Down-regulation of EGFL8 regulates migration, invasion and apoptosis of hepatocellular carcinoma through activating Notch signaling pathway

Our previous studies have reported the down-regulation of EGFL8 correlates to the development and prognosis of colorectal and gastric cancer. The present study is carried out to explore the expression pattern and role of EGFL8 in hepatocellular carcinoma (HCC). EGFL8 expression in 102 cases of HCC tissues matched with adjacent non-tumorous liver tissues, a normal liver cell line and three liver cancer cell lines with different metastatic capacity was detected by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and Western blot. Moreover, the clinicopathological features and prognosis of HCC patients were correlated with expression of EGFL8. Subsequently, the gain-and loss-of-function experiments were carried out to investigate the biological function of EGFL8 in HCC. We also used N-[N-(3,5-Difluorophenacetyl-L-alanyl)]-(S)- phenylglycine t-butyl ester (DAPT), an inhibitor for Notch signaling pathway, in these experiments to verify the involvement of Notch signaling pathway in the effects of EGFL8. Additionally, a mouse model was established to investigate the effect of EGFL8 on metastasis of HCC cells. The expression of Notch signaling pathway in HCC cells and xenograft mouse tumors were detected by Western blot and immunohistochemistory. The expression of EGFL8 was significantly decreased in HCC tissues and cell lines and EGFL8 down-regulation correlated to multiple nodules, vein invasion, high TNM stage and poor prognosis of HCC. Interestingly, the expression levels of EGFL8 in three liver cancer cell lines were negatively associated with their metastatic capacity. In vitro and in vivo experiments indicated that EGFL8 obviously suppressed metastasis and invasion of HCC cells but slightly promoted apoptosis. Meanwhile, the expression of Notch signaling pathway was obviously suppressed in EGFL8 overexpressed HCCLM3 cells and xenograft mouse tumors generated from these cells but markedly elevated in EGFL8 depleted Hep3B cells. Furthermore, the up-regulated expression of Notch signaling pathway and effects induced by EGFL8 knockdown in Hep3B cells could be counteracted by DAPT treatment. The down-regulation of EGFL8 was correlated to progression and poor prognosis of HCC and regulates HCC cell migration, invasion and apoptosis through activating the Notch signaling pathway, suggesting EGFL8 as a novel therapeutic target and a potential prognostic marker for HCC.


Introduction
As the sixth most common cancer in globe, hepatocellular carcinoma (HCC) has become the second leading cause of male cancer death in developing countries, second only to lung cancer [1]. Despite the great improvements in treatments, the long-term survival of patients with HCC remains unsatisfactory, with a 5-year survival rate of less than 20%, mainly due to a high metastasis rate after operation [2][3][4]. Therefore, numerous researches have been carried out to uncover the molecular mechanisms underlying the metastasis of HCC [5]. And quite a lot of signaling pathways have been found to be involved in the metastasis of HCC including Ras/Raf/ MAPK, Notch, HGF/c-Met, Wnt/β-catenin and so on [4][5][6][7][8]. Some other mechanism including autophagy and epigenetics modulation has also been implicated in this pathological process [9,10]. These molecular studies have resulted in the clinical application of sorafenib, a multikinase inhibitor, however, the benefits obtained from sorafenib are very limited. Further researches are still needed to identify potential therapeutic targets for interfering the metastasis of HCC [5,6,8].
In a recent study, we showed the upregulation of epidermal growth factor-like domain 7 (EGFL7), a critical gene in vascular tube formation during embryogenesis [11], in HCC tissue samples and elucidated a novel role of EGFL7/FAK/ EGFR signaling pathway in metastasis of HCC, which firstly evidenced the involvement of a member of EGFL gene family in HCC [12]. As the only know paralog of EGFL7, epidermal growth factor-like domain 8 (EGFL8) shares the same overall domain structure with EGFL7 such as an EGF-like domain, a Ca 2+ binding EGF-like domain, and a N-terminal signal peptide [11,13], which led us to hypothesize that EGFL8 may also be involved in human cancers just like EGFL7. In fact, our subsequent studies did indicate that the expression of EGFL8 was significantly decreased in colorectal and gastric cancer tissues and this decrease correlated significantly to the progression and prognosis of these two malignancies [14,15], suggesting EGFL8 as a novel biomarker for gastroenterological cancers. However, the expression level and biological characteristics of EGFL8 in HCC still remain unclear.
Therefore, the present study was carried out to detect the expression pattern of EGFL8 in human HCC tissues and explore the role of EGFL8 in the development of HCC in vitro as well as in vivo.

