Cystathionine γ-lyase mediates cell proliferation, migration, and invasion of nasopharyngeal carcinoma

Nasopharyngeal carcinoma (NPC) is an epithelia-derived malignancy with a distinctive geographic distribution. Cystathionine γ-lyase (CSE) is involved in cancer development and progression. Nevertheless, the role of CSE in the growth of NPC is unknown. In this study, we found that CSE levels in human NPC cells were higher than those in normal nasopharyngeal cells. CSE overexpression enhanced the proliferative, migrative, and invasive abilities of NPC cells and CSE downregulation exerted reverse effects. Overexpression of CSE decreased the expressions of cytochrome C, cleaved caspase (cas)-3, cleaved cas-9, and cleaved poly-ADP-ribose polymerase, whereas CSE knockdown exhibited reverse effects. CSE overexpression decreased reactive oxygen species (ROS) levels and the expressions of phospho (p)-extracellular signal-regulated protein kinase 1/2, p-c-Jun N-terminal kinase, and p-p38, but promoted the expressions of p-phosphatidylinositol 3-kinase (PI3K), p-AKT, and p-mammalian target of rapamycin (mTOR), whereas CSE knockdown showed oppose effects. In addition, CSE overexpression promoted NPC xenograft tumor growth and CSE knockdown decreased tumor growth by modulating proliferation, angiogenesis, cell cycle, and apoptosis. Furthermore, DL-propargylglycine (an inhibitor of CSE) dose-dependently inhibited NPC cell growth via ROS-mediated mitogen-activated protein kinase (MAPK) and PI3K/AKT/mTOR pathways without significant toxicity. In conclusion, CSE could regulate the growth of NPC cells through ROS-mediated MAPK and PI3K/AKT/mTOR cascades. CSE might be a novel tumor marker for the diagnosis and prognosis of NPC. Novel donors/drugs that inhibit the expression/activity of CSE can be developed in the treatment of NPC.


INTRODUCTION
Hydrogen sulfide (H 2 S) is widely regarded as a key gasotransmitter [1][2][3]. H 2 S is generated from L-cysteine catalyzed by cystathionine γ-lyase (CSE) and cystathionine β-synthase (CBS) which are pyridoxal-5′-phosphate (PLP)-dependent enzymes. CBS and CSE are mainly found in the cytoplasm [4,5]. In the presence of α-ketoglutarate, a PLP-independent enzyme, 3-mercaptopyruvate sulfurtransferase  can act together with cysteine aminotransferase to generate H 2 S. 3-MST can be detected in both cytoplasm and mitochondria [5,6]. In addition, D-amino acid oxidase could metabolize D-cysteine to 3mercaptopyruvate, acting as a substrate for 3-MST to generate H 2 S in the brain and kidney [7]. H 2 S can be scavenged by methaemoglobin and disulfide/metallo-containing molecules that act as sulfane-sulfur and bound-sulfate pools. Oxidation and methylation are another two important pathways for H 2 S metabolism [3,8].
Nasopharyngeal carcinoma (NPC), one of the epithelial malignancies, arises from the most superior portion of the pharynx [9].
NPC has a specific geographic distribution, with the highest incidence in regions of South China, Southeast Asia, and North Africa [10]. A number of factors contribute to the development of NPC, such as host genetics, environmental factors, and EpsteinBarr virus infection [11]. Many patients with NPC at initial diagnosis have locally advanced disease or distant metastasis, leading to a poor clinical prognosis [12,13]. It is urgent to identify new prognostic marker and molecular target in NPC treatment. It has been proposed that endogenous H 2 S can promote cancer cell growth and downregulation of H 2 S-producing enzyme can induce cancer cell death [8]. The effects of endogenous H 2 S on cancer development and progression have been further demonstrated [14][15][16][17][18][19]. Nevertheless, the mechanism of action of endogenous H 2 S on NPC growth remains unknown.
In this study, we determined the role of CSE in the proliferative, migrative, and invasive activities of NPC cells and detected the effect of CSE on NPC xenograft tumor growth. The function of DLpropargylglycine (PAG, an inhibitor of CSE) on NPC cell growth was further detected.

