The cell lines used in the present study were as follows: 293T cell line, the lung fibroblast cell lines MRC-5 and WI-38, the lung squamous cell carcinoma (LUSC) cell lines: MES-1 and H226, and the LUAD cell lines H358, A549, H1299 and H1650. All cell lines were purchased from Fuheng Biotechnology (Shanghai, China), and validated by short tandem repeat (STR) analysis. All the cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; HyClone, Cytiva) and 1% penicillin and streptomycin (Invitrogen; Thermo Fisher Scientific, Inc.). Ferrostatin-1 (Fer-1), ZVAD-FMK, necrostatin-1 (Nec-1), erastin and apoptozole (all from Sigma-Aldrich; Merck KGaA) were used to treat cells. For vectors, CREB-HA and CREB-short hairpin (sh)1/sh2 were acquired from previous studies (15,16) which were constructed by our laboratory (Shanghai Institute of Thoracic Oncology, Shanghai Chest Hospital, Shanghai Jiao Tong University, Shangai, China). EP300-Myc was purchased from Addgene, Inc. Empty vector, GFP-sh and EP300-sh1/sh2 cells were purchased from Shanghai Biolink Co., Ltd. CREB-Del-KID, CREB-Del-bZIP, EP300-Del-KIX, EP300-Del-Bromo and EP300-Del-CBP/p300-HAT were constructed using overlapping PCR and cloned into the pcDNA3.1(+) vector. All the vectors were transfected at a final concentration of 2 µg per 6-well plate using Lipofectamine™ 2000 transfection reagent (Invitrogen; Thermo Fisher Scientific, Inc.). 293T cells were transfected with FBS-free DMEM for 6 h and cultured in FBS-containing medium for another 24 h. Then, target cells were infected with lentivirus-containing-DMEM for 24 h, and the follow-up experiments were performed. All the aforementioned procedures were conducted at 37°C and 5% CO₂. All primers are summarized in Table SI. Tumorous and adjacent lung tissues of patients (mean age ± SD, 63.86±14.12 years; age range, 27.2–88.4 years; 158 males and 138 females) were recruited at the Shanghai Chest Hospital (Shanghai, China) from September 2013 to March 2018. The diagnosis of lung cancer was confirmed by computed tomography and histological analyses by doctors from Shanghai Chest Hospital. Informed written consent was obtained from all patients. The study was approved by the institutional Ethics Committee of Shanghai Chest Hospital.

IB was performed using conventional protocols that have been previously described (17). The primary antibodies used were: Anti-CREB (product nos. 9197 and 9104) and anti-GAPDH (product nos. 5174 and 51332) all from Cell Signaling Technology, Inc. (CST); anti-GPX4 (product code ab125066), anti-cysteinyl-tRNA synthetase (CARS) (product code ab126714), anti-nuclear factor, erythroid 2 like 2 (NRF2) (product code ab62352), anti-heat shock protein family B small member 1 (HSPB1) (also known as Hsp27; product code ab109376), anti-spermidine/spermine N1-acetyltransferase 1 (SAT1) (product code ab105220) all from Abcam; anti-Myc (product nos. 2276 and 2278), anti-HA (product nos. 2367 and 3724) all from CST; and anti-EP300 (product codes ab54984 and ab275378) and anti-SET domain bifurcated histone lysine methyltransferase 2 (SETDB2) (product code ab5517) all from Abcam. The secondary antibodies were anti-rabbit IgG, HRP-linked antibody (product no. 7074; CST) or anti-mouse IgG, HRP-linked antibody (product no. 7076; CST).

IHC was performed using conventional protocols that have been previously described (18). The primary antibodies included anti-CREB (product no. 9197; CST). Immunohistochemical staining was assessed by independent pathologists.

Total RNA was extracted using TRIzol (Invitrogen; Thermo Fisher Scientific, Inc.), and the RNA was reverse transcribed into complementary DNA using the PrimeScript RT Reagent Kit (TaKaRa Biotechnology Co., Ltd.) according to the manufacturer's instructions. qPCR was performed using a SYBR Premix Ex Taq (TaKaRa Biotechnology Co., Ltd.) kit for 40 cycles (95°C for 3 sec and 60°C for 30 sec). The relative expression of mRNA level was quantified using the 2^−ΔΔCq method (19). The primers are listed in Table SI.