Materials and methods
Patients and follow-up HCC and the corresponding adjacent non-tumorous liver tissue (ANLT) specimens were collected from 102 cases of HCC patients who underwent operation at Guangzhou Red Cross Hospital, Jinan University from February 2012 to December 2019. All tissue samples were obtained and frozen in liquid nitrogen immediately after operation and then transferred to an ultra-low temperature freezer (Meling Biology & Medical, Hefei, China) and stored at − 80°C until detection. The diagnoses of HCC were confirmed by histopathological observation. The clinicopathologic data of the patients including age, gender, etiologies, liver cirrhosis status, serum α-fetoprotein (AFP) level, maximal tumor size, tumor cell differentiation (Edmondson-Steiner grade), tumor nodule number, capsular formation, vein invasion (including portal vein invasion, venous invasion or microscopic vessel invasion), and tumor node metastasis (TNM) stage were also collected. Follow-up data were obtained after hepatic resection for all 102 patients. The follow-up period was defined as the interval between the date of operation and that of patient's death or the last follow-up. Deaths from other causes were treated as censored cases. Recurrence and metastasis were diagnosed by clinical examination, serial AFP level mensuration, and ultrasonography or computed tomography (CT) scan. Prior written informed consent of all patients involved in the study was obtained before operation and the protocol of the present study was approved by the Ethics Committee of Guangzhou Red Cross Hospital.
RNA preparation and reverse transcription-quantitative polymerase chain reaction (RT-qPCR) The total RNA of HCC tissues, adjacent non-tumorous liver tissues (ANLTs), liver cancer cell lines was extracted by using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) as previously described [14][15][16]. The protocol of the first-strand cDNA generation and was also performed as described previously [16]. The quantitative polymerase chain reaction (qPCR) was consisted of 40 cycles after an initial denaturation step (95°C for 3 min) and every cycle including 95°C for 5 s and then 60°C for 30 s. As an internal control, β-actin gene in the same samples was also detected. The primers for EGFL8 and β-actin qPCR amplification were designed by Primer Premier 5.0 software (Premier Biosoft International, Palo Alto, CA, USA). EGFL8: forward, 5′-CCCGCTCCAC TACAACGAGT-3′; reverse, 5′-AACGCGGTACATGG TCCTGT-3′. Beta-actin: forward, 5′-GCATGGGT CAGAAGGATTCCT-3′; reverse, 5′-TCG-TCCCAG TTGGTGACGAT-3′. All the qPCR analyses were performed in triplicate and the results were calculated by 2 -ΔΔCt method. The relative expression of EGFL8 mRNA in tissue samples and liver cancer cell lines were normalized to β-actin and the EGFL8 expression was defined as down-regulation when the relative expression of EGFL8 in HCC tissue < the relative expression of EGFL8 in the corresponding ANLT tissue. Egfl8 expression in HCC specimens were also divided into high EGFL8 expression group (EGFL8 expression level ≥ median of EGFL8 expression levels in all HCC specimens) and low EGFL8 expression group (EGFL8 expression level < median of EGFL8 expression levels in all HCC specimens).

Western blot
Total protein of HCC tissues, ANLTs and liver cancer cell lines was extracted and separated by SDS-PAGE and subsequently transferred onto PVDF (Polyvinylidene Fluoride) membrane (Merck Millipore, Etobicoke, Ontario, Canada), which were incubated successively with primary antibody against EGFL8, Notch1, NICD, Hes1, Hey1 and the corresponding secondary antibody. GAPD H protein in the same samples was also detected as a loading control. The expression levels of EGFL8, Notch1, NICD, Hes1 and Hey1 proteins were normalized to GAPDH. Down-regulation of EGFL8 protein was defined as positive when the relative expression of EGFL8 protein in HCC tissue < the relative expression of EGFL8 protein in the corresponding ANLT tissue.

