Inhibitory effects of LOXL2 knockdown on cellular functions of liver cancer stem cells

Background Lysyl oxidase-like 2 (LOXL2) plays a role in tumor microenvironment formation and metastasis of hepatocellular carcinoma (HCC), which has a high mortality burden. Liver cancer stem cells (LCSCs) are related with the major malignant phenotypes of HCC. The function of LOXL2 in regulation of LCSCs remains unknown. Methods CD133+HepG2 and CD133+Hep3B cells were sorted by fluorescence-activated cell sorting (FACS) from two human hepatoblastoma cell lines. Spheroid formation, apoptosis, cell cycle, as well as transwell assays were performed upon LOXL2 knockdown in CD133+HepG2 and CD133+Hep3B cells. Protein and mRNA levels were quantified by Western blotting, immunofluorescence and reverse transcription-PCR (RT-PCR). Results Knockdown of LOXL2 decreased spheroid formation, migration and invasion (P<0.05), also induced apoptosis (P<0.05) and cell cycle arrest (P<0.05) in CD133+HepG2 and CD133+Hep3B cells. Knockdown of LOXL2 effectively inhibited expression of the anti-apoptosis proteins baculoviral inhibitor of apoptosis protein (IAP) repeat-containing 3 (BIRC3) and murine double minute 2 (MDM2) (P<0.01), as well as autophagy marker microtubule-associated protein 1 light chain 3 B (LC3B) and autophagy gene ATG5 in CD133+HepG2 and CD133+Hep3B cells (P<0.01). Conclusions The results revealed that LOXL2 inhibition could reduce the proliferation and expansion of LCSCs, making LOXL2 inhibitors an attractive and novel therapeutic strategy of HCC.


Introduction
Hepatocellular carcinoma (HCC) is very common and has a high mortality burden [1,2] . Hepatitis B and C, metabolic dysfunction associated fatty liver disease and alcohol intake are the main risk factors of HCC [2,3] . Hepatocarcinogenesis is accompanied by resistance to cellular death signals, metabolic changes, and a failure to clear damaged precancerous hepatocytes by the immune system [4] . The poor prognosis and response to therapy of HCC is in part due to the presence of CSCs, an immune-privileged cell population that can evade immune surveillance more effectively than non-CSCs [5,6] . Liver cancer stem cells (LCSCs) are a subset of hepatocarcinoma cells with pluripotent and self-renew properties related to the major malignant phenotypes of HCC [3,7,8] . The expression of CSC-related genes such as NANOG, SOX2 and OCT4 can be promoted by oncogenes and/or overexpression of oncoproteins such as the HBV envelope protein PreS1, thereby driving the malignant biological behaviors of LCSCs [9] . LCSCs can be characterized through cell surface markers, such as CD133 [10,11] . CD133 is a transmembrane cell surface single chain glycoprotein, which is the original marker of hematopoietic stem cells, neural stem cells and liver stem / progenitor cells. It is an important indicator of malignant progression, patient survival and recurrence rate. The expression of CD133 in tumor cells is related to the stem cell-like characteristics [12~14] .
Lysyl oxidase-like 2 (LOXL2) belongs to the lysyl oxidase (LOX) family, members of which are implicated in diverse pathophysiology including developmental regulation, cell motility, cellular senescence, and tumor suppression or promotion [15~17] . The roles played by LOXL2 in brosis, tumorigenesis, and metastasis have been extensively reported, and it has also been recognized as a therapeutic target [16,18,19] . Overexpression of LOXL2 is related with poor prognosis of patients with malignant tumors, such as colon cancer, esophageal cancer, and squamous cell carcinoma [15,20,21] . LOXL2 can induce epithelial to mesenchymal transition (EMT), its overexpression negatively affects the clinicopathological features of different tumor types [22] . In HCC, LOXL2 is upregulated and plays a key role in HCC metastasis by mediating EMT and extracellular matrix remodeling, subsequently enhancing HCC invasion [22,23] . LOXL2 is also crucial in tumor microenvironment formation and metastatic niche in HCC [23,24] . However, the functions and regulatory mechanism of LOXL2 in LCSCs remains unknown.
In this study, we sorted the CD133 + HepG2 and CD133 + Hep3B cells from two human hepatoblastoma cell lines. The cellular function of the two cell lines, including proliferation, migration, invasion as well as apoptosis and cell cycle assay, were demonstrated upon LOXL2 gene knock down. To understand whether LOXL2 bene ted LCSCs from escaping apoptotic cell death, we investigated the change of gene expression of anti-apoptosis proteins baculoviral IAP repeat-containing 3 (BIRC3, also known as cIAP2) and murine double minute 2 (MDM2) in CD133 + HepG2 and CD133 + Hep3B cells upon LOXL2 knockdown. Moreover, to explore whether LOXL2 was related with autophagic cell death, we investigated the expression change of autophagy marker microtubule-associated protein 1 light chain 3 (LC3) B and autophagy gene ATG5 in CD133 + HepG2 and CD133 + Hep3B cells upon LOXL2 gene knock down.