CSE is upregulated in human NPC cells
We firstly determined the levels of H 2 S in culture supernatant and human NPC cells. The concentrations of H 2 S in NPC cells and the supernatant were higher than those in normal nasopharyngeal cells and the supernatant (Fig. 1a). The protein level of CSE was dramatically upregulated in all human NPC cells compared to normal nasopharyngeal cells (Fig. 1b, c). The mRNA level of CSE showed a similar trend (Fig. 1d). To determine the role of CSE in NPC cell growth, overexpression/knockdown of CSE was performed. Transfection of CSE cDNA into CNE-1 and C666-1 cells enhanced the protein level of CSE and transfection of sh-CSE reduced the level of CSE (Fig. 1e). Both protein and mRNA levels of CSE showed similar trends ( Fig. 1f-h). Furthermore, CSE overexpression increased H 2 S level in both NPC cells and the supernatant, while CSE knockdown exhibited the reverse effects (Fig. S1). The above data show that CSE overexpression/knockdown experiment has been successfully performed.
CSE mediates the viability and proliferation of human NPC cells Compared to the Mock group, CSE overexpression promoted the proliferation of CNE-1 and C666-1 cells, whereas CSE knockdown exhibited oppose effects compared to the sh-Scb group (Fig. 2a, b). CSE exerted similar effects on the viability of human NPC cells (Fig.  2c). CSE overexpression enhanced colony formation and CSE knockdown reduced the colony number (Fig. 2d, e). The results of flow cytometric analysis showed that CSE overexpression induced an increased cell population in G2 phase and a decreased population in S phase, whereas CSE knockdown exhibited reverse trends (Fig. 2f, g). It has been shown that a number of cell cycle-related proteins are involved in the regulation of cell cycle progression, such as CDK2/4, cyclin D1/E1, p21, and p27 [20,21]. As shown in Fig. S2, CSE overexpression enhanced the expressions of cyclin D1/E1 and CDK2/ 4, but reduced the expressions of p21 and p27. CSE knockdown exerted reverse effects on the expressions of the proteins. In sum, these data indicate that CSE mediates the viability and proliferation of NPC cells by regulating cell cycle.

CSE mediates the migration and invasion of human NPC cells
We further evaluated the effects of CSE on the migration and invasion of human NPC cells. CSE overexpression increased the migration abilities of CNE-1 and C666-1 cells and CSE knockdown showed opposite effects (Fig. 3a, b). Overexpression of CSE promoted the anchorage-independent growth of CNE-1 and C666-1 cells, whereas the reverse effect was found in sh-CSE group (Fig. 3c, d). The migration/invasion ability of CNE-1 and C666-1 cells was increased in CSE group, whereas the sh-CSE group showed reverse effects ( Fig.  3e-h). To confirm the above results, western blot was conducted to detect the expressions of N-cadherin, E-cadherin, Vimentin, MMP-2, and MMP-9, which can play important roles in cellular migration and invasion [22,23]. The results showed that CSE overexpression reduced the expression of E-cadherin, but promoted the expressions of N-cadherin, Vimentin, MMP-2, and MMP-9. CSE knockdown exhibited reverse effects on the expressions of the proteins (Fig.  S3). The data together suggest that CSE regulates both migration and invasion of NPC cells.

