NVD-BM-mediated genetic biosensor triggers accumulation of 7-dehydrocholesterol and inhibits melanoma via Akt1/NF-ĸB signaling

Aberrant activation of the cholesterol biosynthesis supports tumor cell growth. In recent years, significant progress has been made by targeting rate-limiting enzymes in cholesterol biosynthesis pathways to prevent carcinogenesis. However, precise mechanisms behind cholesterol degradation in cancer cells have not been comprehensively investigated. Here, we report that codon optimization of the orthologous cholesterol 7-desaturase, NVD-BM from Bombyx mori, significantly slowed melanoma cell proliferation and migration, and inhibited cancer cell engraftment in nude mice, by converting cholesterol to toxic 7-dehydrocholesterol. Based on these observations, we established a synthetic genetic circuit to induce melanoma cell regression by sensing tumor specific signals in melanoma cells. The dual-input signals, RELA proto-oncogene (RELA) and signal transducer and activator of transcription 1 (STAT1), activated NVD-BM expression and repressed melanoma cell proliferation and migration. Mechanically, we observed that NVD-BM decreased Akt1-ser473 phosphorylation and inhibited cytoplasmic RELA translocation. Taken together, NVD-BM was identified as a tumor suppressor in malignant melanoma, and we established a dual-input biosensor to promote cancer cell regression, via Akt1/NF-κB signaling. Our results demonstrate the potential therapeutic effects of cholesterol 7-desaturase in melanoma metabolism, and provides insights for genetic circuits targeting 7-dehydrocholesterol accumulation in tumors.


INTRODUCTION
Globally, melanoma is a serious malignant skin cancer, and is estimated to kill approximately 55,500 individuals annually [1]. Increasing melanoma incidences are reported in older Caucasians, with males experiencing higher rates than females [2,3]. Traditional surgical and chemo-radiotherapy treatments often fail to prolong patient survival, especially in those with advanced disease. This is due to metastasis and chemotherapy resistance, primarily due to acquired somatic mutations, e.g., B-Raf oncogene AGING serine/threonine-kinase (BRAF), NRAS proto-oncogene (NRAS) and neurofibromin 1 (NF1) [4]. Targeted therapies such as dabrafenib and vemurafenib are proven melanoma treatments, generating positive results in clinical settings [5][6][7][8][9]. However, drug resistance after a period of treatment, leads to poor responses or relapses in advanced melanoma patients [10,11].
Recent studies have demonstrated that metabolic reprogramming plays important roles in melanoma tumorigenesis, and may be responsible for drug resistance and poor prognoses [12]. Therefore, the identification of novel therapeutic strategies targeting cancer metabolism, and the concomitant development of therapeutic drugs are essential for melanoma cancer therapy.
Cellular cholesterol levels are tightly regulated by an intricate network of transcriptional regulation of cholesterol biosynthesis, deposition, cellular uptake and efflux programs [13][14][15]. Cholesterol metabolism dysregulation is associated with cancers. Oncogenic processes permits cholesterol synthesis in cancer cells, which is further metabolized to support rapid proliferation [16]. Recent investigations have shown that cholesterol biosynthesis gene activation is associated with poor prognoses in leukemia, sarcoma and melanoma [17]. Indeed, perturbed expression of key cholesterol biosynthesis enzymes, such as HMGCR and SQLE, prevents carcinogenesis in colorectal cancer, breast cancer, prostate cancer and hepatocellular carcinoma associated with nonalcoholic fatty liver disease (NAFLD-HCC) [18][19][20][21][22]. However, the role of cholesterol and its metabolites in cancer is still controversial [17,[23][24][25]. In breast cancer, decreased CYP7B1 expression triggers 27-hydroxycholesterol (27HC) accumulation [26], permitting 27HC to act as an estrogen receptor agonist that induces tumor growth and metastasis [25]. Other reports have observed that the cholesterol metabolites, dendrogenin A [27], cholestane-3β, 5α and 6β-triol [28] exert antiproliferative effects in human breast and prostate cancer cells [28,29]. Of note, the cholesterol metabolite, 7dehydrocholesterol (7-DHC) exhibits toxic and proapoptotic effects towards melanoma cells [30], thereby encouraging us to develop a therapeutic strategy incorporating cholesterol conversion to 7-DHC.
Although the cholesterol 7-desaturase conversion of cholesterol to 7-DHC has not yet been reported in mammalian cells, several evolutionary conserved isoenzymes exist in orthologous organisms, such as NVD-BM from Bombyx mori, DAF-36 from Caenorhabditis elegans, NVD-DR from Danio rerio, TTHERM_00310640 from Tetrahymena thermophila SB210 and MGC154819 from Xenopus laevis [31][32][33]. Natural products including 7-DHC have been proven in cancer therapy, however, dose administration and side effects often hinder their clinical applications [34]. Designing and synthesizing genetic regulatory circuits in living cells, to program new biological behaviors, is a promising synthetic biological approach for medicinal research and potential therapeutic applications [35,36]. Reprogrammed human cells can circumvent the shortcomings of conventional therapies, such as frequent administration and self-sufficient doses, providing new strategies to solve the complexities of metabolic disorders [37]. In this study, we performed codon optimization on several cholesterol 7-desaturases to examine their roles in melanoma cell proliferation, migration and tumorigenicity, both in vitro and in vivo. Furthermore, we developed a smart genetic circuit to induce 7-DHC accumulation by sensing transcription factors in melanoma cells.

