Nucleolar protein 7 facilitates melanoma progression and metastasis via HIF-1 (cid:0) /PI3K/AKT/ERK axis

tags for relative and absolute quantication; FBS:fetal bovine serum; PBS:phosphate buffer saline; EMT:epithelial mesenchymal transition; CDK:cyclin-dependent


Gene constructs and generation of stable transfectants
Speci c siRNAs were used for the transient knockdown of NOL7 in B16F10 and A375 cells. CRISPR/Cas9 system was operated to generate the stable NOL7-knockout A375 cells based on a previous protocol [15].
In short, sgRNA for A375 cells was designed by a CRISPR design platform (Zhang Feng Lab, http://crispr.mit.edu/) and then cloned into the PX458 vector. The empty vector was used as a control.
These oligo duplexes were transfected into cells via Lipofectamine 3000 Kit (Invitrogen, MA, USA) through standard procedure. Furthermore, stable NOL7-knockout cells were sorted through GFP-activated cell sorting method by ow cytometer (FACS Aria III, BD, San Diego, USA).

Cell proliferation assay
For cell growth kinetic assay, NOL7-knockdown melanoma cells or control cells were plated in 96-well plates (5×10 3 cells/well). Cell proliferation was monitored every day for 5 days using cell counting kit-8 assay (Promega, Sunnyvale, CA) according to the manufacturer's protocol.
For cell spherical growth assay, A375-sgNOL7 cells or control cells were put into 96-well microplates with ultralow attachment surfaces (Corning). Visually 3D cell spheroids were formed after 24 h, and their growth was observed and photographed at regular time intervals by phase contrast microscopy to monitor the tumor spheres diameter.

Colony formation assay
A total of 1000 target cells per well were plated in 6-well plates. After culture in complete medium for 14 days, cell colonies were washed three times with PBS and xed in 4 % paraformaldehyde for 20 min and then, stained with a 0.2 % crystal violet (Sigma-Aldrich) solution for 40 min at room temperature. An overall image of the cell colonies was mapped by Canon scanner (Shanghai, China). Colonies with 50 or more number of cells were counted under a light microscope (Zeiss, Oberkochen, Germany).

Cell cycle assay
Suspended cells were washed twice with precooled PBS and xed in 70 % ethanol for 24 h at 4°C. The cells were then stained with a fluorescent solution (1 % (v/v) Triton X-100, 0.01 % RNase, 0.05 % PI (Sigma-Aldrich)) for 30 min at 37°C in the dark. Ten thousand events were acquired by ow cytometry to measure the cell cycle distribution. Data were processed by Mod t software 3.2.

Cell apoptosis assay
Using an Annexin V-FITC/PI kit (KeyGen Biotech, China), cell apoptosis was detected by ow cytometry.
Target cells at a density of 3×10 5 cells per well were plated in 6-well plates in triplicate. The cells were preprocessed under the following conditions: one group of cells was cultured in complete medium; one group of cells was treated with 10 μM paclitaxel (Meilunbio, Dalian, China) and another group of cells was kept in a suspended state or in FBS-free medium. All of these cells were collected separately after 24 h, and stained with Annexin V-FITC/PI for 15 min under the lucifuge condition. At least 10 4 cells from each specimen were examined by ow cytometry and data were analyzed by FlowJo software 10.0.

Domestication of anoikis-resistant A375 cells
Anoikis-resistant A375 cells were generated through continuous cycle of culturing on normal (adherent) and soft agar (Sigma-Aldrich) -coated cell culture well (suspended). Brie y, A375 cells were plated in soft agar-coated wells and maintained under suspended conditions for the indicated time intervals. After 24 h incubation, cells were washed and transferred to normal cell culture wells. We measured the anoikisresistance of A375 cells by analyzing the cells apoptosis.
In vitro cytotoxicity assay A MTT (3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide, Sigma-Aldrich) assay was used to assess cell sensitivity to paclitaxel, an broad spectrum anticancer drug [16]. The cells were plated in 96well plates at a concentration of 10 4 cells/0.1 mL per well. The cells were treated with various concentrations (0-100 μM) of paclitaxel for 24 h. Finally, the culture medium was replaced with MTT solution (5 mg/mL in phenol red or serum-free medium) and incubated for an additional 4 h. Formazan crystals formed by metabolically viable cells were dissolved in 150 μL DMSO. Absorbance at 490 nm was recorded by an In nite M200 Pro microplate reader (Tecan, Hombrechtikon, Switzerland).

