Methylation-Mediated Silencing of ATF3 Promotes Thyroid Cancer Progression by Regulating Prognostic Genes in the MAPK and PI3K/AKT Pathways

Abstract

Background:

Aberrant expression of oncogenes and/or tumor suppressor genes (TSGs) drives the tumorigenesis and development of thyroid cancer. We investigated the expression and function of a member of the activating transcription factor (ATF)/cAMP-responsive element-binding protein (CREB) transcription factor (TF) family, ATF3, in thyroid cancer.

Methods:

Data from 80 patients with papillary thyroid cancer (PTC) in the First Affiliated Hospital of Sun Yat-sen University and 510 PTC samples in The Cancer Genome Atlas thyroid cancer database were utilized for gene expression and prognosis analyses. The survival data were analyzed by Kaplan–Meier curves and Cox regression with adjustment for age, sex, multilocality, extrathyroidal extension, lymph metastases, and history of neoadjuvant treatment. DNA methylation was analyzed by methylation-specific polymerase chain reaction (PCR) and bisulfite sequencing PCR. TFs binding to ATF3 promoter were identified by DNA pull-down combined with mass spectrum assay, and confirmed by quantitative PCR (qPCR), luciferase reporter assay, and chromatin immunoprecipitation (ChIP)-qPCR. We conducted functional assays in vitro and in a xenograft mouse model to evaluate the function of ATF3 in thyroid cancer. Integrated analyses based on RNA sequencing, ChIP-seq, and CUT&Tag assays were performed to explore the mechanisms underlying the function of ATF3.

Results:

ATF3 was significantly downregulated in PTC and patients with low ATF3 expression had reduced progression-free survival (adjusted hazard ratio = 0.50 [CI 0.26–0.98], p = 0.043). DNA hypermethylation in ATF3 promoter disrupted the binding of SP1 and MYC-MAX, leading to inactivation of the gene. ATF3 functioned as a TSG by inhibiting the proliferation and mobility of thyroid cancer cells. And ATF3 regulated the expression of a number of genes by binding to the regulatory elements of them, particularly for genes in MAPK and PI3K/AKT pathways. Among these target genes, filamin C was positively regulated by ATF3 and associated with a more favorable thyroid cancer prognosis, while dual specificity phosphatase 10, fibronectin-1, tenascin C, and CREB5 were negatively regulated by ATF3 and associated with a poorer prognosis.

Conclusions:

We observed that the promoter DNA hypermethylation decreased the expression of ATF3, which in turn promoted the progression of thyroid cancer, at least partially, by directly regulating prognosis-related genes in the MAPK and PI3K/AKT pathways.

Introduction

Many patients with thyroid cancer harbor genetic variants in the MAPK and PI3K/AKT pathways, suggesting a key role of these two signaling pathways in the pathogenesis of thyroid cancer.^1,2 Epigenetic alterations have also been reported to be involved in thyroid cancer progression.³ However, the potential role of transcription factors (TFs) in the development and progression of thyroid cancer is largely unknown.

Activating transcription factor 3 (ATF3) belongs to the ATF/cAMP-responsive element-binding protein (CREB) TF family, which shares the basic-region leucine zipper (bZIP) domain and binds to the cAMP-responsive element site in promoters.^4,5 As a TF, ATF3 has been reported to homodimerize and repress transcription⁶ or form heterodimers through its bZIP region with other ATF/CERB family proteins to exert either transcriptional repressor or activator function, depending on the cellular condition and promoter context.^4,6,7 It has been characterized as an adaptive-response gene that is involved in cellular processes to adapt to extra- and/or intracellular changes and stress responses, immune regulation, and participates in human diseases, including several types of cancer.^7
–9

ATF3 may play an important role in cancer development and progression. ATF3 interacts with p53 and increases its stability by blocking its ubiquitination in the genotoxic response.¹⁰ However, increased ATF3 expression attenuates p53-dependent senescence and enhances the tumorigenic potential of skin squamous cancer.¹¹ Moreover, ATF3 has been reported to positively regulate several marker genes of epithelial–mesenchymal transition and mediate transforming growth factor-beta induced metastasis of breast cancer cells.^12,13 It was also reported to suppress RAS-stimulated tumorigenesis and act as a tumor suppressor by limiting the proliferation and invasion of lung and prostate cancer cells.^14
–16 We comprehensively investigated the regulation, function, and downstream targets of ATF3 in thyroid cancer.

