MYCN mediates TFRC-dependent ferroptosis and reveals vulnerabilities in neuroblastoma

MYCN amplification is tightly associated with the poor prognosis of pediatric neuroblastoma (NB). The regulation of NB cell death by MYCN represents an important aspect, as it directly contributes to tumor progression and therapeutic resistance. However, the relationship between MYCN and cell death remains elusive. Ferroptosis is a newly identified cell death mode featured by lipid peroxide accumulation that can be attenuated by GPX4, yet whether and how MYCN regulates ferroptosis are not fully understood. Here, we report that MYCN-amplified NB cells are sensitive to GPX4-targeting ferroptosis inducers. Mechanically, MYCN expression reprograms the cellular iron metabolism by upregulating the expression of TFRC, which encodes transferrin receptor 1 as a key iron transporter on the cell membrane. Further, the increased iron uptake promotes the accumulation of labile iron pool, leading to enhanced lipid peroxide production. Consistently, TFRC overexpression in NB cells also induces selective sensitivity to GPX4 inhibition and ferroptosis. Moreover, we found that MYCN fails to alter the general lipid metabolism and the amount of cystine imported by System Xc(−) for glutathione synthesis, both of which contribute to ferroptosis in alternative contexts. In conclusion, NB cells harboring MYCN amplification are prone to undergo ferroptosis conferred by TFRC upregulation, suggesting that GPX4-targeting ferroptosis inducers or TFRC agonists can be potential strategies in treating MYCN-amplified NB.


Introduction
Neuroblastoma (NB) is the most common extracranial solid tumor in human infants, with 90% of cases diagnosed by the age of 5 1 . Genetically, frequent amplifications of the oncogene MYCN are identified in patients with NB, predicting poor prognosis independent of other factors 2 . Although MYCN inhibition leads to the suppression of NB in vitro, there are technical challenges in targeting MYCN clinically 3,4 . MYCN orchestrates multiple molecular pathways for cell growth, survival, metabolism, and death, dictating cancer cell fate and tumor progression 5,6 . In order to explore new treatments against MYCN-amplified NB, it is necessary to gain a deeper understanding of the biological functions mediated by MYCN.
Among the key cellular processes governed by MYCN, cell death is a puzzling phenomenon. On one hand, it is the cause of tumor growth under nutrient-deprived conditions whose adaptation promotes NB progression 7,8 ; on other hand, abnormal expression of MYCN leads to the direct activation of cell death pathways 9,10 . This contradiction indicates that MYCN finely regulates the balance between cell survival and cell death. Recently, researchers discovered a new type of programmed cell death, namely ferroptosis, which is characterized by intracellular iron disorder and accumulation of lipid peroxides in the cell membrane 11 . Cell culture and animal experiments have shown that induction of ferroptosis can eliminate selective tumor cells that are resistant to apoptosis, such as in kidney cancer 12 , breast cancer 13 , lung cancer 14 , etc., which provides a new venue for overcoming drug resistance. System X c (−) and GPX4 inhibition are the most common methods to induce ferroptosis. System X c (−) imports cystine as a key substrate for synthesizing glutathione (GSH), which is the reducing equivalent used by GPX4 to antagonize ferroptosis 15 . Although MYCN has a role in regulating cellular redox balance 16 , the potential effects of MYCN on ferroptosis remain unclear.
In this study, we identified a new regulatory mechanism of ferroptosis by MYCN. We found that NB cells overexpressing MYCN are particularly sensitive to ferroptosis induced by GPX4 inhibition rather than system X c (−) blockage. Moreover, MYCN amplification fails to cause significant changes in lipid species, suggesting that alternations of iron metabolism are mainly accounting for MYCN regulated ferroptosis. Therefore, we analyzed the expression of iron-related genes and revealed that TFRC, which encodes transferrin receptor 1 as an essential iron transporter on the cell membrane, plays a key role in the process of MYCN regulated ferroptosis. Similar to MYCN, TFRC significantly increases the intracellular iron load, and TFRC upregulation confers cell sensitivity to ferroptosis induced by GPX4 inhibition. Considering that MYCN and TFRC are both deregulated in a variety of tumor types, our finding points to a tumor vulnerability that can be therapeutically exploited.