7-desaturase expression inhibits melanoma cell growth and migration
Since no cholesterol 7-desaturases exists in mammalian cells, we optimized codon usage and synthesized several cholesterol 7-desaturases from other species; DAF-36 from Caenorhabditis elegans, NVD-BM from Bombyx mori, NVD-DR from Danio rerio (zebrafish), TTHERM_00310640 from Tetrahymena thermophila SB210 and MGC154819 from Xenopus laevis (Table 1,  Supplementary Table 1). This approach facilitated anticancer studies in melanoma cells. Using transient transfection, the five humanized cholesterol 7desaturases were expressed in A375 and A2058 melanoma cells by qRT-PCR assay ( Figure 1A, 1B). Cell proliferation assays showed that the expression of NVD-BM significantly slowed cell growth of melanoma cells(A375 and A2058), however, DAF-36 failed to suppress A2058 cell proliferation and other cholesterol 7-desaturases did not inhibit melanoma cell proliferation ( Figure 1C, 1D). Furthermore, the proliferation effects exerted by NVD-BM expression were dose dependent in A375 cells (Figure 2A). We next investigated cell migration capabilities in cells expressing NVD-BM, using the real-time cell analysis (RTCA). These data showed that A375-NVD-BM and A2058-NVD-BM cell migration capabilities were lower than the empty vector negative control (NC) group at 36 h, suggesting that NVD-BM severely impaired melanoma cell migration ( Figure 2B, 2C).

NVD-BM inhibits melanoma cell proliferation in xenograft mice
The proliferation and migration inhibitory capabilities of NVD-BM in melanoma cells suggested anti-tumor AGING approximately four folds smaller significantly than the NC group (p = 0.00063) ( Figure 3A). Moreover, overall tumor structures were almost normal, and displayed no obvious necrosis ( Figure 3B). In contrast, the NC group displayed large areas of necrosis, nuclear degradation and cytoplasmic protein degradation. These data demonstrated that NVD-BM served as an effective tumor suppressor in melanoma derived cells in vivo. AGING

NVD-BM catalyzes cholesterol to 7-DHC in melanoma cells
To test cholesterol 7-desaturase capability in converting cholesterol to 7-DHC, we extracted sterols from A375 cells with the expression of cholesterol 7-desaturases, and determined cholesterol and 7-DHC concentrations using GC-MS ( Figure 4A). Our data revealed that NVD-BM converted cholesterol to 7-DHC, and was 2.1 folds compared with NC group ( Figure 4B, 4C). DAF-36 failed to convert cholesterol to 7-DHC in A375 cells, suggesting that DAF-36 inhibited A375 cell proliferation via alternative, non-cholesterol mediated mechanisms. Therefore, NVD-BM appeared to effectively inhibit melanoma growth via cholesterol degradation, to increase 7-DHC levels in cells. This observation was consistent with the previous report showing that the proliferation of melanoma cells was inhibited by 7-DHC addition (Supplementary Figure 1).