CSE modulates mitochondrial apoptosis in human NPC cells
The apoptosis was reduced in CSE group compared to Mock group but enhanced in sh-CSE group compared to sh-Scb group (Fig. 4a, b). The Bcl-2 family proteins can act as effector molecules in regulating mitochondria-dependent apoptosis, such as Bax, Bad, Bcl-xl, and Bcl-2 [24]. The enhanced Bad/Bcl-xl and Bax/Bcl-2 ratios are key phenomena in mitochondrial apoptosis [25,26]. Cyt C can activate the caspase cascade once it is accumulated in cytosol [27]. Caspase-3 and caspase-9 are important members of the caspase family and induce cell apoptosis via the mitochondriamediated pathway [28]. PARP acts as a cleavage substrate for caspase-3 in the process of apoptosis [29]. CSE overexpression decreased Bad/Bcl-xl and Bax/Bcl-2 ratios and the expressions of Cyt C, cleaved cas-3, cleaved cas-9, and cleaved PARP, while CSE knockdown exhibited reverse effects (Fig. 4c, d). The data reveal that CSE can modulate mitochondria-dependent apoptosis in human NPC cells.
CSE modulates ROS-mediated mitogen-activated protein kinase (MAPK) and PI3K/AKT/mTOR pathways in human NPC cells ROS are widely regarded as oxygen-containing molecules, such as superoxide anion, hydrogen peroxide, and hydroxyl radical [30,31]. At low-to-moderate concentrations, ROS can serve as signaling molecules that activate stress-responsive survival pathways and promote cell proliferation and differentiation. Excessive production of ROS induces damages to cellular components, including lipids, DNA, and proteins [30,32]. The data suggested that CSE overexpression decreased ROS levels and CSE knockdown increased the levels of ROS in CNE-1 and C666-1 cells (Fig.  5a, b), suggesting that CSE can mediate the oxidative stress in human NPC cells. ROS can act as signaling messengers produced during various environmental stresses and play important roles in MAPK and PI3K/AKT/mTOR pathways [21,33]. These data suggested that CSE overexpression downregulated the phosphorylation levels of ERK, p38, and JNK, but upregulated the expressions of p-PI3K, p-AKT, and p-mTOR. However, CSE knockdown exhibited reverse effects on the expressions of the proteins ( Fig. 5c-f). Overall, these data suggest that CSE modulates ROSmediated MAPK and PI3K/AKT/mTOR cascades in human NPC cells.
CSE regulates human NPC xenograft tumor growth CNE-1 and C666-1 cells are used to establish subcutaneous xenograft tumors in nude mice [34,35]. The effects of CSE on NPC xenograft tumor growth were further determined. The tumors were then removed and photographed (Fig. 6a). We observed that CSE overexpression dramatically promoted xenograft tumor growth and CSE knockdown significantly decreased tumor growth ( Fig. 6b-h). As shown in Fig. S4a, b, CSE overexpression significantly increased the protein level of CSE and CSE knockdown exhibited reduced CSE level in tumor tissues. In addition, CSE overexpression increased the DT/DC index but decreased TVDT, CSE knockdown showed the reverse trends (Fig. S4c, d). However, no obvious difference was observed in body weight (Fig.  S4e, f). The proliferation, MVD, and p-AKT of human NPC xenograft tumors were increased in CSE group but decreased in sh-CSE group. Moreover, IHC with p-p38, p21, and cleaved cas-3 antibodies exhibited reverse trends (Fig. S5). In sum, the results show that CSE plays an important role in the regulation of NPC xenograft tumor growth.
CSE inhibitor suppresses human NPC cell growth PAG, a stereoselective compound, is one of the most commonly used pharmacological inhibitors of CSE [36][37][38]. The effects of PAG on the growth of human NPC cells were further determined. As Fig. 1 The expression levels of CSE, CBS, and 3-MST in human NPC cells were detected and CSE overexpression and knockdown experiments were performed. a The concentrations of H 2 S in cells and culture supernatant were determined. b, c Western blot was conducted to detect the protein levels of CSE, CBS, and 3-MST in NP69, CNE-1, CNE-2, HONE-1, and C666-1 cells. GAPDH was used as the loading control. d RT-PCR was performed to detect the mRNA levels of CSE, CBS, and 3-MST in NP69, CNE-1, CNE-2, HONE-1, and C666-1 cells. Data are presented as mean ± SEM of three independent experiments. *P < 0.05, **P < 0.01 compared with human normal nasopharyngeal epithelial cell line NP69. e Fluorescence microscopy of CSE in CNE-1 and C666-1 cells; original magnification ×100. f, g The protein expression of CSE was examined in CNE-1 and C666-1 cells by Western blot. GAPDH was used as the loading control. h RT-PCR was performed to detect the mRNA levels of CSE in CNE-1 and C666-1 cells. Data are presented as mean ± SEM of three independent experiments; **P < 0.01 compared with the Mock group; ## P < 0.01 compared with the sh-Scb group.