MYCN is a regulator of GPX4-dependent ferroptosis
To study MYCN-dependent cellular processes in NB, we examined a collection of 147 genes essential for the survival of MYCN-amplified NB cells 17 through pathway analysis using the Molecular Signatures Database v7.2. We discovered that genes regulating cell death are significantly enriched in this collection (Fig. 1A). Furthermore, we performed RNA-seq and gene set enrichment analysis (GSEA) of six human NB cell lines (MYCN amplified SK-N-BE2, BE(2)-C, NLF, SK-N-DZ, and MYCN non-amplified SHEP, SK-N-AS), and found that MYCN expression is closely associated with levels of ferroptosis-related genes as summarized by Liang et al. 18 (Fig. 1B and Supplementary Table 1). Next, we compared the ferroptosis-related genes with those regulating MYCN-dependent cell survival (Supplementary Table 2), and identified GPX4 as the only overlapping gene (Fig. 1C). GPX4 is a well-established ferroptosis regulator neutralizing the oxidized lipid species 19 . To confirm the above results, multiple NB cell lines were treated respectively with two classic ferroptosis inducer, (1 S, 3 R)-RSL3 (hereafter RSL3 for short), a highly specific GPX4 inhibitor, and Erastin, a cystine-glutamate antiporter (system X c (−)) inhibitor. Interestingly, MYCN amplification significantly increases the sensitivity of NB cells to RSL3, whereas all NB cells exhibit similar resistance to Erastin ( Fig. 1D and Fig. S1A). Consistently, the effects of an alternative system Xc(−) inhibitor sulfasalazine (SAS) 20 are weakly associated with MYCN amplification (Fig. S1B, C). These findings imply that MYCN is a key regulator of ferroptosis, and GPX4 might play an essential role in this process. Indeed, high expression of GPX4 in NB tumor tissues predicts poor prognosis (Fig. 1E). However, RNA-seq data reveal that there is no significant correlation between GPX4 and MYCN expression ( Fig. 1F and Fig. S1D). Therefore, MYCN regulates GPX4-dependent ferroptosis likely through an unknown mechanism.

MYCN confers cell sensitivity to ferroptosis upon GPX4 inhibition
To confirm that MYCN is linked to ferroptosis, we ectopically expressed MYCN in SHEP, a MYCN nonamplified NB cell line, and found that MYCN expression significantly increases lipid peroxide contents. Conversely, lipid peroxides are reduced when MYCN is depleted in SK-N-BE2, a MYCN amplified NB cell line ( Fig. 2A, B). Next, MYCN overexpression significantly increases the sensitivity of SHEP cells to RSL3, but not Erastin. Moreover, MYCN induced sensitivity towards RSL3 can only be rescued by treating cells with ferroptosis inhibitors, but not other cell death inhibitors (Fig. 2C). These results are further supported by the data in MYCN amplified SK-N-BE2 cells with shRNA-mediated MYCN depletion (Fig. 2D).
The function of GPX4 is to convert toxic lipid peroxides to nontoxic alcohol lipids using the reducing equivalent glutathione. The system X c (−) imports cystine, the extracellular form of cysteine, for glutathione synthesis, therefore functioning upstream of GPX4. To test whether the intracellular cysteine level can be regulated by MYCN, we use mass spectrometry to identify altered metabolites upon MYCN overexpression, and found that MYCN does not significantly affect the intracellular cysteine level (Fig. 2E). Meanwhile, our metabolic analysis confirms that MYCN amplification elevates serine levels, which is consistent with previous findings such as by Xia et al. 21 . These results demonstrate that cellular cystine import is independent of MYCN. Together, we conclude that MYCN induces NB cell sensitivity to ferroptosis upon GPX4 inhibition, through a mechanism non-correlated with cysteine metabolism.

MYCN increases the intracellular iron load
Deregulated iron metabolism and abnormal lipid synthesis are both required to induce ferroptosis 22 . To this end, we examined the fatty acid content upon MYCN overexpression by mass spectrometry and found that MYCN has no significant influence on cellular fatty acid pools (Fig. 3A). Next, we analyzed the expression of   several ferroptosis-related genes as summarized by Hassannia et al. 23 , and identified TFRC as the most upregulated one both in RNA-seq data (Fig. 3B) and RT-PCR results (Fig. 3C). TFRC encodes transferrin receptor 1 (TfR1), which is the prevalent iron importer on the cell membrane. Consistently, MYCN upregulation enhances the protein level of TfR1 in NB cells (Fig. 3D), which parallels with TFRC levels in NB tumor tissues (Fig. 3E). Interestingly, we noticed that the expression of SLC40A1, which encodes the only known iron exporter ferroportin, is slightly but reproductively reduced upon MYCN overexpression (Fig. 3C), and RNA-seq analysis of NB tumor tissues also supports this argument (Fig. 3F). These results suggest that MYCN tends to increase the iron load in NB cells.
To confirm this, we tested the level of the total iron pool (ferric plus ferrous ion) upon MYCN manipulation. As expected, MYCN overexpression increases the iron level in SHEP cells, whereas MYCN depletion reduces the iron level in SK-N-BE2 cells (Fig. 3G, H). These results indicate that MYCN mediates GPX4-dependent ferroptosis likely via regulating iron metabolism.