Lentivirus-mediated EGFL8 transfection
Full-length human EGFL8 cDNA was amplificated by polymerase chain reaction as described previously [17]. Then EGFL8 expression vector was generated by subcloning EGFL8 cDNA into the GV143 plasmid (purchased from GeneChem). The package and infection of lentivirus containing EGFL8 expression vector or empty vector were also described previously [17]. The HCCL M3 cells infected with lentivirus containing EGFL8 expression vector were named as HCCLM3 EGFL8 and those HCCLM3 cells infected with lentivirus containing empty vector were named as HCCLM3 Vector . And Western blot analysis was subsequently used to determine the expression levels of EGFL8 protein in HCCLM3 EGFL8 and HCCLM3 Vector cells.

Wound healing assay
Liver cancer cells were seeded on 96-well plates (3 × 10 4 cells/well). The cells were incubated for 24~48 h and when the confluency reached about 80%, a linear wound was made by a cell scraper (1.2 mm width) across the cell monolayer. Then the cells were washed twice and incubated with the high glucose DMEM without FBS. And photographs were taken at 0, 24 and 72 h by a microscope (MicroPublisher 3.3RTV, OLYMPUS, Tokyo, Japan). The 24 h / 72 h wound closure rate = (0 h width of wound -24 h / 72 h width of wound) / 0 h width of wound × 100%. The scattered spherical cells in the scratch wound of the original image, which were scratched out or washed out by PDS and adhered again in the scratch wound after PDS washing, did not influence the assessment of the results. Experiments were carried out in triplicate.

Transwell invasion assay
Transwell invasion assay were carried out according to the same methods described previously [19]. These experiments were performed in triplicate.

MTT assay and cell apoptosis analysis
The protocol of MTT assay and cell flow cytometric for apoptosis analysis were also described previously [17,18,20].

HCC metastatic mouse model
A total of 20 nude mice, 6 to 8 weeks old, were purchased from Tongji University Experimental Animal Center (Shanghai, China) and reared in cages under SPF (specific pathogen free) conditions at 21-25°C with 40-70% humidity, 12/12 light cycles and free access to food and water. These mice were randomized into the HCCL M3 EGFL8 group and the HCCLM3 Vector group (10 mice/ group). HCCLM3 EGFL8 or HCCLM3 Vector cells were subcutaneously inoculated into the mice in HCCLM3 EGFL8 or HCCLM3 Vector group on their right upper flank regions (1 × 10 7 cells/mouse). Since the subcutaneous tumors emerged from these mice, the volume (V) of tumors was measured every other day until the ninth measurement (in 16 days after the emergence of the tumor) and calculated by using the formula: V = a × b 2 /2, in which a means the largest tumor diameter and b means the smallest one. During the whole time of experiment, the health and behavior of these mice were observed every other day. If the tumor burden of a mouse was evaluated to be high or a mouse was found to be difficult to feed, this mouse would be euthanized early. However, no early euthanasia was actually executed. After the ninth measurement of the subcutaneous tumors, which was in 26~31 days after HCCL M3 EGFL8 or HCCLM3 Vector cells inoculation, all 20 mice were euthanized with pentobarbital sodium by intraperitoneal injection at a dose of 150 mg/kg and then the euthanasia was confirmed by cervical dislocation. Subsequently, the subcutaneous tumors were dissected out and cut into 4 μm thick slices for hematoxylin and eosin (H&E) staining and immunohistochemistry. Lung of each mouse was made into serially sections and observed under a microscope after H&E stain as described previously [12]. Once metastatic HCC cells were found on any slide of lung sections of a mouse, which was considered as lung metastasis positive. All animal experiments were designed to minimize pain or discomfort to the animals and complied with the ARRIVE guidelines and the AVMA guidelines for the euthanasia of animals (2013 Edition). The animals were acclimatized to laboratory conditions (23°C, 12 h/12 hlight/dark, 50% humidity, ad libitum access to food and water) for 2 wk. prior to experimentation. All the experiment protocols involving animals were reviewed and approved by the Ethics Committee of Guangzhou Red Cross Hospital and this study was carried out in compliance with the ARRIVE guidelines.