Materials and Methods

Clinical samples

Eighty papillary thyroid cancer (PTC) tissue samples were obtained from a convenience sample of patients who underwent thyroid surgery between 2017 and 2019 in the First Affiliated Hospital of Sun Yat-sen University. Written informed consent was obtained from all patients for the use of their tissue samples for this study. The study was approved by the institutional ethics committee (IEC) of the First Affiliated Hospital of Sun Yat-sen University ([2021]109). The patients were all diagnosed with PTC based on postoperative histopathology, according to the criteria defined by World Health Organization.¹⁷

The study was carried out in accordance with the ethical principles of the Declaration of Helsinki. Criteria for patient inclusion and exclusion are described in the Supplementary Data and a patient recruitment flow diagram is shown in Supplementary Figure S1. The clinicopathological characteristics of the enrolled patients are reported in Supplementary Table S1. In addition, 510 PTC samples with a median follow-up time of 31 months from The Cancer Genome Atlas (TCGA) thyroid cancer (THCA) database were included to investigate the association of ATF3 and its target genes with thyroid cancer clinical outcomes.

Among these samples, 53 progressed during the follow-up period. Specifically, 27 samples had locoregional recurrence, 15 had distant metastasis, 4 had new primary tumor, 4 died without new tumor event, 2 had biochemical evidence of progression, and 1 had new tumor event.

Cell lines and functional analyses

Thyroid cancer cell lines BCPAP, TPC-1, KTC-1, ACT-1, C643, CAL62, Hth7, KHM-5M, TTA-1, and thyroid follicular epithelial cell line Nthy-ori 3–1 were purchased from the Cell Bank of Type Culture Collection of Chinese Academy of Science (Shanghai, China). These cell lines were used for detecting the methylation and expression of ATF3, gene overexpression or knockdown, and subjected to a series of functional assays, including cell proliferation, colony formation, cell migration and invasion, and luciferase reporter assay. The details of cell culture, DNA methylation and expression analyses, and cell-based functional assays are available in Supplementary Data and the primers used in the study are listed in Supplementary Table S2.

DNA pull-down and mass spectrometry

The methylated and nonmethylated biotinylated probes, nonbiotin-labeled DNA products (5 μg), and nuclear protein extracts (500 μg) were mixed and incubated at 4°C overnight. And 50 μL of Dynabeads™ MyOne™ Streptavidin (No. 65601; ThermoFisher) was added into the mixture and incubated at 4°C for 1 hour to form DNA–protein–beads complexes. Then the pulled-down proteins were identified using liquid chromatography-tandem mass spectrometry. The methods in detail are available in Supplementary Data.

High-throughput sequencing

For RNA sequencing (RNA-seq), total RNA isolated from cells was reversely transcribed into the complementary DNA library and sequenced by Illumina NovaSeq 6000 (Illumina, USA) with paired-end 150 bp. The chromatin immunoprecipitation (ChIP)-seq assay was performed using the SimpleChIP Plus Enzymatic Chromatin IP Kit (No. 9005; CST) according to the manufacturer's protocol.

The resultant DNA by anti-ATF3 antibody, along with the corresponding 2% input DNA sample, was used to construct the ChIP-seq libraries, and the library products were sequenced by Illumina NovaSeq 6000. The CUT&Tag assay was conducted using the NovoNGS^® CUT&Tag 3.0 High-Sensitivity Kit (No. N259-YH01; Novoprotein Scientific Inc.). The detailed methods of RNA-seq, ChIP-seq, and CUT&Tag assay are described in Supplementary Data.

Xenograft tumorigenicity assay

The animal experiments were approved by the IEC of the First Affiliated Hospital of Sun Yat-sen University ([2021]746). The four-week-old female nude mice were purchased from the animal center of Sun Yat-sen University and housed in specific pathogen-free facilities. The mice were randomly divided into two groups (n = 6 per group) according to random numbers generated by the RAND function of the Microsoft Excel software. The KHM-5M cells (5 × 10⁶ cells suspended in 100 μL phosphate-buffered saline) with ATF3 overexpression or the empty vector were subcutaneously injected into the flanks of each mouse.

None of the mice was excluded from analyses. Tumor size in each mouse was measured once a day using a caliper where the tumor was visible. The tumor volume was calculated using the formula (width² × length) × 0.5. Then, the mice were euthanized by cervical dislocation after anesthesia at two weeks after injection. Tumors were surgically extirpated, photographed, and weighed. The investigators performing randomization and tumor cell injection were aware of the group allocations. The investigators performing the data collection and analyses were different researchers in a blinded manner.