TFRC photocopies MYCN to increase the labile iron pool
To confirm that MYCN-induced TFRC expression promotes iron uptake, we overexpressed TFRC in SHEP and depleted TFRC in SK-N-BE2 cells, respectively. Indeed, TFRC expression levels correlate with the cellular iron load (Fig. 4A, B). The Fe 3+ ion transported by TfR1 undergoes endocytosis, reduction to Fe 2+ , and importation to the cytosol composing the labile iron pool (LIP). Fe 2+ promotes the formation of lipid peroxides by Fenton reactions. Based on these knowledges, we reasoned that increased cellular iron load would result in LIP upregulation, thereby promoting lipid peroxide formation. To test this, we detected intracellular LIP levels with the calcein acetoxymethyl ester (Cal-AM) probe as previously described 24 . Indeed, the LIP levels are increased upon TFRC overexpression and deceased upon TFRC knockdown (Fig. 4C, D), which photocopy the effects of MYCN (Fig. 4E, F). Therefore, MYCN selectively sensitizes NB cells to ferroptosis by increasing the intracellular LIP.

TFRC mediates MYCN induced, GPX4-dependent ferroptosis
Our data have shown that MYCN upregulates TFRC expression, thereby regulating ferroptosis. Next, we explored whether TFRC mediates GPX4-dependent ferroptosis, in a similar way as MYCN. In MYCN nonamplified SHEP cells, we found that TFRC overexpression makes cells sensitive to ferroptosis induced by both System X c (−) and GPX4 inhibition (Fig. 5A), whereas TFRC knockdown only rescues ferroptosis induced by GPX4 inhibition, but not system X c (−) inhibition (Fig. 5B). In SK-N-BE2 cells with amplified MYCN, we also found that knockdown of TFRC rescues the GPX4-dependent but not X c − -dependent ferroptosis, and the rescue effects are substantially enhanced (Fig. 5C). Moreover, we found that the system Xc(−) inhibitor SAS upregulates TFRC expression, and TFRC-depleted, MYCN-overexpressed NB cells exhibit high resistance to SAS (Fig. S1E-G). The above results suggest that TFRC indeed mediates GPX4dependent ferroptosis, which can be further reinforced by MYCN amplification. Consistent with it, the lipid peroxide amount is significantly elevated with ectopic TFRC expression, and decreased upon TFRC knockdown (Fig. 5D, E).

TFRC is a direct downstream target of MYCN
Given that MYCN is a well-established transcription factor, we speculated that MYCN drives TFRC expression through direct transcriptional regulation. Indeed, we confirmed that TFRC expression is responsive to either up or down-regulation of MYCN (Figs. 3C and 6A). Then, we established an inducible MYCN-ER chimeric protein expression system in SHEP cells, in which nuclear translocation and activity of MYCN can be induced by 4-hydroxytamoxifen (4-HT) treatments. Upon 4-HT stimulation, TFRC expression is elevated at both mRNA and protein levels (Fig. 6B, C), paralleled with lipid peroxide formation in a time-dependent manner (Fig. 6D). To further dissect exactly how MYCN regulates TFRC expression, we performed ChIP-qPCR assays and found that MYCN mainly binds to the P1 and P2 regions, which are located~1.5 kb away from the transcription starting site (TSS) within the TFRC promoter (Fig. 6E, F). In addition, we cloned the TFRC promoter region from −1973 to +120 bp surrounding the TSS as shown in Fig. 6E and generated corresponding truncations. We found that MYCN-promoted luciferase signals are strongly attenuated when the region of −1973 to −1623 is eliminated (Fig. 6G), demonstrating that MYCN regulates TFRC expression by binding to this region. Therefore, we conclude that MYCN induces GPX4-dependent ferroptosis by upregulating TFRC as a direct downstream target.