Motility assay
Cell motility was measured by wound healing assay. Monolayer cells in 6-well plates were scratched with a 100 μL pipette tip (t = 0 h) and maintained in 1640 medium containing 1 % FBS. The wounded area was photographed at 0 h and 24 h by a light microscope (Zeiss). Cell mobility was quanti ed as the difference value between the width of the wound at a given time and the initial width of the wound.

Migration and invasion assays
Falcon cell culture inserts with 8 μm pores (Corning, Cambridge, MA, USA) were placed in 24-well plates. To measure invasion, the upper chamber of a Transwell apparatus was precoated with Matrigel (Corning). Cells in 200 μL culture medium containing 1 % BSA were plated into the upper chamber and 800 μL culture medium containing 20 % FBS was added to the lower chamber. After incubation for 24 h, the noninvasive cells that adhered to the upper chamber were wiped away, while the invasive cells adhered to the lower chamber were xed with 4 % paraformaldehyde for 30 min and stained with crystal violet for 40 min. The invasive cells per eld were assessed by a uorescence microscope (Zeiss). To measure cell migration, the upper chamber of the inserts was not precoated with Matrigel.

Adhesion assay
To measure cells adhesion to endothelial cells, monolayer HPMECs were precultured in 6-well plates and stimulated with IL-1β (10 ng/mL, Sigma-Aldrich) for 4 h. Subsequently, the target cells labeled with rhodamine 123 (Sigma-Aldrich) came into contact with the surface of HPMECs and oscillated at 200 rpm for 30 min. After that, the cells that didn't attach to the HPMECs were washed away gently. The cells adhered to HPMECs were photographed and counted by phase-contrast light microscopy (Zeiss). Similarly, to measure adhesion to the ECM, rhodamine 123-labeled target cells were plated into 6-well plates with bronectin-coated (Sigma-Aldrich) and incubated without shaking for 30 min.

Western blotting
Cell lysates were extracted by RIPA lysis buffer containing 1 μM PMSF at 4℃. Then each equal quality of each sample was separated by SDS-PAGE and transferred to PVDF (Bio-Rad, USA) membranes. After blocking in 5 % skim milk solution for 2 h at room temperature, the PVDF membranes were incubated with dilute primary antibodies overnight at 4°C and secondary antibodies at room temperature for 2 h. Protein blots were visualized using the hypersensitive chemiluminescence kit (Wanlei, Shenyang, China) through ChemiDoc XRS system (Bio-Rad). Protein expression levels were analyzed by Image Lab analysis software 5.0 (Bio-Rad). β-actin was served as the internal control.
Quantitative Real-Time PCR Total RNA was extracted using TRIzol reagent (Invitrogen), and 100 ng of each sample was then used as template for reverse transcription to cDNA via a PrimeScript® reverse transcription kit (Takara, Japan) in accordance with the manufacturer's guide. PCR was performed by SYBR® Premix Ex Taq™ PCR kit (Takara) on a CFX96™ real-time PCR detection system (Bio-Rad). The threshold cycle values were used to quantify the relative mRNA expression and normalized to β-actin. Primer sequences used are listed in Supplementary Table 1.