Statistical analyses

The Shapiro–Wilk normality test and Levene's test for homogeneity of variance were used to assess the model assumptions. Two-tailed Student's t-test was applied to determine the significance of the difference between two groups. The associations between ATF3 expression and clinicopathological characteristics were analyzed by using chi-squared tests. Survival analyses according to ATF3 status were conducted using a Kaplan–Meier curve and the Cox proportional hazard regression analysis (R package survminer version 0.4.9).

The following variables were adjusted in the cox regression analysis model: age, sex, multilocality, extrathyroidal extension, lymph metastases, and history of neoadjuvant treatment. And other analyses were performed using SPSS (version 26.0; IBM SPSS) or GraphPad Prism (version 7). Two-sided p-values <0.05 were considered statistically significant.

Results

DNA hypermethylation-induced decreased expression of ATF3 in thyroid cancer

To explore the role of ATF3 in thyroid cancer, we first investigated the relative expression of ATF3 by analyzing TCGA database. As shown in Figure 1A, ATF3 was found to be significantly downregulated in 510 PTC samples when compared with the 58 normal cases (p < 0.001). A similar pattern was observed in 58 paired PTC and adjacent normal samples (p < 0.001; Fig. 1B). Then, we detected the expression of ATF3 in paired PTC and adjacent normal tissues from 80 patients in our hospital and found that 54 cases (67.5%) showed a significant decrease (Fig. 1C).

FIG. 1.

DNA hypermethylation induced reduction in expression of ATF3 in thyroid cancer. (A) Relative expression of ATF3 in TCGA–THCA samples. (B) Relative expression of ATF3 in the paired PTC and adjacent normal tissues from TCGA–THCA samples. (C) qPCR assay to detect ATF3 transcriptional levels in 80 pairs of PTC tissues and adjacent normal tissues. For each sample, the relative ATF3 level in cancer tissue was normalized to that in the adjacent normal tissue. (D) Kaplan–Meier analysis of PFS stratified by ATF3 expression in TCGA–THCA samples. (E) Schematic plot illustrated the predicted CGIs in the promoter of ATF3. (F) The qPCR assay to detect ATF3 transcriptional levels in thyroid follicular epithelial cell Nthy-ori 3-1 and thyroid cancer cell lines. (G) MSP assay identified the status of ATF3 promoter methylation in the predicted CGIs in Nthy-ori 3-1 cell and thyroid cancer cell lines. M indicated methylation while U indicated unmethylation. (H) MSP assay was performed to determine the methylation levels of ATF3 CGI-1 and CGI-2 regions in C643 and TPC-1 cells treated with 5 μM DAC or DMSO for 5 days. (I) The relative ATF3 levels in C643 and TPC-1 cells treated with different concentration gradients of DAC or DMSO for 5 days were detected by qPCR. Data are presented as mean ± SD. The Student's t-test and the log-rank test were applied to determine the difference between the two groups and compare the statistical significance of Kaplan–Meier survival analysis, respectively. **p < 0.01 and ***p < 0.001. ATF3, activating transcription factor 3; CDS, coding sequence; CGIs, CpG islands; DAC, decitabine; DMSO, dimethyl sulfoxide; FPKM, fragments per kilobase million; GC, guanine and cytosine; HR, hazard ratio; MSP, methylation-specific PCR; PFS, progression-free survival; PTC, papillary thyroid cancer; qPCR, quantitative real-time polymerase chain reaction; SD, standard deviation; TCGA, The Cancer Genome Atlas; THCA, thyroid carcinoma; TSS, transcription start site; UTR, untranslated region.

Moreover, the level of ATF3 was significantly lower in patients with anaplastic thyroid cancers than PTC (Supplementary Fig. S2A). We next assessed the association of ATF3 expression with clinical outcomes of thyroid cancer, and found that patients with high ATF3 expression had a longer progression-free survival (PFS) with a hazard ratio (HR) of 0.53 ([95% confidence interval {CI} 0.29–0.98], p = 0.034; Fig. 1D), and this remained significant after adjusting multiple clinical factors (HR = 0.50 [CI 0.26–0.98], p = 0.043; Supplementary Fig. S2B). Taken together, ATF3 was significantly decreased in thyroid cancer and associated with a more favorable survival prognosis.