Cell lines
The human hepatoblastoma cell lines HepG2 and Hep3B were purchased from Chinese Academy of Sciences (Shanghai, China). Cells were cultured in RPMI 1640 medium (Gibco, Thermo Fisher Scienti c, Waltham, MA, USA) with 10% fetal bovine serum (FBS; PAN, Germany) and 1% penicillin and streptomycin (Gibco) at 37.0°C in 5% CO 2 .
In vitro LOXL2 gene silencing LOXL2 gene in CD133 + HepG2 cells and CD133 + Hep3B cells were stable knockout using siRNA lentiviral vector (LV; GeneChem, Shanghai, China), the control group were transfected with siRNA negative control lentiviral vector. The vectors were encoded with green uorescent protein (GFP). Cells were transduced with 25 multiplicity of infection (MOI) lentiviral particles, and HiTransG P (GeneChem, Shanghai, China) was used to promote transfer e ciency. After 48 to 72 hours, the effect of cell infection was observed under uorescence microscope to make sure that more than 80% of cells show GFP. Cells were transduced for 72 h before treated with 2 µg/mL puromycin for 2 weeks. Transfection e ciency was con rmed by RT-PCR and western blotting.
After stable knockout of LOXL2 by transfected with LOXL2 siRNA lentivirus, the CD133 + Hep3B cell line presented with growth retardation, even died. CD133 + Hep3B cell line was more susceptible to LOXL2 gene deletion. Thus, we adjusted to transient transfection of CD133 + Hep3B cell line by siRNA in the follow-up study after spheroid formation assay. Pre-validated Silencer® select siRNAs against LOXL2, negative control siRNA, and GAPDH siRNA (Thermo Fisher Scienti c) were transfected respectively into 5 × 10 5 CD133 + Hep3B cells at a dose of 25 pmol in six-well plates using the Lipofectamine RNAiMAX™ transfection reagent (Invitrogen, Waltham, MA, USA) for 24 h.

Quantitative real-time PCR
Total RNA was extracted from CD133 + HepG2 cells and CD133 + Hep3B cells. Then, cDNA was synthetized by reverse transcription with the Evo M-MLV RT Kit for qPCR (AG Bio, China). PCR was performed by Applied Biosystems 7500 PCR System (Thermo Fisher Scienti c) with SYBR® Green Premix Pro Taq HS qPCR Kit (AG Bio). GAPDH was taken as the internal control. The ampli cation conditions were as follows: pre-denaturation (95℃, 30s), denaturation (95℃, 5s), annealing and extension (60℃, 30s), all for a total 40 cycles. The experimental Ct values were normalized against GAPDH. 2 −ΔΔCT method was used to analyze the data.
Then, the PVDF membrane was blocked with 5% buffer bovine serum albumin in TBST and exposed to primary antibody at 4°C overnight. Next, the PVDF membrane was washed in TBST (10 min × 3 times) and exposed to secondary antibody (1:5000) at room temperature for 1-1.5 h. The band was incubated in BeyoECL Plus (Beyotime China) and detected by chemiluminescence.
Flow cytometric apoptosis analysis Cells in logarithmic growth phase were collected and digested into single ones by ACCUTASE (Sigma-Aldrichl). 1×10 5 ~ 1×10 6 single cells were washed with cold PBS and 1x binding buffer, respectively. Next, cells were resuspended in 100µL binding buffer with 5µL Annexin V-FITC and 10µL PI staining solution, and incubated in the dark for 15 min at room temperature. Then, apoptosis was analyzed with a FACS Calibur instrument.