Mice and in vivo mouse study
The mice experiment was guided and approved by the Institutional Animal Care and Use Committee (IACUC) of Fuzhou University, the approval number is SYXK-2019-0007. Female BALB/c nude mice 6-8 weeks old and female C57BL/6 mice 6-8 weeks old were obtained from Slac Animal Inc (Shanghai, China). BALB/c nude mice were housed in clean and sterile cage (Sujingantai Biotechnology, Suzhou, China) and fed with sterilized rat food (Fuzhou Wushi Animal Center, Fuzhou, China). and sterilized water. C57BL/6 mice were housed in clean cage and fed with usual rat food (Fuzhou Wushi Animal Center) and clean water. All mice were accessed to food and water ad libitum. The animal house was kept in a 12 hday/night cycle with lights on at 7:00 a.m and in a temperature (26 ± 1°C) and humidity (50 ± 10 %). All mice were randomised into each group by picking random numbers. The mice in all experiments involved in this manuscript were not given any drugs or anesthetic. We have tired our best to minimize animals' pain and suffering.
To obtain the pulmonary metastases of melanoma described above, B16F10 cells (5×10 4 cells in 0.1 mL PBS per mouse, 5 mice per group) were injected into the tail vein of C57BL/6 mice and A375 cells (3×10 6 cells in 0.1 mL PBS per mouse, 5 mice per group) were injected into the tail vein of BALB/c nude mice, respectively. After 7 weeks, all mice were killed by cervical dislocation in a biological safety cabinet (Sujingantai Biotechnology) and the pulmonary metastases were collected for further analysis.
To assess cell metastatic potential, A375-sgNOL7 cells (3×10 6 cells in 0.1 mL PBS per mouse, 5 mice per group) were injected into tail veins of BALB/C nude mice. Mice injected with A375-control cells (3×10 6 cells in 0.1 mL PBS per mouse, 5 mice per group) were used as control group. After 7 weeks, all mice were sacri ced by cervical dislocation and dissected in a biological safety cabinet. The pulmonary metastatic nodules were counted for further analysis.
To assess the cell tumorigenicity, A375-sgNOL7 cells (6×10 6 cells in 0.1 mL PBS per mouse, 5 mice per group) were subcutaneously injected into the left axillae of BALB/C nude mice. Mice injected with A375control cells (6×10 6 cells in 0.1 mL per mouse, 5 mice per group) were used as control group. About 7-10 days later, the subcutaneous tumor grew and tumor volumes were measured every alternate day. After 4 weeks, all mice were killed by cervical dislocation and dissected in a biological safety cabinet. The tumors were collected for further analysis.

Pathological analysis
Excised neoplasms and lung tissues were embedded in para n and sliced into sections at thickness of 5 μm. Afterward, histological slices were subjected to hematoxylin and eosin (H&E) staining and imaged by a light microscope (Zeiss).

Immunohistochemistry (IHC)
For validation of the clinical signi cance of NOL7 in melanoma. Various clinical samples from melanoma patients (normal skin, n = 5; nevus, n = 5; primary melanoma, n = 32; metastases, n = 10) were purchased from Shanghai Zhuoli Biotechnology Co., Ltd (Shanghai, China). NOL7 expression in these samples was evaluated by IHC analysis with the assistance of Shanghai Zhuoli Biotechnology Co., Ltd. The immunostaining score of NOL7 in tumor tissues was quanti ed by the H-score (histochemical score) based on the staining intensity and heterogeneity under a Nano Zoomer S210 (Hamamatsu, Japan). The percentage of immunostaining and the staining intensity were scored as 0, negative; 1, weak; 2, moderate; and 3, strong. H-score was calculated using the following formula: H score = (percentage of cells of weak intensity×1) + (percentage of cells of moderate intensity×2) + (percentage of cells of strong intensity×3) For analysis of NOL7 expression in melanoma formed in mice. NOL7 expression was evaluated by IHC analysis with the assistance of Wuhan Servicebio Co., Ltd. In brief, para n embedded sections were xed in formaldehyde, and a heat-mediated antigen retrieval step in citrate buffer was performed. The samples were blocked in BSA solution for 2 h at 37°C and incubated with anti-NOL7 antibody overnight at 4°C. An horserdish peroxidase conjugated goat anti-rabbit antibody was used as the secondary antibody. NOL7 protein was visualized by AEC staining and examined under the confocal uorescence microscope (Nikon Eclipse C1, Nikon).

