underwent preoperative chemotherapy with FOLFOX4 regimen according to the protocol [6]. Briefly, bolus 5-FU 400 mg/m 2 , infusional 5-fluorouracil 600 mg/m 2 over 22 h on day 1 and 2, bolus leucovorin 100-200 mg/m 2 , and oxaliplatin 85 mg/m 2 , once every 2 weeks. As described in a previous study [4], the appearance of new lesions or the increase of tumor growth > 30% were defined as the FOLFOX4 resistance while the increase of tumor growth < 20% defined as the FOLFOX4 sensitivity following two months of chemotherapy. Based on this, we included 11 FOLFOX4-resistant and 6 FOLFOX4-sensitive patients in this study. Blood samples were collected from the 62 HCC patients on the day before chemotherapy (named as prechemotherapy), the day before operation (named as preoperation or postchemotherapy if patients underwent FOLFOX4 therapy) and on the seventh day after operation (named as postoperation) and 11 healthy volunteers (controls). All the blood samples were allowed to stand for 30 min and then fresh serum was separated by centrifuging at 4 • C, 4000 rpm for 5 min.
All fresh samples were stored at −80 • C until used. The clinicopathological characteristics of all the patients were shown in Table 1. All human experiments were carried out with the approval of the ethics committee of the Nanchong Central Hospital and the written informed consent was obtained from patients.

Quantitative Real-Time Polymerase Chain Reaction (QPCR)
Total RNAs in tissues and cells were isolated using the TRIzol method (Invitrogen) according to the manufacturer's instructions. In the process of serum RNA extraction, 1 ug lyophilized cel-miR-39-3p (Qiagen, Hilden, Germany) was added as exogenous reference according to the provider recommendations and a miRNeasy serum/plasma kit was used to extract serum cell-free miRNAs according to manufacturer's protocol. The Spectrophotometer ND-1000 (NanoDrop Technologies, Wilmington, NC, USA) was used to detect RNA concentration. The expression levels of miRNAs were determined using All-in-One miRNA qRT-PCR Detection Kit (GeneCopeia, Rockville, MD, USA) in the ABI 7500 System (Applied Biosystems, Foster City, CA, USA) according to the protocol. The relative expression of miR-371a-5p in tissues and cells were normalized to U6 and that in serum to cel-miR-39 using the 2 −∆∆Ct method. The primers were as follows: cel-miR-39, 5 -CAGAGTCACCGGGTGTAAAT-3 ; miR-371a-5p, 5 -TGCGGACTCAAACTGTGGGGGC-3 ; u6, 5 -TGCGGGTGCTCGCTTCGGCAGC-3 .

Western-Blot (WB) and Immunohistochemistry (IHC)
WB and IHC were performed according to established methods in the previous study [38]. Primary antibodies were used at appropriate dilution concentration as fol-

Luciferase Reporter Assay
As shown in Supplementary Figure S2b, the wild-type (Wt) sequences of ATG13, ATG14 and BECN1 mRNA 3 UTR containing putative miR-371a-5p binding site and their corresponding mutant-type (Mut) sequences were synthesized by Wuhan GeneCreate Biological Engineering Co., Ltd. (China) and, respectively, inserted into the pmirGLO plasmid (Promega, Madison, WI, USA). DNA sequencing was used to validate the eligibility of recombinant plasmid. The recombinant plasmid and miR-371a-5p mimics were co-transfected into 293T and SMMC-7721 cells using Lipofectamine 3000 (Invitrogen). Following 72 h, the Dual-Luciferase Reporter Assay Kit (Promega) was used to determine the luciferase activity according to the manufacturer's instructions.

Transmission Electron Microscopy (TEM)
Approximately 10 7 candidate HCC cells were harvested and fixed with 2.5% glutaraldehyde in phosphate buffer for 90 min. The cells were postfixed with 1% osmium tetroxide for 30 min followed by a gradient dehydration using ethanol and acetone and then embedded in Epon 812 resin. The 60-nm ultra-thin sections were obtained using an ultramicrotome (Leica Microsystems, Wetzlar, Germany) and placed on uncoated copper grids. The sections were stained with 3% lead citrate-uranyl acetate and then observed under the TECNAI20 electron microscope (Philips, Amsterdam, The Netherlands). The frequencies of autophagic vacuoles per cell were calculated by averaging the frequencies of autophagic vacuoles in 10 cells.

Statistical Analysis
All statistical analyses were performed using SPSS 22.0 software. Quantitative results with a normal distribution are presented as the mean ± SD and these with an abnormal distribution are presented as median. Based on the result of homogeneity test for variance, Paired student t test or Wilcoxon matched-pairs signed rank test was used to analyze statistical significance of the paired data, the independent-samples t test or the Mann-Whitney test was used to analyze statistical significance of the data between two groups, one-way ANOVA or the Kruskal-Wallis test was used to analyze statistical significance of the data among groups. Pearson's correlation analysis was used to determine the correlation. A p value < 0.05 was considered as a significant difference.

Conclusions
In this study, we identify a dual contribution of the miR-371a-5p/BECN1 axis to clinical characteristics and the therapy response in HCC. On the one hand, we find that miR-371a-5p is positively and BECN1 is negatively associated with malignant characteristics of HCC. On the other hand, we reveal that miR-371a-5p is an OXA-sensitive and BECN1 is an OXAresistant gene. Importantly, we illuminate that miR-371a-5p enhances the OXA response by target suppression of BECN1-dependent autophagy. Reasonable intervention of the miR-371a-5p/BECN1-dependent autophagy pathway will be a promising sensitization strategy in HCC treatment with the OXA-based chemotherapy regimen.