Discussion
The MYC family member MYCN plays important role in the development of the central nervous system 25,26 , and its abnormal expression correlates with the poor prognosis of NB patients. Meanwhile, elevated activity of c-MYC, the predominant MYC isoform, also accelerates the progression of low-grade NB 27 , indicating that MYC family members share similar functions during NB development. Ferroptosis is a novel type of programmed cell death characterized by iron-dependent, unleashed lipid peroxidation, which has been identified in recent years. Cystine deprivation rendered by system X c (−) blockage, or redox imbalance by GPX4 inhibition are two classic approaches to induce ferroptosis 15 . Our study revealed that MYCN sensitizes NB cells to ferroptosis, which is largely dependent on GPX4 inhibition rather than cystine deprivation. This is mainly because the abnormal level of MYCN stimulates the expression of TFRC, increases the intracellular levels of iron load and LIP, ultimately leads to enhanced lipid peroxidation and a tendency towards ferroptosis.
One important function of MYCN in NB is upregulating glutamine transporters, such as ASCT2, thereby promoting glutamine addiction 28,29 , a hallmark of tumor  metabolic reprogramming. In NB cells, the glutamic acid level is decreased upon MYCN amplification (Fig. 2E), confirming that MYCN promotes glutamine consumption. It was reported that glutaminolysis and iron are both essential for Erastin-induced ferroptosis in an ischemia/ reperfusion model 30 , implying that MYCN might induce ferroptosis through system X c (−) in that context. However, our results reveal that MYCN induces ferroptosis in NB cells independent of system X c (−), suggesting that the molecular mechanism underlying MYCN-induced ferroptosis is highly context-dependent.
Notably, Floros et al. recently reported that MYCNamplified NB cells are more sensitive to the system X c (−) inhibitor sulfasalazine (SAS) 31 . Indeed, MYCN-amplified NB cells are slightly sensitive to SAS as reflected by their smaller IC50 values compared to non-amplified cells, although no statistical significance was achieved in our hands (Fig. S1C). This trend is seemingly contradictory to what we observed with Erastin. However, SAS equally inhibits other targets such as NFκB and reduced folate carrier 32,33 in addition to System X c (−), which may help explain this discrepancy. Interestingly, Floros et al. noticed that SAS treatment of MYCN-amplified NB in vivo upregulate TFRC expression through an uncharacterized mechanism, which can be reproduced by us in vitro (Fig. S1E). Since we demonstrated that enhanced TFRC expression is responsible for increased cellular sensitivity to ferroptosis, we reasoned that SAS may exert its effect through a similar mechanism. Indeed, TFRCdepleted, MYCN-overexpressed NB cells exhibit high resistance to SAS (Fig. S1F, G), demonstrating that SAS sensitizes MYCN-amplified NB cells to ferroptosis at least partially through TFRC upregulation. This argument can be further supported by the study from Floros et al. For instance, SAS is capable of inducing~70% cell death in MYCN-amplified NB cells. However, knockdown of two System X c (−) components SLC7A11 and SLC3A2 induce only~25% cell death in the same cell line 31 , confirming that SAS harbors considerable off-target effects. In conclusion, the complex effects of SAS treatments leading to ferroptosis cannot be solely linked to the sensitivity of NB cells to System X c (−) inhibition.
There are multiple major differences between our studies and the work by Floros et al. First, they were focused on targeting the redox balance of MYCN-amplified NB cells to induce ferroptosis, using different small molecule inhibitors such as buthionine-(S,R)-sulfoximine (BSO), sulfasalazine (SAS), and auranofin. All these inhibitors function to antagonize the ROS-scavenging activities in MYCN-amplified NB cells. In contrast, our work highlights TFRC as the major downstream target of MYCN to confer cell sensitivities to ferroptosis inhibitors. We particularly demonstrated that TFRC depletion greatly suppresses MYCN-mediated ferroptosis, as companied by the reduction of the intracellular iron pool. Moreover, TFRC overexpression phenocopies MYCN amplification to increase the intracellular iron pool and sensitize these tumor cells to ferroptosis. Given that ferroptosis requires both elevated iron levels and dampened ROS-neutralizing activities, these two studies compensate with each other in understanding the vulnerability of MYCN-amplified NB cells towards ferroptosis. Second, work by Floros et al. suggested that MYCN binds to the TFRC promoter and upregulates TFRC expression by analyzing relevant data in a public ChIP-seq database. Conversely, we performed ChIP-PCR analyses to identify the MYCN-binding region within the TFRC promoter, and further confirmed this result by luciferase reporter assays. Third, they found that MYCNamplified NB cells are sensitive to the System X c (−) inhibitor SAS, whereas our study suggested that these cells are resistant to the System X c (−) inhibitor Erastin.
Iron is an essential element for cell growth and survival, the deregulation of which determines cell fate. In the process of ferroptosis, iron acts as a catalyst for lipid peroxidation through Fenton reactions 23 . TfR1 encoded by TFRC is the main iron transporter in cells, and it has been reported that SLC39A14 is an additional iron transporter 34 , although our experiment shows that MYCN fails to regulate SLC39A14 expression (Fig. 3C). In addition, several classical MYCN binding motifs (CANNTG) are present in the TFRC promoter, and our ChIP-qPCR analysis reveals that MYCN precisely binds to a motif-containing region distal to TSS. The specific recognition and regulation of TFRC by MYCN is consistent with previous findings 35 . Recently, the Stockwell group reported that TfR1 clusters on the cell membrane can be used as a detection marker for ferroptosis 36 . They observed that TfR1 expression is increased during ferroptosis induction, but the mechanism is unknown. Our study supports the MYC signaling as an important candidate. Moreover, it has been reported that TfR1 depletion confers cell resistance to Erastin-induced ferroptosis under abnormal RAS activities 37 . However, we found that TFRC knockdown has no significant impact on Erastin-induced ferroptosis in MYCN-amplified cells. These results suggest that Erastin-induced ferroptosis harbors a regulatory mechanism independent of TFRC, which is likely decided by different oncogenic signals. Although MYC is one of the first identified oncogenes 38,39 , it can function to induce p53dependent cell death, similar to the case that oncogenic RAS can initiate cell senescence 40,41 . Whether and how different oncogenes cooperate to determine cancer cell fate is pivotal for developing new cancer therapeutics.