Statistical analysis
Date for all experiments was managed using GraphPad Prism software 5.0 and represented as the mean ± s.d. A paired t test was used for two group comparison, and a one-way ANOVA was used for multiple group comparison. P < 0.05 was considered representative of a signi cant difference and statistical signi cance was symbolized by asterisks corresponding to * P < 0.05, ** P < 0.01 and *** P < 0.001.

NOL7 expression correlates with melanoma progression
Post-metastatic cell lines possess an inherently increased capacity for tumorigenesis and metastasis [17]. As described previously, we established the post-metastatic melanoma cell line B16F10M, derived from the pulmonary metastases of the parental pre-metastatic B16F10 cells. The paired cell lines B16F10 and B16F10M were used to probe the global proteomics signature that marks melanoma metastasis to the lung and secondary tumor growth through iTRAQ screening [14]. The iTRAQ analysis showed that NOL7 was upregulated in B16F10M cells compared to B16F10 cells (Fig. 1A). Similarly, another post-metastatic cell line A375M was derived from the pulmonary metastases of the parental pre-metastatic A375 cells (Fig. 1B). In both metastatic cell lines, we con rmed the increased expression of NOL7 and PCNA (Proliferating cell nuclear antigen) both at the transcriptional and translational levels, compared with that in the parental B16F10 and A375 cells (Fig. 1C-D).
To further explore the role of NOL7 in melanoma progression, we examined the expression of NOL7 in clinical melanoma samples. Various tissue samples including normal skin, nevus, primary melanoma and melanoma metastases were used. As shown in Fig. 1E, NOL7 expression was nearly undetected in normal skin and nevus samples. In contrast, NOL7 expression was markedly elevated in primary melanoma, and further increased in metastatic melanoma samples. Cumulatively, these ndings demonstrate by multiple means that NOL7 is increased in melanoma metastases. As the primary tumors also show augmented levels of NOL7 as compared with the normal skin or healthy nevus, NOL7 might be expected to additionally regulate other malignant characteristics in melanoma. These observations prompted us to sequentially investigate the effects of NOL7 on multiple melanoma activities, from proliferation to in vivo metastasis, as detailed in the subsequent sections.

NOL7 knockdown suppresses melanoma cell proliferation and induces cell cycle arrest in vitro
As the rst step to investigate sequentially the roles NOL7 plays in melanoma progression, we investigated whether NOL7 was required for malignant cell proliferation of melanoma in vitro by speci c siRNAs knockdown in B16F10 and A375 cells (B16F10-siNOL7 and A375-siNOL7 cells). After siRNA treatment, NOL7 protein expression was barely detectable, and its mRNA levels were decreased by ca.80% and ca.70% in the two cell lines, respectively, compared with control B16F10 and A375 cells, where siRNAs whose sequences did not react on any known genes were transfected (see Methods, Fig. 2A).
With this approach, we observed that NOL7 knockdown signi cantly reduced the cell proliferation rate and the number of cell colonies, compared with the control group ( Fig. 2B-D). Since melanogenesis is a marker of differentiated melanoma cells [18], we examined whether NOL7 knockdown interfered the melanin production in melanoma cells. As shown in Supplementary Fig. 1, NOL7 knockdown disrupted melanin biosynthesis and the activity of tyrosinase (enzyme catalyzing the initial rate-limiting reaction in melanin biosynthesis in B16F10 cells [19]).
We next used ow cytometry to assess the involvement of NOL7 in the cell cycle progression. As shown in Fig. 2F, NOL7 knockdown markedly added the cell populations at the S-phase, but reduced the cell populations at the G0/G1-phase. Furthermore, various cell cycle-related proteins were altered by NOL7 knockdown (Fig. 2E). Speci cally, the expression levels of p21 and p27 were signi cantly increased, and those of CDK2, cyclin A and cyclin E -signi cantly decreased upon the knockdown of NOL7. These changes in the levels of cell cycle regulators are consistent with the S-phase arrest induced by NOL7 knockdown. Indeed, the CDK inhibitors p27 and p21 exert inhibitory effects on cell cycle progression through inhibition of the G1 cyclin/CDK complexes, such as cyclin A/CDK2 and cyclin E/CDK2 [20,21]. Sphase cell-cycle arrest has been previously described to be induced by up-regulation of p21 and p27, reducing formation of active cyclin E/CDK2 and cyclin A/CDK2 complexes [22,23,24,25]. In contrast, levels of Rb were not affected by NOL7 knockdown, which agrees with the prior observation that Rb is upstream of NOL7 [10].