Flow cytometry cell cycle analysis
Cells were collected and digested into single ones by ACCUTASE (Sigma-Aldrichl). Then, the cells were washed with cold PBS, and xed with 70% ethanol at 4°C overnight. After being washed, cells were incubated with 2mL staining solution (propidium iodide) in the dark for 30 min at room temperature. Then, cell cycle was analyzed with a FACSCalibur instrument (BD Biosciences).

Transwell assay for migration and invasion
Cells were starved for 24 hours, and suspended with RPMI 1640 medium. 5 x 10 4 cells in 200 µL RPMI 1640 medium were added into the upper chambers of 8 µm pore size transwells. 700 µL medium with 10% FBS were added into the lower ones. For the invasion assay, the bottom of the transwell chamber was laid with diluted Matrigel (BD Biosciences). The cells in transwell were cultured at 37℃ in 5% CO 2 for 24 hours. Next, the transwell was treated with 4% paraformaldehyde for 30 min, and 0.1% crystal violet for 25 min. After PBS washing, cells in the upper chambers were wiped off, and cells at the bottom of the chamber were observed under the microscope, and ve elds were randomly selected for counting (×100 objective).

Immuno uorescence assay
Cells were xed in 4% paraformaldehyde for 30 min and permeabilized in 0.25% Triton X-100 for 15 min.
After blocked by 5% goat serum at room temperature for 45 min, cells were exposed to primary antibody (1:500; Abcam, Cambridge, UK) at 4°C overnight. Next, cells were exposed to uorescein (FITC)conjugated goat anti-rabbit antibody (1:250; Abbkine, China) in the dark for 1 h at room temperature. Fluorescence images were acquired by a uorescence microscope (Olympus, Tokyo, Japan).

Statistical analysis
Statistical analysis was performed in Prism 8 software (GraphPad Software, La Jolla, CA). Data of continuous variables are shown as mean ± standard deviation (S.D). The difference between two independent groups was analyzed with two-sided unpaired Student's t-test. p-value <0.05 is regarded as statistically signi cant.

Knock down of LOXL2 impairs LCSCs spheroid formation
To understand the effect of LOXL2 knockdown on the cellular functions of LCSCs, LOXL2 or control siRNA lentiviral vectors were transferred into CD133 + HepG2 and CD133 + Hep3B cell lines. Results of Western blot and PCR showed that LOXL2 was effectively knocked down in the two cell lines (Fig. 1a, b). The ability of colony formation depends on the replication immortality of tumor cells, which is one of the hallmarks of tumor [25] . To explore whether knock down of LOXL2 gene could impair tumor spheroid formation of LCSCs, CD133 + HepG2 and CD133 + Hep3B cells were digested into single cells and cultured in SFC medium for 1-2 weeks. For CD133 + HepG2 cells, spheroid formation was signi cantly decreased upon silencing LOXL2 (Fig. 1c). When switching to CD133 + Hep3B cells, a more signi cant inhibition effect was observed (Fig. 1d). CD133 + Hep3B cell line was more susceptible to LOXL2 gene deletion. After stable knockout of LOXL2 by transfected with LOXL2 siRNA lentivirus, the growth of CD133 + Hep3B cell line signi cantly slowed down, stagnated or even died. These results indicate that LOXL2 knockdown impairs LCSCs spheroid formation.