Immunohistochemistry assay
As described previously, the expression of EGFL8 and Notch1 in mouse xenograft tumors was determined by immunohistochemistry (IHC) assay [12,17,18]. Briefly, the sections made from xenograft tumors were deparaffinized, rehydrated and incubated with 3% H 2 O 2 . Then, these sections were put in citrate buffer (0.01 M, pH 6.0) and heated by a microwave oven at high power for two times, each for 7 min, for antigen retrieval. After rinsing with PBS, the sections were incubated successively with primary antibodies and HRP-conjugated second antibodies. Finally, the sections were visualized by using 3,3diaminobenzidine tetrahydrochloride (DAB) and counterstained with hematoxylin.

DAPT treatment
To verify the role of Notch signaling pathway in the biological effects of EGFL8 on liver cancer cells, EGFL8 depleted Hep3B cells were treated with DAPT (50 μM, dissolved by DMSO), an inhibitor of Notch signaling pathway, for 48 h [21].

Statistics analyses
The expression levels of EGFL8 in HCC tissues were compared with those in ANLTs by Mann-Whitney test. Data come from three or five independent experiments are presented as mean ± SEM (standard error of mean). The differences between two groups were examined by unpaired t test. The comparisons of multi-group were performed by one-way analysis of variance (ANOVA) with Tukey test as post hoc test. The growth curves of liver cancer cells and xenograft tumors were compared by ANOVA with Tukey test. Survival curves were constructed using the Kaplan-Meier method and evaluated using the Log-Rank test. These analyses were all completed by using Graphpad Prism 7.0 software (Graphpad Software, La Jolla, CA, U.S.A.). All analyses were twosided and P < 0.05 was considered as significant.