The development of a dual-input signal genetic biosensor
To enhance the efficacy and specificity of NVD-BM, based on a genetic circuit, we were ought to scan highly expressed transcription factors as dual-input sensors. We compared gene expression profiling data from melanoma tissue with normal skin tissue, and identified several differentially expressed melanoma genes. The top 10 highly expressed transcription factors, RELA, STAT1, FOXD1, POU3F4, SOX5, ELF1, TBX2, LEF1, HOXD13 and POU3F2, were then evaluated for their ability to drive luciferase expression ( Figure 5A). Five repetitive transcription factor binding motifs were separately cloned into firefly luciferase vectors, driven by mini-promoters, to determine firefly luciferase expression in A375 and A2058 cells (Supplementary Table 2, Figure 5B). As shown in Figure 5C, RELA, STAT1, POU3F2 and POU3F4 were the top four input AGING signals in A375 and A2058 cells. Next, vectors with different input signals, were divided into 16 permutation methods. Heatmaps showed that the optimum match was RELA and STAT1 in A375 and A2058 cells, respectively ( Figure 5D). Finally, RELA and STAT1 were selected as dual-input signals for a genetic biosensor in melanoma cells.

The activated genetic biosensor inhibits melanoma cell growth and migration
We next developed a molecular switch combined with CRISPR-Cas9 and TET-ON, with four modules to activate NVD-BM expression and inhibit melanoma cell growth and decrease migration rates ( Figure 6A). Five repetitive motifs from RELA (5'-GGGAATTTCC-3'), Cas9 and sgRNA, targeting the lacI gene were assembled into one construct, facilitating RELA binding to induce Cas9 expression. The Cas9 protein with sgRNA-lacI in the next module, inhibited lacI expression. Five repetitive motifs from STAT1 (5'-CATTTCCCGGAAACC-3'), lacO and rtTA were constructed in one vector, recruiting STAT1 binding to induce rtTA expression. In case of toxicity, NVD-BM was not expressed without doxycycline. When doxycycline was added to the culture medium, it entered cells and bound rtTA. Then, rtTA and doxycycline bound to the TRE gene to activate NVD-BM expression. Thus, when RELA and STAT1 were up-regulated in a tumor, parallel to doxycycline added in vitro, the genetic circuit was then switched on ( Figure 6B).
To test genetic circuit availability in melanoma cells, we co-transfected constructs into A375 and A2058 cells, and administered doxycycline at final concentration of 1 μM. Quantitative RT-PCR data showed that NVD-BM expression was activated by doxycycline ( Figure 7A), and A375 and A2058 proliferation and migration rates were significantly impaired by NVD-BM activation ( Figure 7B-7D). Thus, our genetic circuit effectively sensed the dualinput of RELA and STAT1 signals, and induced NVD-BM actuator expression to rebuild cholesterol metabolism by converting cholesterol to 7-DHC, promoting melanoma cell regression.

NVD-BM functions as a tumor suppressor via Akt1/NF-κB signaling
Enhanced NF-κB activation, as a result of upstream signaling dysregulation, is a main molecular characteristic in melanoma [38]. Indeed, GSEA analysis of GSE3189 transcriptomic data between normal skin and melanoma samples, demonstrated that NF-κB signaling was up-regulated in melanoma patients ( Figure 8A). Thus, we determined the expression of known NF-κB targets in melanoma, e.g., MMP9, CCL20 and IL1B [39,40]. RT-PCR data demonstrated that MMP9, CCL20 and IL1B expression was significantly down-regulated by NVD-BM ( Figure 8B). Similarly, by investigating the cellular localization of free RELA, which is an important transcription factor implicated in NF-κB signaling, we noted that RELA accumulated in the cytoplasm by western blot assay, in a dose dependent NVD-BM manner ( Figure 8C), thus, suggesting that RELA translocation to the nucleus was inhibited under these conditions. Next, we determined the phosphorylation status of an upstream NF-κB kinase by western blot assay, and found that Akt1-ser473 phosphorylation was decreased by NVD-BM, in a dose dependent manner ( Figure 8D). Collectively, these data indicated that NVD-BM promoted tumor regression by decreasing phosphorylated Akt1 levels, and inhibiting free RELA translocation into the nucleus in melanoma cells.

