Screening and Validation of a Carvacrol-Targeting Viability-Regulating Protein, SLC6A3, in Liver Hepatocellular Carcinoma

Background Liver hepatocellular carcinoma (LIHC) is the second leading cause of tumor-related death in the world. Carvacrol was also found to inhibit multiple cancer types. Here, we proposed that Carvacrol inhibited LIHC. Methods We used MTT assay to determine the inhibition of Carvacrol on LIHC cells. BATMAN-TCM was used to predict targets of Carvacrol. These targets were further screened by their survival association and expression in cancer using TCGA data. The bioinformatic screened candidates were further validated in in vitro experiments and clinical samples. Finally, docking models of the interaction of Carvacrol and target protein were conducted. Results Carvacrol inhibited the viability of LIHC cell lines. 40 target genes of Carvacrol were predicted, 8 of them associated with survival. 4 genes were found differentially expressed in LIHC vs. normal liver. Among these genes, the expression of SLC6A3 and SCN4A was found affected by Carvacrol in LIHC cells, but only SLC6A3 correlated with the viability inhibition of Carvacrol on LIHC cell lines. A docking model of the interaction of Carvacrol and SLC6A3 was established with a good binding affinity. SLC6A3 knockdown and expression revealed that SLC6A3 promoted the viability of LIHC cells. Conclusion Carvacrol inhibited the viability of LIHC cells by downregulating SLC6A3.


Introduction
Liver hepatocellular carcinoma (LIHC), the most common type of primary liver cancer, is the second leading cause of tumor-related death in the world [1]. There are 905.7 new liver cancer cases and 830.2 death from liver cancer per 100 000 people in 2020 in the world [2]. According to 2021 Cancer Statistics, liver and intrahepatic bile duct cancer accounted for 42,230 new cases and 30,230 cancer death in 2020 [3]. Over the past several decades, the study in LIHC management has made a limited improvement, and the outcome of LIHC treatment remains undesirable. The earliest FDA-approved anti-LIHC agents for late-stage LIHC treatment included sorafenib regorafenib and lenvatinib. These drugs are all subjected to low response rates, thus, further progress is required for their application in clinics [4][5][6]. Traditional medicine and naturally occurring compounds have been being studied intensively for their applications in the management of human diseases [7][8][9][10][11][12]. Traditional medicine has been applied wildly in clinical cancer treat-ment as complementary and supplementary medicine, especially in China, Korea, and Japan where traditional medicine is part of the healthcare system [13]. However, so far, there is no natural medication that has been approved by the FDA for the treatment of liver hepatocellular carcinoma. Hence, more understanding of the effect of natural medication on this type of cancer is required for future clinical applications. Carvacrol, a phenol that is a natural monoterpene derivative of cymene, has been used for antifungal, antiviral, treatment for cancer, and regulation of inflammatory activities [14]. Carvacrol was first discovered as a nonspecific inhibitor for the transient receptor potential melastatin-like 7 channel (TRPM7) [15], which is a potential target for cancers [16,17]. Carvacrol was also found to inhibit multiple cancer type, including breast cancer cell lines [18,19], cervical cancer [20,21], ovarian cancer [22], prostate cancer [23][24][25][26][27][28], colon cancer [29,30], lung cancer [31], and oral cancer [32]. However, so far, few studies have reported its effect on LIHC. An in vitro study has reported the antiproliferative and proapoptotic effect of Carvacrol on human hepatocellular carcinoma cell line HepG-2 [33]. Our hospital has applied Carvacrol in-hospital preparations for LIHC patients as supplementary medicine and has achieved desirable outcomes in many cases. Although no systematic data regarding this issue has been published, we proposed that Carvacrol might potentially inhibit LIHC. This study provided preclinical evidence to support the clinical application of Carvacrol for LIHC. In addition, although many studies reported potential targets and mechanisms of Carvacrol, pharmacological targets of Carvacrol remain largely unidentified. In this study, we screened pharmacological targets of Carvacrol in LIHC. This study identified a potential target of Carvacrol and is conducive to the optimization of the clinical application of Carvacrol in LIHC treatment.  Primary cells MEF and THLE-2 were used as normal control. Cells were exposed to 10-300 μM Carvacrol for 24 hours, and the viability was determined using MTT assay.

Disease Markers
Hep-G2 with an inhibition rate of up to 90% at 300 μM Carvacrol. The least sensitive cell lines were SNU-182 and SNU-389, both of which with an inhibition rate of 40% at 300 μM Carvacrol. In addition, we used primary cells MEF and THLE-2 as controls to demonstrate the cancer specificity of Carvacrol. Results showed that the viability of MEF and THLE-2 only significantly decreased at 300 μM ( Figure 1). Thus, we suggest that Carvacrol can inhibit the viability of LIHC cells. 200 μM was used in the subsequent study as it did not significantly affect the control cells. The viability IC50 values of Carvacrol were displayed in Table 1.