Results
The decreased expression of EGFL8 in HCC tissues and its association with the clinicopathological features and survival of HCC The expression level of EGFL8 in HCC tissues was significantly decreased compared with the corresponding ANLT specimens (median, 5.651 versus 9.475; P < 0.0001) and the expression of EGFL8 was downregulated in 73.81% (76/102) of HCC patients (Fig. 1a). To verify the results of RT-qPCR, we also detected the expression of EGFL8 protein in 40 cases of HCC and the corresponding ANLTs, which is included in 102 cases of HCC and ANLTs, by Western blot and the results evidenced the decreased expression of EGFL8 protein in HCC tissues (median, 0.260 versus 0.406, P < 0.001; Fig. 1b) and down-regulation of EGFL8 protein was positive in 70% (28/40) of HCC patients. Our results further showed that EGFL8 down-regulation correlated significantly to multiple tumor nodes (P = 0.019), vein invasion (P = 0.012), and high TNM stage (P = 0.031) of HCC (Table 1). However, there was no significant correlation between EGFL8 down-regulation and the other clinicopathologic features of HCC. To explore the prognostic implication of EGFL8 expression in HCC, we divided all 102 cases of HCC patients into low EGFL8 expression group (n = 51) and high EGFL8 expression group (n = 51) according to the results of RT-qPCR and compared the overall and progression-free survival between these two groups. Our results showed that the Fig. 1 The expression of EGFL8 in HCC tissues and its association with prognosis of HCC patients. a The reverse transcription-quantitative polymerase chain reaction (RT-qPCR) results obtained from 102 cases of HCC tissues and matched ANLTs showed that EGFL8 mRNA expression was down-regulated in HCC tissues (the long bar in this figure represents median and the short bars represent 25 and 75% percentiles of the expressions of EGFL8). ***, P < 0.001. b The representative Western blot results showed that Egfl8 protein in HCC tissues was significantly lower than those in ANLTs (the long bar in this figure represents median and the short bars represent 25 and 75% percentiles of the expressions of EGFL8). T, HCC tissues; N, ANLTs; ***, P < 0.001. c Estimated overall survival according to the expression of EGFL8 mRNA in 102 cases of HCC tissues (the Kaplan-Meier method). Log-Rank test shows that HCC patients within low EGFL8 expression group had poorer overall survival than those in high EGFL8 expression group (P = 0.0140). d Progression free survival was analyzed in the same cohort of HCC patients and the results showed that HCC patients in low EGFL8 expression group also had poorer progression free survival than those in high EGFL8 expression group (P = 0.0465) HCC patients within low EGFL8 expression group had either worse overall survival (median survival time, 420 days versus 985 days, P = 0.0140; Fig. 1c) or worse progression free survival (median progression free survival time, 380 days versus 760 days, P = 0.0365; Fig. 1d) than those within high EGFL8 expression group.
The expression of EGFL8 and its overexpression and knockdown in liver cancer cell lines RT-qPCR results showed that the expression levels of EGFL8 in three liver cancer cell lines were also significantly lower than that in HL-7702 liver cell line. Interestingly, EGFL8 expression level in high-metastatic liver cancer cell line HCCLM3 was obviously lower than lowmetastatic liver cancer cell lines HepG2, in which the expression level of EGFL8 was also lower than almost nonmetastatic liver cancer cell lines Hep3B, suggesting the involvement of EGFL8 in metastasis of liver cancer cells (Fig. 2a). Therefore, EGFL8 was overexpressed by lentivirus-mediated gene transduction in HCCLM3 cells and depleted by lentivirus-mediated short hairpin RNA (shRNA) in Hep3B cells. Western blot results subsequently indicated that EGFL8 protein levels in HCCL M3 EGFL8 and Hep3B shEGFL8 cells were increased by 3.66 times (0.2267 ± 0.0203 versus 0.8300 ± 0.0231, P < 0.0001) and decreased by 4.45 times (1.113 ± 0.0524 versus 0.2500 ± 0.0231, P = 0.0001) compared with the corresponding negative control cells, respectively (Fig. 2b).

EGFL8 suppresses the migration and invasion of HCC cells
Since the expression of EGFL8 correlated with the metastatic potential of liver cancer cells (Fig. 2a), we firstly investigated the effects of EGFL8 on HCC cell migration and invasion. As shown in Fig. 2c, though a few scattered spherical cells were found in the scratch wound, the wound healing assay showed that the closure rate of wound in HCCLM3 EGFL8 was significantly lower than HCCLM3 Vector (for 24 h: 3.33% ± 0.33% versus 17.33% ±  Fig. 2d) but the number of Hep3B shEGFL8 cells invaded passed through the matrigel was more than Hep3B shCtrl cells (105.30 ± 6.98 versus 52.33 ± 4.26, P = 0.0008; Fig. 2d), further suggesting a suppressing role of EGFL8 on the metastasis and invasion of HCC cells.

EGFL8 does not affect the proliferation of HCC cells but slightly promotes apoptosis
The MTT results showed no obviously difference between the growth curves of HCCLM3 EGFL8 and HCCL M3 Vector cells (Fig. 2e). Similarly, there was also no significant difference between the growth curves of Hep3B-shEGFL8 and Hep3B shCtrl cells, although the proliferation rates of Hep3B shEGFL8 cells seem slightly higher than Hep3B shCtrl cells in every timepoint (Fig. 2e). On the other hand, the apoptosis level increased in HCCL M3 EGFL8 cells than HCCLM3 Vector cells (6.08% ± 0.11% versus 9.11% ± 0.16%, P < 0.001) and decreased in Hep3B shEGFL8 cells than Hep3B shCtrl cells (2.77% ± 0.07% versus 4.60% ± 0.13%, P = 0.001; Fig. 2f), implicating a slightly promotive effect of EGFL8 on apoptosis of liver cancer cells.