Cells were plated at an initial density of 2×10⁵ cells/well and cultured for 24 h at 37°C and 5% CO₂. Cell viability was measured using a Cell Titer-Glo luminescent cell viability assay (cat. no. G7572; Promega Corporation) according to the manufacturer's instructions. MDA, 4-HNE and Fe²⁺ were measured using kits from Abcam (MDA kit, product code ab118970, 532 nm; 4-HNE kit, product code ab238538, 450 nm; Fe²⁺, product code ab83366, 593 nm) according to the manufacturer's instructions. The luminance or absorbance was measured by a reader from BioTek Instruments, Inc.

ChIP was performed using a kit from Active Motif according to the manufacturer's instructions. Cells (2×10⁷) were fixed using 1% formaldehyde at room temperature for 10 min, washed with PBS and lysed using lysis buffer from the kit. Following sonication, protein-DNA complexes were incubated with antibody-coupled protein G beads at 4°C overnight. The antibodies used were anti-CREB (product no. 9197; CST), anti-IgG (product no. 3900; CST) and anti-hepatocyte nuclear factor 4α (HNF4A; cat. no. PP-H6939-00; R&D Systems, Inc.). On the second day, DNA was eluted in 1% SDS/0.1 M NaHCO₃, reverse crosslinked at 65°C, purified via phenol/chloroform extraction and ethanol precipitation, and subjected to qPCR. The primers are listed in Table SI.

Luciferase activities were measured using a dual-luciferase kit (Promega Corporation) according to the the manufacturer's instructions. Wild-type (WT)- and mutant (Mut)-GPX4-promoter luciferase reporters were constructed using the pGL4.21 vector at our laboratory (Shanghai Institute of Thoracic Oncology, Shanghai Chest Hospital, Shanghai Jiao Tong University, Shanghai, China), and co-transfected into lung cancer cells with Renilla plasmids using Lipofectamine™ 2000 transfection reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Cells were transfected with FBS-free DMEM for 6 h and cultured in FBS-containing medium for another 42 h in 37°C and 5% CO₂. Then, cells were harvested and then lysed in the passive lysis buffer from the kit. The fluorescence intensity of the luciferase reporters was then examined and normalized to the Renilla luciferase activity.

The concentrations of CREB were evaluated by ELISA. The tissue samples were diluted (1:4) in a dilution buffer provided by the manufacturer, and 50 µl of each diluted sample was added to 96-well microtiter plates for analysis. A CREB ELISA kit (cat. no. TX11637) was purchased from Lichen Biotech, Ltd. ELISAs were performed in strict accordance with the manufacturers' guidelines. The signals were determined by measuring the absorbance at 450 nm with a microplate reader (BioTek Instruments, Inc.).

First, basement membrane extract (BME) was seeded in a 96-well plate at 50 µl/well and warmed at 37°C for 30 min. Then, cells were seeded on top of the plate coated with BME at a density of 10,000 cells/well. After 7 days, cells were stained with SYTOX Green (1 µM; Invitrogen, Thermo Fisher Scientific, Inc.) at 37°C for 30 min. Images were collected using a light microscope, and the relative spheroid size and amount (Φ>30 µm) were counted and calculated.

For co-IP, cell lysates (containing 2×10⁷ cells) were incubated with antibody-conjugated protein A/G magnetic beads (Thermo Fisher Scientific, Inc.) in Western/IP lysis buffer (Beyotime Institute of Biotechnology) at 4°C overnight. Immunoprecipitates isolated with the magnetic beads were washed five times with Western/IP lysis buffer before being subjected to IB. The antibodies used for co-IP were: Anti-Myc (product no. 2276; 1:100; CST), anti-HA (product no. 2367; CST, 1:100), anti-CREB (product no. 9104; 1:100; CST), anti-EP300 (product code ab54984; 1:100; Abcam) and anti-SETDB2 (product code ab13712; 1:50; Abcam).

Lipid ROS generation was measured by adding C11-BODIPY (Invitrogen; Thermo Fisher Scientific, Inc.) to a final concentration of 1.5 µM for 20 min before cell harvest. Lipid ROS-positive cells were finally assessed by a BD FACSCanto II flow cytometer.

Data concerning CREB expression in 515 LUAD and 59 normal lung specimens were obtained from The Cancer Genome Atlas (TCGA) and further analyzed using UALCAN database (http://ualcan.path.uab.edu) (20). CREB binding motif was acquired from JASPAR database (21). Potential CREB-binding methyltransferase/acetyltransferase was detected using the STRING database (22). Uniprot database was used to analyzed the domains in CREB and EP300 proteins (23).