Fig. 2
Effects of CSE on the proliferation and viability of human NPC cells. a DNA replication activities of CNE-1 and C666-1 cells in each group were examined by EdU assay; original magnification ×100. b The proliferation rate of each group was analyzed. c The percentages of viable cells were determined using MTS and the cell viability of the control group was taken as 100%. d The clonogenic capacity was determined in CNE-1 and C666-1 cells. e The numbers of colonies were calculated. f Flow cytometry assay was used to determine cell cycle distribution. g Cell cycle distribution was analyzed. Data are presented as mean ± SEM of three independent experiments; *P < 0.05, **P < 0.01 compared with the Mock group; # P < 0.05, ## P < 0.01 compared with the sh-Scb group.    shown in Fig. S6, PAG reduced the protein level of CSE in a dosedependent manner. The concentrations of H 2 S in both cells and supernatant showed similar trends. We found that PAG dosedependently suppressed proliferation, viability, migration, and invasion of NPC cells (Figs. S7 and S8). In addition, PAG dosedependently increased ROS levels and the protein expressions of p-ERK, p-p38, and p-JNK, but decreased the protein levels of p-PI3K, p-AKT, and p-mTOR (Fig. S9). Furthermore, PAG suppressed CSE expression and human NPC xenograft tumor growth in a dose-dependent manner (Figs. 7a-d and S10). PAG dosedependently downregulated the proliferation, MVD, and p-AKT of human NPC xenograft tumors, but showed reverse effects on p-p38 and p21 expression levels and apoptotic index (Fig. 7e-g). However, there was no morphological difference of heart, liver, spleen, lung, kidney, and brain among groups. Moreover, no obvious difference in organ index and body weight was observed among groups (Fig. 8a-d). The data together suggest that CSE inhibitor could suppress human NPC cell growth via ROSmediated MAPK and PI3K/AKT/mTOR pathways without significant toxicity.