NOL7 is bene cial to cell survival by regulating apoptosis resistance
We next moved to study the role of NOL7 in apoptosis resistance of melanoma cells when subjected to a disadvantageous microenvironment. For this purpose, B16F10-siNOL7 and A375-siNOL7 cells were treated with paclitaxel. Additionally, A375-siNOL7 cells were placed in suspended culture conditions (loss of anchorage) and FBS-free culture conditions (loss of nutrition). Apoptosis assays consistently indicated a higher apoptotic index in cells with NOL7 knockdown (Fig. 3A, Supplementary Fig. 2). The MTT assay showed decreased cell viability upon paclitaxel treatment and IC 50 values in NOL7-knockdown cells compared with those in control groups (Fig. 3B-C).
Subsequently, we domesticated anoikis-resistant A375 cells in a continuous cycle between the adherent culture and suspended culture conditions. After 3 cycles, suspended A375 cells were almost nonapoptotic (Fig. 3D, Supplementary Fig. 3). It suggested that domesticated A375 cells were resistant to anoikis. Interestingly, we found a signi cant upregulation of NOL7 expression in anoikis-resistant cells, suggesting that that NOL7 was associated with the anoikis resistance (Fig. 3E). Similarly, we put the A375-sgNOL7 cells (see below for the details on the CRISPR/Cas9-mediated NOL7-knockout) in a continuous cycle between the adherent culture and suspended culture conditions. Within the 3rd cycles, A375-sgNOL7 cells showed a signi cantly higher apoptotic index than the parental A375 cells. It suggested that NOL7 depletion impaired anoikis resistance of melanoma cells (Supplementary Fig. 3). Furthermore, NOL7 knockdown inhibited the expression of survivin, but promoted the expression of caspase 3, caspase 9, bad and bax (Fig. 3F, Supplementary Fig. 4). Cumulatively, these results suggest the participation of NOL7 in melanoma cell self-adaptation and protection against apoptosis induced by chemotherapy and starvation, and against anoikis.

NOL7 knockdown depresses melanoma aggressiveness in vitro
Having demonstrated the importance of NOL7 in melanoma cell proliferation, colony formation, and apoptosis resistance, which are largely characteristics of the primary tumor growth, we then moved to study the functions on NOL7 in metastasis. To start, we utilized the wound healing assay, the transwell assay and the cell-ECM/HPMEC adhesion assay in order to evaluate the impact of NOL7 knockdown on cell motility, migration or invasion, and adhesion abilities in melanoma, respectively. As shown in Fig. 4, control B16F10 and A375 cells were faster to cover the wound area than the NOL7 knockdown cells, indicating that the motility of melanoma cells was inferior upon decreasing the NOL7 levels. In the adhesion assay, NOL7 knockdown inhibited cells adhesion to gelatin matrix and HPMECs. In comparison to control cells, the number of B16F10-siNOL7 and A375-siNOL7 cells adhered to gelatin matrix or HPMEC was reduced by 30-40% (Fig. 4B). Meanwhile, falcon cell culture insert systems were used to monitor the process of melanoma cell migration or invasion (migration through Matrigel). Upon NOL7 knockdown, the migratory capabilities were reduced by ca.80% and ca.50% in B16F10 and A3757 cells, and invasive capabilities were reduced by ca.70% and ca.60% in B16F10 and A375 cells, respectively.