The differences between groups were examined using Student's t-test, one-way ANOVA followed by Bonferroni's post hoc test, Fisher's exact test, χ² test and Spearman rank-correlation analysis. P<0.05 was considered to indicate a statistically significant difference. The statistical analysis was performed using Graphpad Prism 8 (GraphPad Software, Inc.) or SPSS version 21 (IBM Corp.).

After screening TCGA data using the UALCAN database, it was revealed that CREB was significantly upregulated in 515 LUAD specimens compared to 59 normal lung specimens (Fig. 1A). By using IB and IHC, it was revealed that compared to that of the lung fibroblasts MRC-5 and WI-38, and the LUSC cell lines MES-1 and H226, CREB was highly expressed in LUAD cell lines, especially in A549 and H1299 cells (Fig. 1B). After detecting the expression of CREB in 250 paired samples of LUAD patients by ELISAs, it was determined that the expression of CREB in the tumor tissues was higher than that in adjacent normal tissues (Fig. 1C and D). The tissues of patients 1–10 with the most significant increase in CREB according to the ELISA data of Fig. 1C and D, were also selected to analyze their CREB expression. It was revealed that CREB expression was significantly higher in the LUAD tissues than in the adjacent normal tissues; in addition, the upregulated degree of CREB expression in tumor tissue in Fig. 1E was more obvious than that in Fig. 1A. A similar phenomenon was observed in immunohistochemical analysis by measuring the tissues of no. 1–3 patients (Fig. 1F). CREB expression was also analyzed in squamous cell lung cancer (LUSC) and small-cell lung cancer (SCLC) tissues by ELISAs. It was revealed that CREB was not significantly increased in the LUSC and SCLC tissues compared to their adjacent tissues (Fig. S1). These data indicated that CREB was highly expressed in LUAD tissues.

In LUAD, CREB was reported to be a stimulator of cell growth (24,25). However, whether CREB is a regulator of cell death has not been investigated in detail. It was observed that knockdown of CREB inhibited cell viability, and 3D cell growth, whereas these effects could be reversed by the apoptotic inhibitor ZVAD-FMK and the ferroptotic inhibitor Fer-1. However, the CREB knockdown-induced effects could not be regulated by the necrotic inhibitor necrostatin-1 (Nec-1) (Fig. 2A). Compared to ZVAD-FMK, Fer-1 reversed the decrease of cell viability and 3D cell growth induced by CREB knockdown to a greater extent (Fig. 2A). It was also confirmed that CREB knockdown upregulated the level of the lipid peroxidation product MDA and the ferroptotic biomarkers lipid ROS and Fe²⁺, and these effects could not be reversed by ZVAD-FMK and Nec-1 (Fig. 2A and C). Moreover, it was revealed that ectopically expressed CREB could reverse the apoptotic stimulus Apoptozole- and ferroptosis stimulus erastin-induced decrease of cell viability and 3D cell growth (Fig. 2B), as well as erastin-induced MDA, Fe²⁺ increase and lipid ROS generation (Fig. 2B and C). These data indicated that knockdown of CREB concurrently induced apoptosis- and ferroptosis-like cell death.

To confirm the target of CREB in ferroptotic regulation and to explore whether their levels were regulated by CREB, several factors (including GPX4, ACSL4, CARS, NRF2, HSPB1 and SAT1) (4,26–30) were selected, which have been recently reported to exert important roles during the ferroptotic process in cancer. It was observed that among these factors, only the mRNA and protein levels of GPX4 were positively regulated by CREB in A549 and H1299 cells (Fig. 3A-C). In the tissues of LUAD patients 1–10, it was observed that the mRNA and protein levels of CREB and GPX4 were upregulated in the LUAD tissues compared to the adjacent normal tissues, and their levels in the LUAD tissues were significantly correlated with each other (Fig. 3D-G). However, parallel experiments revealed that the levels of SAT1 were not significantly increased in the LUAD tissues (Fig. 3D and F). The aforementioned data indicated that GPX4 expression was positively regulated by CREB.