As promoter hypermethylation is one of the major mechanisms causing gene transcriptional repression, we next evaluated whether ATF3 was regulated by DNA methylation. After searching potential CpG islands (CGIs) in the promoter region of ATF3 by MethPrimer, we found that there were five predicted CGIs (Fig. 1E). Then, according to the ATF3 transcriptional levels in thyroid cancer cell lines (Fig. 1F), TPC-1 and C643 cells with relatively low ATF3 expression as well as BCPAP, Hth7, and TTA-1 with relatively high ATF3 expression, together with Nthy-ori 3–1 cell, were randomly selected for methylation-specific polymerase chain reaction (PCR) assay to determine methylation status.

The results revealed that CGI-1 and CGI-2 of ATF3 were hypermethylated in TPC-1 and C643 cells, while the other three CGIs were unmethylated in all the cells we tested (Fig. 1G). Then, we assessed the detailed methylation status of the individual CpG sites within ATF3 CGI-1 and CGI-2 regions by bisulfite sequencing PCR. Consistently, hypermethylation was observed in TPC-1 and C643 cells (Supplementary Fig. S3A, B).

To test whether promoter DNA methylation was involved in the epigenetic silencing of ATF3, TPC-1 and C643 cells were treated with DNA demethylating agent decitabine (DAC), and it was shown that the methylation levels of CGI-1 and CGI-2 in the ATF3 promoter were significantly decreased (Fig. 1H; Supplementary Fig. S3C, D), and the expression of ATF3 was correspondingly increased after DAC treatment (Fig. 1I). Consequently, promoter hypermethylation mediated the epigenetic silencing of ATF3 in thyroid cancer.

DNA hypermethylation prevented binding of SP1 and MYC-MAX to ATF3 promoter

To further explore the mechanism by which promoter hypermethylation represses the transcription of ATF3, DNA pull-down assays were performed to screen the potential binding TFs in CGI-1 and CGI-2. We initially identified 104 and 111 candidate proteins binding specifically to the unmethylated CGI-1 and CGI-2, respectively (Supplementary Table S3). Next, we performed an in silico analysis to predict the potential TFs binding to the two CGIs and then combined them with the candidates from DNA pull-down assays. As a result, we identified 7 and 12 TFs that might be capable of binding to ATF3 CGI-1 and CGI-2, respectively.

Among these candidates, we selected SP1, MYC, MAX, USF1, and USF2 for validation (Fig. 2A). Knockdown of SP1, MYC, and MAX significantly decreased the expression of ATF3 at the transcriptional level in two cell lines with ATF3 hypomethylation (Fig. 2B, C), whereas USF1 or USF2 knockdown had no such effect (Supplementary Fig. S4A, B). In addition, the regulatory effects of SP1, MYC, and MAX on ATF3 were validated by luciferase assays (Fig. 2D).

FIG. 2.

DNA methylation blocked the binding of SP1 and MYC-MAX to ATF3 promoter. (A) Schematic diagram of exploring the potential binding TFs to ATF3 promoter, and the relevant DNA pull-down and MS experimental procedures performed using C643 nucleoproteins, biotin-labeled methyl probe or unmethyl probe. The nonbiotin-labeled PCR products were served as NC. And the wild type specific binding proteins were identified by excluding the NC group binding proteins and the methyl probe binding proteins. The predicted TFs binding to the ATF3 CGI-1 or CGI-2 region were downloaded from JASPAR database. (B, C) Relative expression of ATF3, SP1, MYC, and MAX in KHM-5M (B) and BCPAP (C) cells treated with siRNA (20 nM) as indicated for 48 h was determined by qPCR. (D) Luciferase reporter assays in KHM-5M cells. Luciferase reporters containing ATF3 CGI-1 and CGI-2 region were treated with SP1-specific siRNA (20 nM) and MYC, MAX-specific siRNA (20 nM), respectively. (E, F) ChIP-qPCR assay detected the SP1 occupancy at CGI-1 region and MYC or MAX occupancy at CGI-2 region in hypomethylated KHM-5M cells (E) and in hypermethylated C643 cells (F) treated with 5 μM DAC or DMSO for 5 days. IgG served as control. Data represented the average ± SD of triplicate samples. *p < 0.05, **p < 0.01, ***p < 0.001, by two-tailed Student's t-test. ChIP, chromatin immunoprecipitation; IgG, immunoglobulin G; LC-MS/MS, liquid chromatography-tandem mass spectrometry; MAX, Myc-associated factor X; mRNA, messenger RNA; MS, mass spectrometry; MYC, MYC Proto-Oncogene; NC, negative control; SP1, specificity protein 1; TFs, transcription factors; USF1, upstream stimulatory factor 1.