NOL7 depletion represses melanoma growth and metastasis in vivo
Further to examine the role of NOL7 on melanoma growth in vivo, and to assess an independent means of reducing NOL7 levels, we utilized the CRISPR/Cas9 system to generate a stable NOL7-knockout A375 cell line (A375-sgNOL7 cells). The expression of NOL7 in A375-sgNOL7 cells was heavily depleted (Fig. 5A, inset). Separately, we measured the cell proliferation and colony formation capabilities of A375-sgNOL7 cells. Same as in Fig. 2B-D, NOL7 knockout markedly reduced cell proliferation and cell colony formation in A375 cells (Supplementary Fig. 5). In addition, the 3D spheroid formation assay was employed to grow cells in a 3D globular cluster to simulate the in vivo cancer cell growth. After culturing in equal amounts of complete medium for 10 days, A375-sgNOL7 cells had a signi cantly smaller cell spheroid diameter than control cells (Supplementary Fig. 6).
Next, an orthotopic xenograft nude mouse model was established. A375-sgNOL7 cells and control A375 cells were injected subcutaneously into the left axillae of nude mice. Tumor growth was monitored by measuring tumor volumes every two days. In contrast to mice injected with control cells, which ripened large tumors, all ve mice injected with A375-sgNOL7 cells appeared to harbor either noticeably smaller tumors or no tumors at all within 4 weeks (Fig. 5A). Then, we surveyed whether NOL7 depletion repressed melanoma cell metastasis. Paired A375-sgNOL7 cells and control A375 cells were injected into the lateral tail veins of nude mice and evaluated for formation of distant metastasis. The end-point analysis revealed that the ve mice injected with the control cells formed 54 to 134 tumor nodules per lung. In contrast, the ve mice injected with A375-sgNOL7 cells formed 10 to 26 tumor nodules per lung (Fig. 5B). Three out of ve mice injected with control cells were sacri ced in advance because they became underweight within 3 weeks, and notable metastases were observed in other locations -in a sharp contrast with no such metastases induced upon the NOL7 knockout cell injection (Fig. 5C). Also, NOL7 was observed to be associated with the overall survival of mice in this model (Fig. 5D). Pathological analysis con rmed the appearance of tumors from A375-sgNOL7 cells, and IHC staining con rmed essentially undetectable NOL7 expression in A375-sgNOL7 cells (Fig. 5A-B).