DISCUSSION
Metabolic reprogramming has received considerable attention due to its ability to support growth and antiapoptosis mechanisms in cancers [23]. Disrupting the expression of key enzymes in cholesterol biosynthesis, such as HMGCR and SQLE, appears to prevent carcinogenesis mechanisms in colorectal, breast, prostate and NAFLD-HCC cancers [18][19][20][21][22]. Though comprehensive studies showed that high cellular cholesterol contributes to tumor growth and metastasis, it is still not clear in melanoma yet. Indeed, transcriptome data shows that cholesterol biosynthesis gene activation is associated with poor prognoses in melanoma patients [17], however, the expression of cholesterol biosynthesis limiting enzymes HMGCR was down-regulated in melanoma patients according to analysis of TCGA-SKCM and GSE3189 transcriptome data (Supplementary Figure 2). In present study, administration of 7-DHC in vitro suppressed melanoma cell growth [38], suggesting that the strategic manipulation of cholesterol metabolism may be a potential strategy for melanoma treatment. Because no cholesterol 7-desaturases exist in mammalian cells, we screened and cloned evolutionary homologous cholesterol 7-desaturases from several species and tested them against several melanoma cell pathological characteristics, such as cell proliferation and migration. When compared with DAF-36, NVD-BM from Bombyx mori exhibited a stronger ability to inhibit growth in in vitro melanoma cells, and growth in an in vivo xenograft mouse model. Consistent with GC/MS data, DAF-36 failed to convert cholesterol to 7dehydrocholesterol. Thus, we believed that NVD-BM is a novel tumor repressor in melanoma.  AGING Synthetic cell-based therapies are ideal approaches for metabolic disorders, as they circumvent the shortcomings of conventional therapies, such as frequent administration and self-sufficient doses. Genetic circuits are typically composed of a sensor, processor and actuator [35]. From our results, NVD-BM was an ideal actuator in promoting 7-DHC accumulation in skin cancer. To ensure specificity, safety and efficacy of the genetic circuit, the sensor and processor were composed of four independent modules, which sensed highly expressed transcription factors in tumors, and activated downstream step by step. The CRISPR-Cas9 system and the Tet-on switch were central to our genetic circuit, guide RNA and Cas9 device were detached on separated module and dimerized once sensing up-regulated transcription factors. RELA and STAT1 binding motifs were blocked, and the switch was activated by doxycycline.
Our results showed that RELA and STAT1 activated NVD-BM expression in the presence of doxycycline, and slowed melanoma cell growth and migration. However, endogenous transcription factors usually AGING function as "master regulators", exerting control over processes that specify cell types and developmental patterning in normal cells. Hence, preventing signal leakage and accurately setting specific threshold values that distinguish tumor and normal cells to trigger circuit activation, should be investigated in the future.
Taken together, we identified NVD-BM from Bombyx mori as a novel melanoma tumor suppressor. It decreased Akt1 phosphorylation levels and inhibited RELA translocation into nuclei. We also developed an effective genetic circuit, based on an "AND" gate approach, to activate NVD-BM expression by sensing RELA and STAT1. Our results provide insights to against cancer with cholesterol accumulation by depletion of cholesterol. More importantly, our genetic circuit dual-inputs and actuators are replaceable depend on cancer types, thus providing a general metabolic reprogramming strategy for cancer treatment.

Cell lines and cell culture
A375 malignant melanoma (A375) and 293T embryonic kidney (HEK293T) cells were purchased from Cell Resource Center of Peking Union Medical College (IBMS, CAMS/PUMC). The A2058 melanoma cell line was kindly provided by Dr. Fang from the Beijing Institute of Genomics. All cells were cultured in highglucose DMEM complete medium (Gibco), with 1% penicillin-streptomycin (Solarbio), and 10% fetal bovine serum (FBS) (Gibco) in a 5% CO2, 37°C environment.

Subcutaneous tumor formation
BALB/c-nu/nu mice, weighing on average 10 g, were purchased from Beijing Huafukang Bioscience Co. Inc. Lentiviral vectors for NVD-BM and pCDH were transfected into A375 cells (n = 4/group), and selected using 1 μg/ml puromycin for two weeks. We then collected 1×10 7 cells for subcutaneous injection into mice. After four weeks, we measured mouse tumors, and tumor tissues were fixed in 10% neutral buffered formalin for hematoxylin-eosin (HE) staining. The animal protocol was reviewed and approved by the Animal Ethical and Welfare Committee (AEWC) (approval No. IRM-DWLL-2019102).

Cell transfection and infection
All cells were transfected with plasmids using Lipofectamine 3000 (Invitrogen) to facilitate DNA transient transfections. For cell infection, viral vectors were transfected into HEK293T cells, after which lentiviruses were collected and concentrated. A375 and A2058 cells were incubated in viral media and polybrene, and incubated in 1 μg/ml puromycin for two weeks.

Real-time quantitative PCR assay
Total RNA was isolated from cells using Trizol reagent (Invitrogen), and cDNA was generated using the reverse transcription kit (Genecopoeia). Real-time PCR was performed using SYBR PCR mix (TransGen), and GAPDH was used as a reference housekeeping gene. We calculated expression fold changes using the 2 −ΔΔct method. All primers for qRT-PCR were designed using Snapgene and Primer 5 software (Supplementary Table 3).