Cell culture
Human NB cell lines with MYCN amplification (SK-N-BE2, BE(2)-C, NLF, SK-N-DZ, Kelly) and without MYCN amplification (SHEP, SK-N-AS, SY5Y) were grown in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin-streptomycin, under 5% CO 2 at 37°C. Cell authentication was confirmed by the short tandem repeat analysis, and mycoplasma tests were weekly performed. Cell stocks were created within five passages, and experiments were completed within ten passages.

Constructs
MYCN and TFRC expression plasmids were constructed by cloning the open reading frame of the corresponding cDNA into the multiple cloning sites of the PCDH vector. shRNA sequences were cloned into PLKO.1 vector according to Addgene's protocol. For the luciferase assay, promoter sequences were cloned into pGL3 Luciferase Reporter Vectors (E1741, Promega, Wisconsin, USA).

RNA-seq analysis
Total RNA was extracted from each sample using the standard TRIzol protocol. The Agilent 2100 Bioanalyzer was employed to verify RNA quality. cDNA libraries were generated and processed for sequencing following the Illumina TrueSeq version 2 library preparation kit's protocol. The cDNA libraries were then sequenced by Illumina HiSeq 2500 with paired-end 2 × 75 bp reads using the HiSeq Control Software (version 2.0.10). The quality of raw reads was evaluated by FastQC, and reads alignment was performed using STAR (version 2.3.0). We used SAMTools to sort the aligned reads, and performed genelevel read quantifications with HTSeq (version 0.10.0).

Bioinformatic analyses
Gene expression correlation and survival analyses for NB patients (retrieved from GSE45547 42 and GSE49710 43 ) were conducted using the R2 Platform (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi), and the resulting data and p values were calculated according to online instructions. Gene Ontology (GO) analysis of MYCN-dependent genes 17

Viral production and transduction
HEK-293T cells were seeded on a 10-cm culture dish with 50% confluence. After attachment, cells were cotransfected with the lentiviral expression plasmid (10 μg), viral packaging plasmid (psPAX2, Addgene 12260, 5 μg), and envelope plasmid (pMD2.G, Addgene 12259, 2 μg) using PEI following the manufacturer's protocols. Viral supernatants were collected and filtered after transfection. For viral transduction, target cells were seeded and incubated with a medium supplemented with viral particles and 8 mg/mL polybrene for 6-8 h. Virus-infected cells were selected with corresponding antibiotics.

Cell viability assay
Cell viability was detected with CCK-8 according to the manufacturer's protocol. Briefly, 5 × 10 3 cells were seeded in 96-well plates per well and treated with different compounds. Twenty-four hours later, cells are incubated with CCK-8 for 2 h at 37°C. Viability values were obtained by a plate reader at 450 nm.