Prediction of Targets of Carvacrol in LIHC.
In this study, we first used BATMAN-TCM to predict potential targets of Carvacrol. For each compositive compound, the predicted candidate targets whose scores given by the target prediction method exceed a cutoff batman score of >40 were considered as the potential targets of Carvacrol and were presented ( Figure 2(a)). BATMAN-TCM used a similarity-based method to predict potential targets of Carvacrol, the core idea of which was to rank potential drug-target interactions based on their similarity to the known drug-target interactions [34]. The batman score was calculated as the product of the drug similarity score and the target similarity score in the known drug-target interactions. Using this algorithm, we obtain 40 target genes of Carvacrol. These target genes were further constructed into a protein-protein interaction network using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) (Figure 2(b)).

Survival Association of Targets of Carvacrol in LIHC.
To identify potential effective targets of Carvacrol in LIHC, we screened the association of all these targets using log-rank analysis using TCGA LIHC cohort. Results showed that eight target genes were significantly associated with the overall, including two protective genes, DRD1 (HR 0.  (Figure 3(a)). These eight genes were identified as potential effective targets of Carvacrol in LIHC and were further screened in the subsequent study. In addition, we further plotted the MK curves for these significant genes

The Overexpression of Target Genes in LIHC.
To identify the cancer-specific targets in LIHC, we compared the expression of these target genes in cancer vs. noncancer tissues using the TCGA LIHC cohort. First, we analyzed the coex-pression of these target genes. The most correlated genes pair was GABRA3 and GABRAQ with a coefficient factor of 0.59 ( Figure 4(a)). These data provide potential interactions of these genes. Then, we compared the expression of these target genes in cancer vs. noncancer tissues using TCGA LIHC cohort with GETx liver tissue cohort. Results showed that expressions of DRD1, GABRA3, SLC6A3, GABRQ, and SCN4A in cancer were significantly higher than those in noncancer tissues (Figure 4(b)). In addition, we further analyzed the expression level of these target genes in LIHC paired samples from TCGA data. Cancer and noncancer data from the same patients were compared and analyzed by paired t-test [35][36]. Results showed that GABRA3, SLC6A3, GABRQ, SCN4A, and GABRG3 were overexpressed in cancer compared with normal liver tissues ( Figure 4(c)). Therefore, based on these results, we suggested that GABRA3, SLC6A3, GABRQ, and SCN4A might be cancer-specific targets of Carvacrol which were further screened in the subsequent study.

Effect of Carvacrol on Expressions of Target Genes in LIHC
Cells. To further screen the target genes, we determined the effect of Carvacrol on gene expression of these target genes in two LIHC cell lines Hep-G2 and SNU-182 using QPCR. Cells were exposed to 200 μM Carvacrol for 24 hours before the assay. Results revealed that in Hep-G2 cells, the expression of SLC6A3 was significantly increased by Carvacrol, the expression of SCN4A was significantly decreased by Carvacrol, and the expressions of GABRA3 and GABRQ were not affected ( Figure 4(d)). In addition, in SNU-182, the expression of SLC6A3 was significantly increased by Carvacrol, the expression of SCN4A was significantly decreased by Carvacrol, and the expressions of GABRA3 and GABRQ were not affected ( Figure 4(e)). Thus, we suggested that SLC6A3 and SCN4A might be direct targets of Carvacrol and will be analyzed in the subsequent study.

Correlation of Target Expression and Sensitivity to
Carvacrol. In the subsequent study, we determined the protein levels of SLC6A3 and SCN4A in eight LIHC cell lines,  ). Furthermore, we determined the viability of these cell lines with or without the 24-hour exposure of 200 μM Carvacrol and calculated the viability suppression rate of these cell lines. Results showed that SNU-182 and SNU-398 had the highest viability suppression rate while Hep-G2 had the lowest viability suppression rate after the exposure to Carvacrol ( Figure 5(c)). We also further calculated the correlation of expression of SLC6A3 and SCN4A in these cell lines and their sensitivity to Carvacrol. Results showed that the expression of SCN4A in these cell lines was not correlated with their sensitivity ( Figure 5(d)), but the expression of SLC6A3 in these cell lines was significantly correlated with their viability suppression rate with a coefficient of 1.937 ( Figure 5(e)). These results indicated that SLC6A3 might mediate the effect of Carvacrol.