EGFL8 overexpression suppresses the metastasis of HCC cells in vivo
To confirm the results obtained from in vitro studies, we explored the in vivo relevance of EGFL8 in HCC by a mouse HCC metastasis model. The results showed no significant difference in the sizes, weights or growth curves of xenograft tumors between HCCLM3 EGFL8 cells and HCCLM3 Vector cells groups (Fig. 3a-c). Meanwhile, we also have not found any significant difference in body weights of mice between HCCLM3 EGFL8 and HCCL M3 Vector groups (Fig. 3d). Since the in vitro data suggesting a role of EGFL8 in metastasis of liver cancer cells, we further detected the lung metastasis of HCCLM3 cells in these mice. The lung metastasis of HCCLM3 cells was found in four mice in the HCCLM3 EGFL8 group (four of ten, 40%), much less than the lung metastasis rate in the HCCLM3 Vector group (ten of ten, 100%; P = 0.011), as shown in Fig. 3e. Additionally, H&E stain confirmed the diagnosis of HCC for the subcutaneous xenograft tumors of two groups of mice and the results of IHC assay showed a significantly increased EGFL8 expression and an obviously decreased Notch1 expression in the tumors of HCCLM3 EGFL8 cells compared with those tumors of HCCLM3 Vector cells (Fig. 3f).

EGFL8 regulates metastasis, invasion and apoptosis of HCC cells through suppressing the notch signaling pathway
For a recent study of Subhan et al. suggested the involvement of the Notch signaling pathway in the role of EGFL8 [22], we explored the expression of the Notch signaling pathway in EGFL8 overexpressed or depleted liver cancer cells. Our results showed that the expression (See figure on previous page.) Fig. 2 The gain-and loss-of-function experiments indicated the involvement of EGFL8 in migration, invasion and apoptosis of HCC cells. a The expression of EGFL8 in HCCLM3, Hep2B and Hep3B liver cancer cell lines was significantly lower than that in HL-7702 normal liver cell line. EGFL8 expression level was lower in high-metastatic liver cancer cell line HCCLM3 than low-metastatic liver cancer cell lines HepG2. EGFL8 expression in the latter was also lower than almost non-metastatic liver cancer cell line Hep3B. Data were showed as mean ± SEM. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. b The overexpression or inhibition efficiency of EGFL8 protein in HCCLM3 or Hep3B cell lines after lentivirus infection was determined by Western blot. GAPDH was also measured as loading control. Data were showed as mean ± SEM. ***, P < 0.001. Full-length blots are presented in Supplementary Fig. 1. c Wound healing assay showed that 24 h and 72 h wound closure rates were all remarkably reduced in HCCL M3 EGFL8 cells but notedly elevated in Hep3B shEGFL8 cells. DAPT (N-[N-(3,5-Difluorophenacetyl-L-alanyl)]-(S)-phenylglycine t-butyl ester) treatment could significantly decrease wound closure rate of Hep3B shEGFL8 cells, whatever 24 h or 72 h. Magnification, × 200. Data were showed as mean ± SEM. *, P < 0.05, **, P < 0.01, versus HCCLM3 Vector cells; # , P < 0.05, versus Hep3B shCtrl cells; ### , P < 0.001, versus Hep3B shCtrl cells; N.S., no significance, versus Hep3B shCtrl cells; § § § , P < 0.001, versus Hep3B shEGFL8 cells. d The results of Transwell invasion assay showed the number of HCCLM3 EGFL8 cells invaded through the matrigel was significant decreased than that of HCCLM3 Vecter cells. However, the number of Hep3B shEGFL8 cells permeated through the matrigel was obviously increased than that of Hep3B shCtrl cells and DAPT treatment could significantly reduce this number. Magnification, × 200. Data were showed as mean ± SEM. **, P < 0.01; ***, P < 0.001. e The MTT assay showed there was no significant difference between the growth curves of HCCLM3 EGFL8 and HCCLM3 Vector cells. Although a very subtle increase of proliferation could be found in Hep3B shEGFL8 cells compared with Hep3B shCtrl cells, which did not lead to a significant difference between the proliferation of these two groups of cells. DAPT treatment did not exhibit a significant influence on the proliferation of Hep3B shEGFL8 cells. N.S., no significance. f The apoptosis levels of HCC cells were determined by flow cytometric analysis and the cell percentage in each quadrant was shown. The percentage of apoptotic cells in every group of HCC cell was shown by histogram. Data were showed as mean ± SEM. **, P < 0.01; ***, P < 0.001. N.S., no significance levels of Notch1, NICD, Hes1, and Hey1 were obviously suppressed in EGFL8 overexpressed HCCLM3 cells (Fig. 4a) but significantly elevated in EGFL8 depleted Hep3B cells (Fig. 4b), suggested EGFL8 as a suppressing modulator for the Notch signaling pathway. When we treated EGFL8 depleted Hep3B cells with DAPT, a γsecretase inhibitor, the up-regulated expression of the Notch signaling pathway induced by EGFL8 depletion was remarkably declined (Fig. 4b) and the results from wound healing assay showed the migration of Hep3B   Fig. 2c). Moreover, the Transwell assay also evidenced a reduced number of Hep3B shEGFL8 cells invaded pass through the matrigel after DAPT treatment (69.33 ± 3.18 versus 105.30 ± 6.98, P = 0.0057; Fig. 2d). MTT assay showed no influence of DAPT treatment on the proliferation of Hep3B shEGFL8 cells (Fig. 2e), while the apoptosis of DAPT-treated Hep3B shEGFL8 cells was mildly higher than those without DAPT treatment (4.10% ± 0.29% versus 2.77% ± 0.07%, P = 0.0054; Fig. 2f).