Total iron analysis
Total iron concentrations of cells were measured by a colorimetric ferrozine-based assay as previously reported 44 . Briefly, 5 × 10 5 cells were seeded in 6 cm dishes for 24 h, collected with 50 mM NaOH, and incubate at room temperature (RT) for 2 h. Iron ion was extracted with iron-releasing buffer (1.4 M HCl with 4.5% KMnO 4 , 1:1) at 60°C for 2 h, followed by incubation with 30 μL detecting buffer (6.5 mM Ferrozine, 6.5 mM Neocuproine, 2.5 M ammonium acetate, 1 M ascorbic acid) at RT for 30 min. Absorbance was measured with a 96-well plates reader at 550 nm. The iron contents were calculated based on FeCl 3 standard curves and normalized to protein concentration.

Labile iron pool (LIP) analysis
Calcein acetoxymethyl ester (Cal-AM) is a low-toxic and non-fluorescence dye that can easily pass through the cell membrane. Cal-AM emits fluorescence after being digested by esterase in living cells and quenched in the presence of Fe 2+ , which can be used as a LIP probe as previously described 24 . In this study, 2 × 10 5 cells were seeded on six-well plates for 24 h, after washing with PBS three times, and incubated with 20 nM Cal-AM in HBSS at 37°C for 30 min. Then, Cells were trypsinized, suspended, and analyzed using a BD Accuri C6 flow cytometer with a 488 nm laser on an FL1 detector.

Measurement of lipid peroxides
Cells were incubated with 200 nM C11-BODIPY in HBSS at 37°C for 30 min, and analyzed using a BD Accuri C6 flow cytometer with a 488 nm laser on an FL1 detector.

ChIP-qPCR
Assays were performed by following a modified Abcam's Chromatin immunoprecipitation protocol. Briefly, cells overexpressing biotinylated-tagged proteins (avi tag) were fixed in 1% formaldehyde at RT for 10 min and stopped by the addition of 0.125 M glycine. Cells were then harvested in ChIP Lysis Buffer (50 mM HEPES-KOH pH = 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 1X protease inhibitor cocktail). Lysates were sonicated and DNA sheared to an average length of 200-500 bp. Genomic DNA (Input) was prepared by treating aliquots of chromatin with RNase A, Proteinase K, and 65°C heat for decrosslinking, followed by ethanol precipitation. The resuspended pellet was incubated with magnetic streptavidin beads (Z5482, Promega, Wisconsin, USA) for 4 h rocking at 4°C, washed four times, and eluted from beads with SDS elution buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl pH = 8). Crosslinks were reversed by high salt incubation for 4 h at 65°C. ChIP DNA was purified by phenol-chloroform extraction, followed by quantitative real-time PCR analysis. Primers used for ChIP-qPCR:

Luciferase reporter assay
Dual-Luciferase Reporter Assay System (E1910, Promega, Wisconsin, USA) was used. Briefly, pGL3 vectors carrying TFRC promoter sequences were transfected into SHEP cells. Cells were then transferred to six-well plates with 2 × 10 5 per well. MYCN activity was induced with 40 nM 4-HT treatments. Luciferase signals were measured 48 h later. Firefly luciferase signals were divided by Renilla luciferase signals, and rescaled to set the control signals equal to 1.

Metabolic analysis
For GC-MS assays, cells were washed with pre-cold physiological saline buffer and harvested in HPLC grade methanol containing internal standards (Ribitol). After sonication and 20,000 g centrifugation at 4°C, the supernatants were transferred to new 1.5 mL tubes. The supernatants were then transferred into the rotary evaporator and evaporate completely. After treatment with derivatization reagent, samples were analyzed with GC-MS according to the manufacturer protocol. For LC-MS assays, cells were washed with pre-cold physiological saline buffer and harvested in HPLC grade methanol. Then, lipids were extracted with chloroform after 12,000 g centrifugation at 4°C. The lower phase was evaporated with nitrogen and esterified in 5% sulfuric acid in methanol. The extracts were analyzed by capillary electrophoresis time-of-flight mass spectrometry. Peak areas were normalized to standards and cell numbers.

Statistical analysis
Data were presented as mean ± SD. Statistical significance was analyzed using a two-tailed unpaired t-test and assessed by p values as reported in figure legends. Kaplan-Meier srvival analyses employed the log-rank test.