Knock down of LOXL2 inhibits LCSCs migration, and invasion
As LOXL2 gene plays a key role in HCC metastasis, subsequently enhancing HCC invasion, we asked whether silencing LOXL2 could inhibit LCSCs expansion. The transwell migration and invasion assays were performed in CD133 + HepG2 (Fig. 2a, b) and CD133 + Hep3B cells (Fig. 2c, d) which subjected to siRNA-mediated LOXL2 knockdown. The migration and invasion effect of both two cell lines were decreased upon LOXL2 knockdown compared to the controls. It indicates that LOXL2 knockdown inhibits LCSCs migration, and invasion.

Knock down of LOXL2 induces apoptosis and cell cycle arrest in LCSCs
Anti-apoptosis is one of the signs of malignant tumors. We assessed apoptosis of CD133 + HepG2 and CD133 + Hep3B cells by ow cytometry using Annexin V-FITC and PI. The results showed that knock down of LOXL2 signi cantly increased the proportion of cell apoptosis in both CD133 + HepG2 (Fig. 3a) and CD133 + Hep3B cells (Fig. 3b). Furthermore, we explored whether knock down of LOXL2 could also affect cell cycle progression in the two cell lines. We found that the proportion of CD133 + HepG2 cell decreased in G0/G1 phase, increased in G2/M phase, and was unaltered in S phase upon LOXL2 knockdown (Fig. 3c). Correspondingly, the proportion of CD133 + Hep3B cells increased in G0/G1 phase, decreased in S phase, and did not change in G2/M phase (Fig. 3d). These results indicate that knock down of LOXL2 makes CD133 + HepG2 cell cycle arrest in G2/M phase, and CD133 + Hep3B cells arrest in G0/G1 phase. LOXL2 knockdown induces the apoptotic phenotype and regulates the cell cycle of LCSCs.

Knock down of LOXL2 inhibits the expression of the anti-apoptosis proteins BIRC3 and MDM2 in LCSCs
The activation of anti-apoptosis proteins bene ts tumor cells from escaping apoptotic cell death [4,26] . To understand whether LOXL2 favored the survival of LCSCs by anti-apoptosis, we explored the expression of BIRC3 and MDM2 in CD133 + HepG2 and CD133 + Hep3B cells upon LOXL2 knockdown. The results showed that LOXL2 knockdown signi cantly decreased protein and mRNA expression of BIRC3 and MDM2 in the two cell lines compared to controls (Fig. 4a-d). Immuno uorescence analysis of CD133 + HepG2 cells was consistent with the western blotting and RT-PCR results, with reduced level of BIRC3 and MDM2 upon LOXL2 knockdown (Fig. 4e, f). These results indicate that one of the molecular mechanisms of apoptosis induced by LOXL2 knockdown in LCSCs may lies in the downregulation of anti-apoptosis protein BIRC3 and MDM2.

Knock down of LOXL2 inhibits the expression of autophagy marker LC3B and autophagy gene ATG5 in LCSCs
As to CD133 + Hep3B cells proliferated signi cantly slow or even died after stable knockout of LOXL2, we wondered whether LOXL2 took a part in regulation of autophagic cell death of LCSCs besides antiapoptosis. So, we further explored the expression of autophagy marker LC3B and autophagy gene ATG5 in CD133 + HepG2 and CD133 + Hep3B cells upon LOXL2 gene knock down. The results showed that LC3B and ATG5 expression at protein and mRNA level were both signi cantly decreased in the two cell lines upon LOXL2 knockdown (Fig. 5a-d). Also, it was found using immuno uorescence that inhibition of LOXL2 reduced the expression of LC3B and ATG5 in CD133 + HepG2 cells. (Fig. 5e, f). These results indicate that LOXL2 knockdown inhibits the expression of autophagy marker LC3B and autophagy gene ATG5 in LCSCs.