NOL7 is induced by HIF-1 under hypoxia and participates in hypoxia-induced EMT and drug resistance in melanoma
Having determined the importance of NOL7 for multiple aspects of melanoma progression, from tumor cell proliferation to metastasis, we then set to unravel the potential molecular mechanisms behind.
Hypoxia/HIF-1 signaling inducing target gene expression to promote cancer progression and metastasis has been widely recognized [26,27]. Therefore, we examined whether the hypoxia/HIF-1 signaling activated NOL7 expression. Western blotting was used to test the expression of HIF-1 and NOL7 in A375 cells when exposed to hypoxic conditions (1% O 2 ) for different durations. HIF-1 expression was triggered in response to hypoxia at 0.5-1 days, peaked at 2 days, and stable HIF-1 expression was maintained within 5 days. Remarkably, NOL7 expression was induced by hypoxia with a delay as compared to HIF-1 , peaking at 2 days, and stable NOL7 expression was maintained within 5 days (Fig. 6A). The delay in NOL7 expression under hypoxia could indicate that NOL7 was induced by HIF-1 . To test this possibility, we employed YC-1, a chemical compound that accelerates HIF-1 degradation and blocks HIF-1 expression, during 2 days-culturing under hypoxic conditions (1% O 2 ). As a result, the protein levels of NOL7 and HIF-1 returned to their initial levels (Fig. 6B).
HIF-1 advances cancer development through conferring drug resistance to cancer cells and stimulating the EMT [28,29]. To identify the role of NOL7 in the hypoxia/HIF-1 signaling, A375-sgNOL7 cells and control cells were cultured under hypoxic conditions (1% O 2 for 2 days). As shown in Fig. 6C, D and Supplementary Fig. 7, NOL7 knockout has no effect on the morphology of A375 under normoxic condition. However, hypoxic condition motivated the conversion of the non-polar (upon the 2D culturing conditions) epithelial phenotype to the polar mesenchymal phenotype, and altered the EMT-associated protein expression of E-cadherin, N-cadherin, Vimentin and Fibronectin. A375 cells invasion and migration were strikingly enhanced after the hypoxic treatment. Interestingly, NOL7 depletion reversed the hypoxiainduced EMT and impaired the elevated invasion and migration characteristics (Fig. 6E-F).
Furthermore, to investigate the contribution of NOL7 to chemoresistance caused by hypoxia, we inspected the sensitivity of A375-sgNOL7 cells to paclitaxel under hypoxia. As expected, decreased cytotoxicity of paclitaxel and improved IC 50 values were observed under hypoxia. Further, NOL7 knockout reverted the hypoxia-induced desensitization of A375 cells to the drug (Fig. 6G-H). Thus, NOL7 is induced by HIF-1 under hypoxia and is required for HIF-1 -induced EMT, invasiveness, and chemoresistance.

NOL7 modulates EMT and the PI3K/AKT/ERK pathway in melanoma
To further explore the molecular mechanism(s) of the NOL7-induced melanoma malignancy, we investigated the participation of NOL7 in EMT and the AKT, ERK, JNK, and Wnt/β-catenin signaling pathways, which are widely recognized as activators of cancer progression [30,31]. As shown in Fig. 7A, NOL7 knockdown/knockout increased the levels of the epithelial marker E-cadherin, and correspondingly depressed the levels of mesenchymal makers N-cadherin, MMP9, Vimentin and Twist both in B16F10 and A375 cells. Meanwhile, NOL7 knockdown/knockout in B16F10 and A375 cells led to decreased levels of p-AKT and p-ERK, while GSK3β, a negative regulator of the PI3K/AKT/ERK signaling and Wnt/β-catenin signaling, was upregulated [32] (Fig. 7B). However, JNK, p-JNK, β-catenin and rhoA levels were unaffected after NOL7 knockout in B16F10 or A375 cells (Supplementary Fig. 8). These results indicate that NOL7 modulates EMT and the PI3K/AKT/ERK pathway and, potentially, the Wnt/β-catenin signaling, to promote melanoma progression. At the same time, we nd no impact of modulation of NOL7 levels on such established players in the EMT program as TSP-1, SMAD2/3, and pSMAD2/3 [33,34] ( Supplementary  Fig. 9).