Effect of Carvacrol on Expressions of Target Genes in LIHC Tissue.
To validate the regulatory effect of Carvacrol on the expression of SLC6A3, we collected LIHC tissues from 18 patients with Carvacrol treatment and 18 patients without Carvacrol treatment. The prescription of Carvacrol treatment depended on the clinical condition of the patients, and the Carvacrol was applied as a supplementary therapy for patients. Hence, the dose of Carvacrol might vary from patient to patient. Nevertheless, the comparison of samples from patients with or without Carvacrol treatment can provide a hint at the effect of Carvacrol. Results showed that cancer tissue from patients with Carvacrol treatment expressed significantly lower SLC6A3 at both mRNA and protein levels compared with cancer tissue from patients without Carvacrol treatment (Figures 6(a)-6(c)). The protein staining of SLC6A3 in LIHC samples collected from patients with or without Carvacrol treatment further confirmed that SLC6A3 expression was downregulated by Carvacrol treatment (Figure 6(d)).

Binding
Potential of Carvacrol to SLC6A3. As we suggested Carvacrol might exert a direct effect on SLC6A3, we established cavity-detection guided blind docking models of the interaction of Carvacrol and SLC6A3 protein using the CB-Dock. The structure used in the docking model was from the Pubchem and the AlphaFold. The docking predicted five potential biding configurations of the interaction of Carvacrol and SLC6A3 protein, with vina scores of -7.1, -5.3, -4.5, -4.4, and -4.2 kCal/mol, respectively ( Figure 7). Vina scores of -10 or lower usually represent a very good binding, and scores of -7 to -10 kCal/mol might be considered good binding. Only one of our models passed the cutoff score of -7 kCal/mol, therefore, we suggested model one was the most likely binding configuration.  (Figures 8(a) and 8(b)). The MTT assay revealed that the overexpression SLC6A3dependently increased the viability of cells (Figure 8(c)). In      (Figures 8(a) and 8(b)). The MTT assay showed that the knockdown SLC6A3dependently decreased the viability of cells (Figure 8(c)). These results indicated that SLC6A3 positively regulated the viability of LIHC cell line SNU-449.