A CREB motif (image from JASPAR database) was observed in the ~-104 to −93 promoter region of GPX4 and two luciferase reporters named WT-GPX4-promoter (containing the CREB motif) and Mut-GPX4-promoter (with the CREB motif deleted) were constructed (Fig. 4A). Via dual-luciferase experiments using these two reporters, it was observed that the promoting effect of CREB for GPX4 promoter was dependent on the CREB motif (Fig. 4B and C). ChIP experiments revealed that CREB bound to the ~-250 to −1 region of the GPX4 promoter (Fig. 4D and E), and the parallel experiments indicated that CREB could not bind to the ~-1000 to −250 region of the GPX4 promoter, and HNF4A and control IgG could not bind to the ~-250 to −1 region of the GPX4 promoter (Fig. 4D and E). As for tissues from no. 1 to no. 10 patients, it was revealed that CREB bound to the ~-250 to −1 region of the GPX4 promoter in both the LUAD and adjacent normal tissues. However, the binding intensity in the LUAD tissues was significantly higher than that in the adjacent normal tissues (Fig. 4F). These data indicated that CREB directly bound to a CREB motif in the GPX4 promoter.

Since histone modification, especially methylation and acetylation, is critical for transcription (31,32), it was further investigated whether methyltransferase or acetyltransferase played important roles in CREB-induced GPX4 transcription. After a STRING analysis detecting potential CREB-binding methyltransferase/acetyltransferase, it was revealed that EP300 had the highest possibility of binding with CREB (Fig. 5A). The co-IP experiments confirmed that CREB could interact with EP300, whereas an obvious interaction was not detected between CREB and SETDB2 (Fig. 5B). The UniProt database revealed (https://www.UniProt.org) that the CREB protein contains two important domains, KID and bZIP, and three important domains, KIX, Bromo and CBP/300-HAT, are located in the EP300 protein (Fig. 5C). Reciprocal co-IP experiments revealed that deletion of the bZIP or CBP/300-HAT domain completely abolished the interaction between CREB and EP300, suggesting that these two domains are essential for the CREB-EP300 interaction (Fig. 5D and E). Additionally, ChIP experiments revealed that CREB or EP300 knockdown (EP300 knockdown efficiency is presented in Fig. S2) reduced the H3K27Ac levels [a hallmark of open chromatin related to EP300 (33)] and inhibited the enrichment of CREB and EP300 around the CREB motif in the GPX4 promoter, and these effects could not be reversed by ectopically expressed EP300 or CREB (Fig. 5F and G), implying that CREB and EP300 were both critical for GPX4 transcription. Furthermore, it was observed that ectopically expressed CREB or EP300 increased the H3K27Ac levels, and stimulated the enrichment of CREB and EP300 around the CREB motif in the GPX4 promoter, whereas these effects could be blocked by deletion of the bZIP domain and CBP/p300-HAT domain, respectively (Fig. 5H-K). In addition, EP300 was knocked down in CREB-overexpressing cells and the cell viability and MDA levels were analyzed. It was determined that CREB could reverse the erastin-induced decrease in cell viability and the MDA increase; however, these effects were abolished by further knocking down EP300 (Fig. 5L and M). Collectively, these data demonstrated that EP300 was essential and had a promoting role in CREB-induced GPX4 transcription.

Other paired LUAD tissues were selected to investigate the correlation among CREB, GPX4, EP300 and 4-HNE, a reactive breakdown product of the lipid peroxides that execute ferroptosis. It was observed that the mRNA levels of CREB, GPX4, and EP300 were significantly higher in the tumor tissues than in the normal tissues, while the level of 4-HNE was significantly higher in the normal tissues than in the tumor tissues (Fig. 6A-D). In addition, the level of 4-HNE was negatively associated with the CREB, EP300 and GPX4 levels, whereas significant positive correlations were observed between GPX4 and CREB, EP300 and CREB, and EP300 and GPX4 (Fig. 6E-J).

Through the TCGA data from the UALCAN database, it was revealed that the CREB level was positively associated with the EP300 level in 515 LUAD specimens (P<0.001) (Fig. 6K). Furthermore, high expression of CREB was significantly associated with poor prognosis in 52 LUAD patients (P=0.008) (Fig. 6L). It was also determined that low 4-HNE levels were associated with more advanced tumor stages and larger tumor diameters (Fig. 6M and N), whereas high CREB, GPX4 and EP300 levels were associated with more advanced tumor stages and larger tumor diameters (Tables I and II). It was also revealed that high levels of CREB, GPX4 and EP300 were associated with advanced N factors (Table II). Furthermore, CREB, GPX4 and EP300 were not associated with patient age, sex or smoking habits (Tables I and II). Collectively, CREB, GPX4, EP300 and 4-HNE, which are related to lipid peroxidation, were closely related to tumor size and stage, and the tumors with a high degree of malignancy were more likely to have a low degree of lipid peroxidation.