Discussion
LCSCs, a subset of hepatocarcinoma cells with CSC-like properties, are an important therapeutic target in HCC [3,27] . Here we show that knock down of LOXL2 inhibits the proliferation, invasion and migration of CD133 + HepG2 and CD133 + Hep3B cells, and induces cell apoptosis and cell cycle arrest. This is consistent with the conclusions of Wong et al., they demonstrated that LOXL2 facilitates HCC metastasis by inducing bone marrow-derived cells recruitment into the metastatic site [23] . These results indicated that LOXL2 inhibition could reduce the proliferation and expansion of LCSCs. Targeting LCSCs by LOXL2 inhibitors may become an attractive and novel therapeutic strategy in HCC.
We tried to explore the potential mechanisms of LOXL2 inhibiting LCSCs function. In the current study, the growth of CD133 + HepG2 and CD133 + Hep3B cells signi cantly slowed down after stable knockout of LOXL2. The CD133 + Hep3B cell was even stagnant or dead. Apparently, knock down of LOXL2 may lead to LCSCs death via some mechanism. Cell death is subdivided into apoptosis, autophagy death and necrosis [28] . The activation of anti-apoptosis proteins bene ts tumor cells from escaping apoptotic cell death [4,26] . We rst investigated whether knock down of LOXL2 blunted the anti-apoptosis mechanism of LCSCs. We found that knock down of LOXL2 effectively downregulated the anti-apoptosis proteins BIRC3 and MDM2 expression in CD133 + HepG2 and CD133 + Hep3B cells. This is consistent with our previous nding by microarray that BIRC3 and MDM2 are possible LOXL2 downstream genes in SMMC-7721 human hepatoblastoma cell lines [29] . BIRC3, one of the human inhibitor of apoptosis protein (IAP) family, is often overexpressed in HCC tissues and indicates a poor prognosis of HCC patients [30~32] . BIRC3 also induces EMT, proliferation, migration of hepatoblastoma cells in vitro [32] . Moreover, MDM2 is an upregulator of XIAP, which also belongs to the IAP family. It favors resistance of neoplastic cells to irradiation-induced apoptosis [33,34] . MDM2 is also a negative-regulatory factor of p53, which is a central player in apoptosis [33,35~37] . We speculate that LOXL2 might bene t LCSCs from escaping apoptotic cell death by regulating BIRC3 and MDM2. Knockdown of LOXL2 could induce LCSCs apoptosis through downregulation of BIRC3 and MDM2, which may become a promising strategy of targeting LCSCs.
Another mechanisms of LOXL2 inhibiting LCSCs function is that LOXL2 might play a role in regulation of LCSCs autophagy. We investigated the expression change of autophagy marker LC3B and autophagy gene ATG5 upon LOXL2 gene knock down to understand whether LOXL2 was related with autophagic cell death. Our results show that knock down of LOXL2 downregulated the expression of LC3B and ATG5 in CD133 + HepG2 and CD133 + Hep3B cells. This phenomenon could be explained by the previous reports that autophagy favors LCSCs resistance to hypoxia and nutrient de ciency in the tumor microenvironment of HCC [38~40] . Autophagy positively regulates LCSCs by suppressing p53, which would normally be upregulated by mitochondrial PINK1 to downregulate CSC-related NANOG [41] . Moreover, BIRC3 was upregulated in the active autophagy response and inhibition of autophagy suppressed the expression of BIRC2 and BIRC3 [42,43] . Thus, we speculate that LOXL2 might take a part in regulation of LCSCs autophagy, favoring LCSCs survival under cellular stress and hypoxia.
However, the speci c mechanism of LOXL2 in LCSCs is not well established. The relationship between LOXL2, apoptosis and autophagy need to be further explored. Moreover, BIRC3, MDM2 and other possible LOXL2 downstream genes need further investigations, such as p53 which is known associated with MDM2 and autophagy.
In summary, knock down of LOXL2 inhibits the proliferation, invasion and migration of LCSCs, and induces cell apoptosis and cell cycle arrest. Knockdown of LOXL2 effectively inhibited the expression of anti-apoptosis proteins BIRC3 and MDM2, as well as autophagy marker LC3B and autophagy gene ATG5 in LCSCs. LOXL2 inhibition could reduce the proliferation and expansion of LCSCs, making LOXL2 inhibitors an attractive and novel therapeutic strategy of HCC.