Discussion
A previous study has reported the potential preventive effect of Carvacrol against diethylnitrosamine-induced LIHC in rats [37]. In our study, data supported that Carvacrol inhibited the viability of multiple LIHC cell lines which might account for the preventive effect of Carvacrol against LIHC in rats. In breast cancer cells, Carvacrol at 50-500 μM significantly inhibited cell viability. For LIHC cells, our data revealed a similar effective concentration range at 10-300 μM. In addition, Carvacrol at 100-600 μM has been found to inhibit prostate cancer [22], and a higher concentration at over 500 μM was required to significantly inhibit the viability of cervical tumor cell HeLa [21]. Based on these results, we suggested that different cancer types might have different sensitivity to Carvacrol.
However, whether Carvacrol has common targets among these cancer types remains unknown. In this study, we design a novel target screening study for Carvacrol in LIHC, which can also be used for other cancer types or even pancancer studies. BATMAN-TCM used a similarity-based method to predict potential targets of Carvacrol, the core idea of which was to rank potential drug-target interactions based on their similarity to the known drug-target interactions [34]. In this study, we obtained 40 target genes, among which, a large group of them were gamma-aminobutyric acids associated. Some of them might potentially interact with ion channels that affect cancers, such as two-pore channels [38]. Furthermore, many of these are also potential tar-gets of anesthetic agents. Studies have suggested that anesthetics might potentially affect cancers [39][40][41][42][43], thus, Carvacrol might have actions to these effects.
We calculated the survival association of these target genes because we wanted to obtain potential targets that affect survival. So far, TCGA data were widely used in prognostic studies [44][45][46]. The potential association of gene expression and overall survival might identify biomarkers for cancer prognosis or functional cancer regulators for cancer. In this study, we used the survival association analysis to screen the potential functional molecule among the Carvacrol target genes. Eight target genes were identified including DRD1 and SCN4A, GABRA3, SLC6A3, GABRQ, PDE4D, GABRG3, and ALOX5. DRD1 was a gene associated with breast cancer [47] and lung cancer [48,49]. SCN4A was sodium channel genes that might also affect cancer cells [50]. But the protective effect of DRD1 and SCN4A against LIHC has not been reported. In addition, GABA-associated genes (GABRA3, GABRQ, and GABRG3) are most expressed and function in the neurotransmitter in the mammalian brain. SLC6A3 is a dopamine transporter that is a member of the sodium-and chloride-dependent neurotransmitter transporter family [51]. Another target gene PDE4D was found functioning in colon cancer [52] and bladder cancer [53]. The last target gene, ALOX5, was reported to play a role in colon cancer [54], breast cancer [55], and lung cancer [56]. So far, these genes have not been studied in LIHC.
Among these 8 genes, we identified four genes that were expressed differently in LIHC and normal liver tissue. We suggested that the expression difference between cancer and noncancer tissue might indicate the potential mediation of this target for the cancer specificity effect of Carvacrol on LIHC treatments. Our results also found that Carvacrol can affect the expression of SLC6A3 in both LIHC cell lines and LIHC in patients. Clinical samples of this study were taken from liver cancer patients who were on separate medications, which weaken the consistency of the objects. We admitted that the doses were not chosen based on any standard 7 Disease Markers criteria for bodyweight or medicine they were taking. The criteria we used to prescript was based on traditional Chinese medicine (TCM) theory which is difficult to descript in the paper. Technically, there was no preference for which patient used TCM or not because TCM is not necessary for the treatment according to the clinical instruction. Carvacrol was applied as a supplementary therapy for patients. Nevertheless, although not necessarily every one of them, it was clear that some of the patients with Carvacrol were significantly reduced in SLC6A3. We suggested the inconsistency of the medication might account for the inconsistency in the decrease of SLC6A3 in the Carvacrol treated group. More systematic evidence from clinical trials is required in the future.
Knockdown and overexpression experiment further confirmed that SLC6A3 was a biomolecule that promotes the viability of LIHC cells. Therefore, our data suggested that SLC6A3 mediated the inhibition of Carvacrol on the viability of LIHC cells. SLC6A3 has previously been reported as a potential circulating biomarker for gastric cancer detection and progression monitoring [57]. In addition, SLC6A3 was also found overexpressed and functioning in kidney cancer [58] and was suggested as a biomarker for patients with renal cell carcinoma [59]. However, to date, the role of SLC6A3 in LIHC has not been reported. In this study, we were the first to report the promotion effect of SLC6A3 on LIHC. We also predicted the binding conformation of the interaction between Carvacrol and SLC6A3 protein. Animal in vivo models have been widely used for medical study [8,60], we hope the conclusion can be further validated with in vivo experimental evidence in the future. In addition, an alternative therapeutic method for LIHC, targeting cancer stem cells, has been proposed as a promising approach [61]. As Carvacrol affected the viability of LIHC cells, we proposed that its effect might be mediated by cancer stem cells.
To conclude, in the present study, we screened pharmacological targets of Carvacrol in LIHC. We identified SLC6A3 as a potential target of Carvacrol. This study is conducive to the optimization of the clinical application of Carvacrol in LIHC treatment. [62] was used to predict potential targets of Carvacrol. The LIHC TCGA mRNA-seq data with clinical information were accessed from The Cancer Genome Atlas (TCGA) [63] in January 2020. R foundation for statistical computing (2020) version 4.0.3 and ggplot2 [64] [65, 66](v3.3.2) was used to conduced bioinformatic analysis. The structure file of Carvacrol was downloaded from Pubchem [67]. The protein structure of SLC6A3 was predicted by AlphaFold [68], a state-ofthe-art AI system developed by DeepMind. Cavitydetection-guided blind docking models of the interaction of Carvacrol and SLC6A3 protein were conducted using the CB-Dock [69].

The
Collection of LIHC Tissues. LIHC tissues were collected from 36 patients with surgical treatment or biopsy including 18 patients with Carvacrol treatment and 18 patients without Carvacrol treatment. The prescription of Carvacrol treatment depended on the clinical condition of the patients, and the Carvacrol was applied as a supplementary therapy for patients. Samples were fixed, embedded in paraffin, and stored in 4°C. All donors were over 18 years old and have given formal consent to the use of their samples. The study has been approved by the Ethics Committee of the Tumor Hospital Affiliated To Nantong University (no. 2020-082).  4.6. QPCR. The mRNA expressions were determined using a QPCR assay [70]. RNA was extracted using the RNeasy Mini kit (Qiagen, Germantown, MD, USA). The PrimeScript RT Reagent kit with gDNA Eraser (Takara Bio, Japan) and the PowerUp™ SYBR™ Green Master Mix (Thermo, Beverly, MA, USA) was used to conduct the retro transcription and QPCR. The Applied Biosystems StepOnePlus instrument (Thermo, Beverly, MA, USA) was used to run all the reactions. The results were normalized using the 2 -ΔΔCT method.

12
Disease Markers were presented in means ± standard deviations in the bar charts. A t-test or ANOVA was used to assess the significance (p < 0:05). Dunnett's post hoc tests were used to test the difference between groups. The GraphPad Prism (version 8) was used to calculate statistics.

Data Availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest
The authors declare no conflict of interest.