Discussion
Although the expression of EGFL8 is known to be highly expressed in kidney, brain, thymus, and lung of adult mouse [13], its expression pattern in human HCC tissue remains unknown. The present study therefore firstly Fig. 4 The expression levels of Notch signaling pathway in EGFL8 overexpressed or depleted HCC cells. a Western blot results showed the expression of Notch1, NICD, Hes1, and Hey1 in HCCLM3 EGFL8 cells was obviously lower than that in HCCLM3 Vecter cells. All the data were showed as mean ± SEM. ***, P < 0.001, versus HCCLM3 Vector cells. Full-length blots are presented in Supplementary Fig. 2. b Western blot results showed the expression of Notch1, NICD, Hes1, and Hey1 in Hep3B shEGFL8 cells was markedly higher than that in Hep3B shCtrl cells. However, the expression levels of Notch1 signaling pathway in Hep3B shEGFL8 cells were significantly decreased after DAPT treatment. All the data were showed as mean ± SEM. *, P < 0.05, versus Hep3B shCtrl cells; ***, P < 0.001, versus Hep3B shCtrl cells; ## , P < 0.01, versus Hep3B shEGFL8 ; ### , P < 0.001, versus Hep3B shEGFL8 . Full-length blots are presented in Supplementary Fig. 2 showed the downregulated expression of EGFL8 in HCC tissues and this down-regulation was evidenced in most (73.81%) of the HCC patients. These results are in accordance with our previous results, showing the decreased EGFL8 expression in colorectal and gastric cancer tissues, and further indicate the down-regulation of EGFL8 as a common pathological event in the development of malignancies of human digestive system [14,15]. Recently, Lu et al. [23] has showed a consistently high methylation level of EGFL8 in 18 cell lines from 9 types of human tumors such as colorectal cancer, suggesting hypermethylation as a potential cause of the down-regulation of EGFL8 in human cancers. When correlating the down-regulation of EGFL8 with the clinicopathologic features of HCC, we found that EGFL8 was more often down-regulated in HCC with multiple nodes than those with solitary node. In addition, EGFL8 down-regulation was also closely associated with vein invasion. For these two clinicopathological characteristics are all acknowledged markers for metastasis of HCC [24,25], it is unsurprisingly that EGFL8 down-regulation was also closely related to high TNM stage, which is also in accord with our previous studies indicating the correlation between EGFL8 downregulation and high TNM stage of colorectal and gastric cancers. Together, these results therefore suggested that EGFL8 might be involved in the metastasis of HCC and the expression of EGFL8 may be down-regulated in accompany with the progression of human digestive cancers including HCC. For TNM stage is a well-accepted marker for the prognosis of HCC [2,6], our Kaplan-Meier analysis also found that the HCC patients with low EGFL8 expression had worse overall survival and progression free survival than those with high EGFL8 expression, suggesting EGFL8 as a potential prognostic biomarker for HCC, which, of cause, should be further verified in the future studies.