Discussion
Melanoma is one of the most undesirable clinical events, due to its dramatic tendency toward distal metastasis [35,36]. Over the years, we have focused on uncovering the identity and mechanisms of key molecules that regulate how the few disseminated cancer cells survive and metastasize [17,37]. In the current study, we identi ed NOL7, one of the top altered genes from the metastatic melanoma signature, and subjected it to further intensive studies. Herein, we presented that NOL7 is a novel and multifunctional oncogene in melanoma operating in the tumor progression and dissemination mechanisms. These ndings contrast with the prior description of NOL7 as a tumor suppressor in cervical cancer [9]. Previous studies have shown that melanoma is commonly associated with an ampli cation of the chromosome region 6p, particularly 6p21-23 where the NOL7 gene resides, and that this region frequently undergoes heterozygous loss in cervical cancer [38]. It might therefore be predicted that NOL7 exhibits a different expression pattern and plays different roles in melanoma and cervical cancer.
Our data suggest that NOL7 is induced by HIF-1 to promote melanoma progression, potentially through modulation of the PI3K/AKT/ERK pathway. Hypoxia is a fundamental hallmark of tumor microenvironment, and the hypoxia/HIF-1 signaling is critical for cancer progression [27]. Previous reports have shown that HIF-1 can induce many molecules and further activate several downstream signaling pathways, including the PI3K/AKT, ERK and MAPK pathways, to regulate malignant behaviors in cancer cells [39]. We observed that NOL7 expression was stimulated by the hypoxia/HIF-1 signaling and mediated the HIF-1 -triggered EMT, cell aggressiveness and chemoresistance (Fig. 6). Numerous published studies have indicated that the PI3K/AKT, ERK, JNK and rhoA signaling pathways are frequently dysregulated in human cancers [40,41]. Aberrant activation of these pathways leads to extensive cellular transformation, tumorigenesis, disease deterioration and resistance to treatment. Here, we disclosed that NOL7 promoted the AKT and ERK phosphorylation and decreased the GS3Kβ expression, but had no effect on JNK, p-JNK, β-catenin and rhoA (Fig. 7, Supplementary Fig. 8).
Cancer cell growth is associated with acceleration of cell cycle progression and acquisition of apoptotic resistance [42,43]. As the core components of the cell cycle machinery, cyclins and their catalytic partners cyclin-dependent kinases (CDKs) drive cell cycle progression in cancer cells; on the contrary, CDK inhibitors such as p21 or p27 negatively regulate this process [44,45]. Furthermore, apoptosis regulatory proteins, such as the pro-apoptotic caspase family proteins and bax maintain the immortality of cancer cells [46,47]. The PI3K/AKT/ERK pathway has been shown to regulate CDKs, p21, and cyclins to expedite the cell cycle progression, and to alter the expression of caspases proteins to defend against apoptosis, which results in drug resistance [48]. In the current study, we showed that NOL7 depletion altered protein levels of p27, p21, CDK2, cyclin A and cyclin E to induce the S-phase arrest in melanoma cells. Paclitaxel is a chemotherapy drug conventionally applied to melanoma patients in China [49] and drives apoptosis through cell cycle arrest in the G2/M phase [50]. We kept in mind the potential for the synergistic action of the drug on top of the NOL7 knockdown. Data also revealed that NOL7 could decrease the sensitivity of melanoma cells to paclitaxel treatment and to anoikis (Fig. 2-3).
The connection between the PI3K/AKT/ERK pathway and EMT has been well-documented. During EMT, cancer cells change their morphology, extend pseudopodia, degrade the surrounding basement membrane and invade through the ECM. Activation of the PI3K/AKT/ERK pathway leads to EMT through the induction of MMP9 and increased activity of GSK3β [51]. As observed in the current research, NOL7 promoted the PI3K/AKT/ERK pathway and induced EMT to improve the metastatic potential of melanoma cells. Importantly, the observation that depletion of NOL7 reversed the EMT and arrested the cell cycle, as well as opposed chemoresistance and metastasis is consistent with the data on the role of NOL7 in the hypoxia/HIF-1 signaling. Collectively, these data depict the molecular mechanism whereby the hypoxia/HIF-1 signaling induces NOL7 expression to further govern the revitalization of the PI3K/AKT/ERK pathway.

Conclusion
Taken together, our research reveals the crucial role of NOL7 in melanoma. From a functional perspective, NOL7 is bene cial for cell survival and contributes to tumor growth and metastasis of melanoma through promoting cell proliferation, cell cycle progression, aggressiveness, as well as acquisition of chemo-and anoikis-resistance. From a mechanistic perspective, NOL7 emerges as a novel player within the HIF-1 /PI3K/AKT/ERK axis to ultimately activate regulators of the cell cycle, apoptosis and EMT to exert its oncogenic function.
In summary, we identify NOL7 as a multifunctional regulator of malignant activities, proposing it as a novel biomarker of melanoma progression and potential drug target for future therapeutics.