DISCUSSION
NPC, an epithelia-derived malignancy, has a very distinctive geographic distribution, including South China, Southeast Asia, and North Africa [9,10]. In recent years, the developments in application of chemotherapy, radiotherapy technology, and accurate disease staging have dramatically improved the treatment of NPC [39][40][41]. However, in light of local recurrence and distant metastasis, the prognosis of patients with NPC is unsatisfactory [41,42]. There is an urgent need for elucidating the underlying mechanism of NPC development for novel therapeutic strategies [43]. CSE is reported to be overexpressed in both PLC/PRF/5 and HepG2 hepatoma cells compared to normal liver cell line HL-7702 [44]. Another study indicates that CSE level in breast cancer tissue from lymph node metastatic patients is higher than that in breast cancer of lymph node nonemetastatic samples [17]. Furthermore, it has been revealed that CSE level is upregulated in bone-metastatic PC3 cells [45]. Similar to the previous studies, our data indicated that CSE level was dramatically upregulated in human NPC cells, suggesting that CSE can act as an important biomarker in NPC development. It has been revealed that CSE is involved in the growth of liver cancer [44], breast cancer [17], and prostate cancer [45]. Nevertheless, the effect of CSE on NPC growth remains unknown. CNE-1 and C666-1 cell lines have been widely used to determine the effect of different agent [34,35]. Then CNE-1 and C666-1 cells were adopted to detect the role of CSE in NPC cell growth. CSE contributes to the growth of hepatoma cells [44] and gastric cancer cells [46]. In addition, CSE promotes breast cancer metastasis through vascular endothelial growth factor signaling pathway [17], as well as enhances the progression and metastasis of prostate cancer via interleukin-1β/nuclear factor-kappa B (NF-κB) signaling pathways [45]. However, overexpression of CSE could induce spontaneous apoptosis of human melanoma cells via the suppression of NF-κB activity and inhibition of AKT and ERK pathways [47]. Another study indicates that CSE is involved in apoptosis induction in clear cell renal cell carcinoma [48]. These studies together suggest that CSE may exert different effects on tumor development depending on the cell types. Our data showed that CSE overexpression increased the viability, proliferation, migration, and invasion capabilities of NPC cells. Furthermore, overexpression of CSE enhanced the population of cells in G2 phase and reduced the cell percentage in S phase. However, CSE knockdown exhibited completely opposite effects. Collectively, the data indicate that CSE can play important roles in mediating the growth, migration, invasion, and cell cycle of NPC cells.
Apoptosis is responsible for tissue homeostasis and development in multi-cellular organisms [49]. Two main types of apoptosis exist in mammals: the death receptor and mitochondria-mediated pathways [50]. Cyt C release by the mitochondria is a key feature of mitochondria-mediated apoptosis [51]. Bax protein contributes to the transfer of cytochrome c across the mitochondrial membrane and then activates caspase-3 and caspase-9 [27,52]. PARP is further proteolytically cleaved by caspase-3 and can result in the occurrence of apoptosis [53]. CSE may play a role in AGS gastric cancer cell proliferation probably via antiapoptotic effects [46]. Furthermore, inhibition of endogenous CSE/H 2 S pathway could significantly enhance mitochondrial disruption and further induce DNA damage and apoptosis [44]. Similarly, our results suggested that CSE overexpression reduced both Bad/Bcl-xl and Bax/Bcl-2 ratios and the expressions of Cyt C, cleaved cas-3, cleaved cas-9, and cleaved PARP, whereas CSE knockdown exhibited reverse effects. Thus, we can conclude that CSE modulates mitochondria-dependent apoptosis in human NPC cells.
Considering different levels of ROS could cause many biological responses, the regulation of ROS levels is essential in cellular homeostasis [30]. Low to moderate levels of ROS contribute to cancer progression either by promoting genomic DNA mutation or acting as signaling molecules. In contrast, high levels of ROS could induce cellular damage and promote cancer cell death [30,32]. Our data indicated that CSE overexpression decreased the ROS levels and CSE knockdown promoted ROS generation in human NPC cells. Many studies have revealed that MAPK pathway can be regulated by ROS in cancer cells [33,54]. PI3K/AKT/mTOR pathway is dysregulated in cancer cells and ROS can act as upstream regulators of the pathway [24,55]. Furthermore, it has been reported that both of these pathways are involved in several types of cancer, such as prostate cancer [56], hepatocellular carcinoma [57], and gastric cancer [58], suggesting that these two pathways may play synergistic effects in the development of cancer. Moreover, H 2 S can promote autophagy and induce apoptosis in human hepatocellular carcinoma cells and melanoma cells via inhibition of the PI3K/Akt/mTOR pathway [59,60]. However, H 2 S exhibits anti-cancer effects by triggering apoptosis though activation of p38 MAPK pathway in rat C6 glioma cells [61]. Another study indicates that H 2 S dose-dependently decreases the viability of human prostate cancer PC-3 cells via activation of p38 MAPK and JNK [62]. We found that CSE overexpression decreased the expressions of p-ERK, p-JNK, and p-p38, but enhanced the expressions of p-PI3K, p-AKT, and p-mTOR, while CSE knockdown showed reverse effects on the expressions of the proteins. Our data reveal that CSE can modulate ROS-mediated MAPK and PI3K/ AKT/mTOR cascades in NPC cells.
Many studies have indicated that CNE-1 and C666-1 cells can be successfully used to establish xenograft tumor models [34,35]. We then examined the effect of CSE on NPC xenograft tumor growth. CSE overexpression significantly enhanced NPC xenograft tumor growth, while CSE knockdown dramatically decreased tumor growth. Ki67, a cell cycle-related protein, is a key marker in detecting cancer cell proliferation [24,[63][64][65]. The density of CD31 has been regarded as the tumor MVD [66,67]. p21 plays an important role in cell cycle arrest in many types of cancer [68,69]. Cleaved caspase-3 exerts the central effect on the progression of apoptosis [21,70]. In this study, CSE overexpression increased the proliferation index and MVD, but reduced the ratio of p21 positive cells and the apoptotic index. However, CSE knockdown exhibited completely reverse trends. The data together indicate that CSE can regulate NPC xenograft tumor growth by mediating proliferation, angiogenesis, cell cycle, and apoptosis.
PAG, a selective CSE inhibitor, could suppress the progression of different types of cancer [36]. It has been reported that pretreatment with PAG drastically reduces the migration of tumor-derived endothelial cells [71]. Another study indicates that PAG concentration dependently suppresses the growth of AGS human gastric cancer cells [46]. In addition, PAG could decrease the proliferation and migration of SW480 human colon cancer cells and reduce tumor xenograft growth [72]. Furthermore, combination of 3,3′-diindolylmethane with PAG synergistically inhibits proliferation and migration but increases apoptosis in human gastric cancer cells [73]. In according to previous findings, our data suggested that PAG dose-dependently suppressed the viability, proliferation, migration, and invasion of NPC cells via ROS-mediated MAPK and PI3K/AKT/mTOR pathways. Moreover, no obvious difference was found in body weight, relative organ weight, and morphologies of heart, liver, spleen, lung, kidney, and brain among groups, suggesting no obvious systemic toxicity. Therefore, PAG can be used to inhibit NPC cell growth without significant toxicity.