MTT assay
Cells transfected with different plasmids were cultured in 96-well plates at 5×10 3 cells/well. After 48 h at 37°C in 5% CO2, 10 μl MTT solution was added to each well, and incubated for another 4 h. After this, the medium was removed and 110 μl/well DMSO added. Absorbance was measured at 570 nm on a plate reader (Thermo).

Real-time cell analysis assay
A375 and A2058 cell migration was analyzed by realtime cell analysis (RTCA). Both cell types were seeded into the upper chamber of an individual CIM-Plate-16 (5×10 3 cells/well) in 130 μL medium, without FBS. The upper chamber was placed on a lower CIM-Plate-16 chamber, which contained growth medium supplemented with 10% FBS (attractant). Impedance changes resulting from cell migration onto lower membranes were recorded every 5 min, and monitored over 36 h.

Sterol extraction and analysis
Sterol extraction and analysis were performed as recently described, with some modifications [41]. Approximately 1×10 7 cells/sample were collected and washed twice in phosphate buffer saline (PBS). Cells were then centrifuged, resuspended in 3 M HCl, and boiled for 3 min. This was performed three times. Cell fragments were collected by centrifugation, and pellets washed in distilled water. Cell precipitates were resuspended in a 2 M NaOH-methanol solution, and incubated at 60°C overnight. After cooling to room temperature, n-hexane was added to extract sterols, after which the n-hexane phase was removed in a vacuum desiccator. Eventually, desiccated sterol products were derived by N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA) at 30°C for 4 h to generate samples for GC-MS analysis.
Sterols were separated on an Agilent 6890 gas chromatograph (GC) (USA), coupled to a Waters time-of-flight mass spectrometer (TOF-MS) (USA). The GC was equipped with a DB-5 fused-silica capillary column (30 m × 0.25 mm internal diameter, film thickness of 0.25 μm (J&W Scientific, CA, USA). Ions were generated by a 70 eV electron beam in EI mode, at an ionization current of 40 μA. Mass spectra were acquired in the 50-800 m/z range. The ion source temperature was 250°C, and the injection site temperature was 260°C. The temperature was initially 70°C for 2 min, then increased by 30°C/min to 250°C, and finally increased to 280°C at 10°C/ min. 280°C was maintained for 15 min, and finally increased to 290°C at 5°C/min. The final temperature was maintained for 5 min. Standards (cholesterol and 7-DHC) were purchased from Sigma-Aldrich (USA).

Western blotting
Protein lysates from cells were lysed in RIPA buffer, with a protease inhibitor cocktail. For western blotting, proteins (20 μg/sample) were separated on 12% SDSpolyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membranes (Millipore). Membranes were blocked in 5% non-fat milk in TBST, and incubated with primary antibodies against Akt1

Luciferase assay
Cells were co-transfected with firefly luciferase and Renilla luciferase plasmids for 48 h. Growth medium was then removed, and monolayers washed gently and thoroughly in PBS, to avoid disruption. PBS was complexly removed. Then 1×PLB (Promega) was added to each well, and plates rocked at room temperature for 15 min. 100 μl LAR II (Promega) was added to luminometer tubes, including 20 μl cell lysates. Absorbance was measured at 560 nm. 100 µl Stop & Glo reagents (Promega) were then added to each tube, and absorbance measured at 560 nm. Luciferase results were determined by dividing absorbance at 560 nm by absorbance at 470 nm.

Bioinformatics analysis
R software was used to analyze bioinformatics data from GEO (https://www.ncbi.nlm.nih.gov/gds/), TCGA (https://www.cancer.gov/about-nci/organization/ ccg/research/structural-enomics/tcga) and GTEX (http://genome.ucsc.edu/gtex.html) databases. Firstly, we downloaded transcriptome data from melanoma patient samples from the TCGA database, and normal skin sample data from the GTEX database, which are skin cutaneous melanoma (813 normal tissues, 471 tumor tissues). Then we downloaded one GEO dataset, GSE3189, i.e. transcriptome melanoma data comprising 7 normal and 45 melanoma tissues. Secondly, to compare gene expression between normal and tumor samples, we normalized RNA-Seq data from these public databases, and used the limma R package to analyze gene expression patterns, to screen significantly differentially expressed genes (log Fold Change > 1, adj-p value < 0.05). The clusterProfiler R package was used for Gene Set Enrichment Analyses (GSEA). In addition, statistical significance parameters (p value) for all data were calculated by T tests.