We also determined the expression pattern of EGFL8 in a normal liver cell line and three liver cancer cell lines with different metastatic abilities by RT-qPCR. The expression of EGFL8 were all down-regulated in all three liver cancer cell lines compared with the normal liver cell line, which confirmed the down-regulation of EGFL8 in HCC tissues. Of particular interest is the relationship between EGFL8 down-regulation and the metastatic capacity of these liver cancer cell lines. Hep3B and HepG2 cells exhibit almost non-and low-metastatic ability respectively [26,27], whereas HCCLM3 cells has a high metastatic ability to form lung metastases by either subcutaneous or orthotopic inoculation [28]. EGFL8 expression declined in these three liver cancer cell lines in the order by their metastatic potential descending, suggesting an obviously involvement of EGFL8 in the metastasis of HCC.
To understand the role of EGFL8 in HCC development, we employed lentivirus-mediated gene transfer or shRNA to enhance or suppress the EGFL8 expression in HCCLM3 or Hep3B cell line, which had the lowest or highest expression level of EGFL8 among the three liver cancer cell lines we tested respectively. Our results of these gain-and loss-of-function experiments showed at the first time that EGFL8 obviously suppressed the metastatic capacity of HCC cells. The negative regulation of EGFL8 in metastasis of liver cancer cells was further validated by an HCC metastasis mouse model, which showed that the pulmonary metastatic ratio in the EGFL8 overexpression group was significantly lower than the controlled group, indicating EGFL8 as an important modulator in the metastasis of HCC. In view of the possibility that cell proliferation could influence cell migration, we carried out an MTT assay to compare the proliferation of EGFL8 overexpressed HCCLM3 cells or EGFL8 depleted Hep3B cells with the corresponding negative control cells. Interestingly, there was no significant difference was found between the growth curves of these two groups of liver cancer cells. However, our results indeed showed a weakly promotive effect of EGFL8 on the apoptosis of liver cancer cells. These data are consistent with a recent study which showed mouse recombinant EGFL8 protein could inhibit the survival of mouse thymocytes [22].
For Subhan et al. [22] have found that EGFL8 restrains mouse thymocyte proliferation and induces apoptosis by negative regulating the expression of the Notch downstream proteins including Hes1 and Hey1, which are also involved in the metastasis of HCC [21,27]. We therefore presumed that EGFL8 might regulate the metastasis and apoptosis of liver cancer cells through inhibiting the Notch signaling pathway. To verify this hypothesis, we detected the expression levels of Notch signaling pathway including Notch1, NICD, Hes1, and Hey1 in liver cancer cells by Western blot, and the results exhibited a significant decrease of these proteins in EGFL8 overexpressed HCCLM3 cells and an obvious increase of their expression in EGFL8 depleted Hep3B cells. Moreover, IHC assay also evidenced a markedly down-regulation of Notch1 protein in the xenograft tumors came from EGFL8 overexpressed HCCLM3 cells. All these in vitro as well as in vivo data have indicated an inhibition effect of EGFL8 on Notch signaling pathway in HCC. Furthermore, we treated EGFL8 depleted Hep3B cells with DAPT, an inhibitor of Notch signaling pathway [21,29], and found this treatment obviously inhibit the expression of Notch signaling pathway and significantly suppressed the migration, invasion, and survival of Hep3B cells which had been remarkably enhanced by EGFL8 knockdown, suggesting the modulation of EGFL8 on liver cancer cell metastasis and apoptosis is, at least partially, depend on the inhibition of Notch signaling pathway. However, the specific molecular mechanism underlying the regulation of EGFL8 on Notch signaling pathway remains to be further explored.