CONCLUSIONS
In conclusion, this study reveals that CSE mediates the proliferation, migration, and invasion of NPC cells via ROS-mediated MAPK and PI3K/AKT/mTOR pathways (Fig. 8e). In light of its role in the progression of human NPC cells, CSE could act as a potential biomarker for the diagnosis and prognosis in NPC patients. In addition, CSE may be an important therapeutic target and novel donors/drugs that inhibit the expression/activity of CSE can be developed in the treatment of NPC.

Cell viability and proliferation assays
The viability was determined with the CellTiter 96 AQueous one solution assay kit (Promega, Madison, WI, USA). 5-Ethynyl-2′-deoxyuridine (EdU) assay was carried out by using the cell-light EdU apollo 567 in vitro imaging kit (RiboBio, Guangzhou, Guangdong, China) to determine cell proliferation. The cells were observed and counted as the ratio of positive cells to total cells [63].

Colony formation assay
The cells were cultured at 37°C in 6 well plates (1 × 10 3 /well). After 2 weeks, the colonies were fixed with methanol and stained with crystal violet. Then the plates were scanned and the colony number was counted [24].

Flow cytometry
The cells were trypsinized, fixed in 70% ice-cold ethanol, and incubated with propidium iodide and RNase A for 30 min at room temperature. The cell cycle was detected using the FACSVerse flow cytometer (BD, San Jose, CA, USA).

Scratch assay
A sterile 200 μL pipette tip was used to scrape the confluent cells. The migration was photographed using a CKX41 inverted microscope (Olympus, Tokyo, Japan) and calculated using ImageJ software (NIH, Bethesda, MD, USA). The migration rate (MR) was analyzed using formula as below: MR = (A − B)/A × 100%, where A and B is the width at 0 h and 24 h, respectively [64].

Soft agar assay
Cells were cultured in the medium with 10% FBS and 0.6% agarose. Then the mixture was overlaid into a basal layer of 1% agarose in 6 well plate (1 × 10 4 /well). After 14 days, the colonies were photographed and counted under an Olympus CKX41 inverted microscope [63].

Transwell assay
Transwell assay was conducted as previously described [65]. The number of stained cells was analyzed by a Zeiss Axioskop 2 microscope (Thornwood, NY, USA).
TdT-mediated dUTP-biotin nick end labeling (TUNEL) assay TUNEL was carried out using the in situ cell death detection kit (Beyotime, Haimen, Jiangsu, China). Apoptotic cells were photographed using a Nikon Eclipse Ti fluorescent microscope. The percentages of positive cells to total cells were analyzed using ImageJ software.

Reactive oxygen species (ROS) detection
The measurement of cellular ROS was performed using the ROS detection assay kit (Beyotime).

Western blot
Western blot analysis was conducted to determine the proteins levels.

Animal study
Animal experiments were approved by the Committee of Medical Ethics and Welfare for Experimental Animals of Henan University School of Medicine (HUSOM-2017-191). Animal studies were carried out according to a previously reported method with slight modifications [63]. BALB/C nude mice (n = 6/group, 4-week-old, male) were obtained from Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). 2 × 10 6 CNE-1 and C666-1 cells with overexpression/knockdown of CSE were inoculated subcutaneously into the right flanks of nude mice. For the CSE inhibitor experiment, PAG (5, 10, 20, and 40 mg/kg/day) was administered subcutaneously for 4 weeks. The control group was treated with PBS for 4 weeks. The tumor volume and body weight of nude mice were monitored daily. The tumor volume was measured: volume (V) = 1/2 × L × W 2 , where L is the longest dimension parallel to the skin surface and W is the dimension perpendicular to L [24]. The parameter of DT/DC (%) was measured, where DT = T − Do and DC = C − Do (T/C represents the volume of the treated/untreated tumor; Do represents the initial average tumor volume) [65]. The tumor volume doubling time (TVDT) was measured as TVDT = (T -T 0 ) × log2/log(V2/V1), where (T -T 0 ) is the time interval and V2/V1 respectively represents the tumor volume at two measurement time [64]. Tumor growth was weekly monitored by bioluminescent imaging (IVIS® Lumina III In Vivo Imaging System, PerkinElmer, Hopkinton, MA, USA). Then the tumors were removed, weighted, and imaged. The inhibition rate (IR) = (A − B)/A × 100%, where A/B is the average tumor weight of the control/treatment group [24]. The value for the blank was subtracted from both the samples and standard controls. A calibration curve was plotted relating the concentration of each standard solution on the X axis to the OD value on the Y axis. The standard curve linear regression equation was created and the H 2 S concentration was calculated from the standard curve.

Hematoxylin and eosin (HE) staining
Tumor tissues were routinely formalin-fixed, paraffin-embedded, 4 µM sectioned, and stained with HE. The results were photographed using a Zeiss Axioskop 2 microscope.

Immunohistochemistry (IHC)
Cluster of differentiation 31 (CD31) is a key biomarker for vascular endothelial cells. Its density has been regarded as microvessel density (MVD) [66]. Tumor tissues were respectively stained with anti-CD31 (CST), anti-Ki67 (CST), anti-p-AKT, anti-p-p38, anti-p21, and anti-cleaved cas-3 antibody. The results were photographed using a Zeiss Axioskop 2 microscope. Then the MVD was measured and the proliferation index, apoptotic index, p-AKT positive cells, p-p38 positive cells, and p21 positive cells were determined by the ratios of positive cells to total cells.

Statistical analysis
All data are expressed as mean and standard error of the mean (SEM). The differences among groups were analyzed using one-way analysis of variance using SPSS statistical software followed by Tukey's test. P < 0.05 was regarded as statistically significant.

DATA AVAILABILITY
The data of the study are available from the corresponding author on reasonable request.