As shown in Supplementary Figure S5, considering that there were several potential SP1 binding sites in ATF3 CGI-1 and several MYC-MAX binding sites in CGI-2, respectively, we next performed ChIP assay to confirm the binding of these TFs to ATF3 promoter, and found that there were significantly increased occupancies of SP1 in the CGI-1 and MYC and MAX in the CGI-2 region in the hypomethylated cell line KHM-5M (Fig. 2E). In contrast, in the cells with hypermethylation of ATF3, the occupancies of SP1, MYC, and MAX were relatively low but dramatically enriched after treatment by DAC (Fig. 2F). Collectively, ATF3 promoter hypermethylation blocked SP1 and MYC-MAX induced transcriptional activation of ATF3.

ATF3 functioned as a tumor suppressor in thyroid cancer

To explore the biological function of ATF3 in thyroid cancer, we manipulated the expression of ATF3 by gene overexpression and knockdown in several thyroid cancer cell lines (Supplementary Fig. S6A–D), and the effects on cell phenotype were investigated. The results revealed that overexpressing ATF3 significantly inhibited cell proliferation (Fig. 3A), colony formation (Fig. 3B), and migration and invasion of thyroid cancer cells (Fig. 3C–E).

FIG. 3.

ATF3 overexpression inhibited the aggressive behavior of thyroid cancer cells. (A, B) The cell proliferation (A) and colony formation assays (B) were performed in C643, CAL62, and KHM-5M with or without ATF3 overexpression. (C–E) Wound healing (C), cell migration (D), and invasion (E) assays were conducted in the C643, CAL62, and KHM-5M cells. (F, G) The effect of ATF3 on tumor formation in a nude mice xenograft model. The tumor growth curve (F) and representative images of tumors (G) from the ATF3-overexpressing and control groups (n = 6 per group). (H) Statistical analysis of the weight of tumors surgically excised from the xenograft mice. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, and ***p < 0.001, by two-tailed Student's t-test.

To address the function of ATF3 in vivo, we generated xenograft mice models using KHM-5M cells with or without ATF3 overexpression, and found that overexpressing ATF3 significantly inhibited tumor growth (Fig. 3F, G) and tumor weight (Fig. 3H) in nude mice. In contrast, inhibiting ATF3 significantly boosted the aggressive characteristics of thyroid cancer cells (Supplementary Fig. S6E–H). Collectively, these data suggested that ATF3 functions as a tumor suppressor in the progression of thyroid cancer.

ATF3 preferred to target and regulate genes in MAPK and PI3K/AKT signaling

To study the mechanism underlying the function of ATF3, ChIP-seq, CUT&Tag assay, and global transcriptome analysis (RNA-seq) were performed to explore its target genes and related signaling pathways. As shown in Figure 4A, the majority of ATF3 binding sites were near transcription start sites, suggesting that the major role of ATF3 was transcriptional regulation. The ChIP-seq identified 10,781 high-confidence binding peaks in 5091 genes, 31.75% of the peaks were located at promoters within a distance ≤1 kb (Fig. 4B).

FIG. 4.

ATF3 bound to and regulated genes in MAPK and PI3K/AKT pathways. (A) Plot represented the density of the ATF3 binding profiles within a 4 kb window centered on the TSS in ChIP-seq and CUT&Tag assays performed in CAL62 cells with ATF3 overexpression. (B, C) Pie chart showed the genomic distribution of ATF3 binding peaks by ChIP-seq (B) and CUT&Tag assay (C). (D) Putative consensus ATF3 motif enrichment and the corresponding p-values in ChIP-seq and CUT&Tag conducted by HOMER are shown. (E) Volcano plot displayed the differentially expressed genes in the ATF3-overexpressing CAL62 cells compared with the vector control cells by RNA-seq with three biological replicates. (F) Venn diagram shows the overlap genes among ChIP-seq, CUT&Tag, and RNA-seq. (G) Pathway enrichment analysis based on the Kyoto Encyclopedia of Genes and Genomes signaling pathway. (H, I) The heat map showed the expression of ATF3 target genes in the MAPK (H) and PI3K/AKT signaling (I) after ATF3 overexpression, respectively. Two-tailed Student's t-test was used to determine the significance of the difference between the two groups. RNA-seq, RNA sequencing.

Among 50,159 binding peaks in 6381 genes identified by CUT&Tag, 20.89% of the peaks resided in promoters within a distance ≤1 kb (Fig. 4C). Motif analysis of these peaks revealed significant enrichment of the consensus ATF3 binding motif (Fig. 4D), and a number of de novo motifs were also identified (Supplementary Fig. S7A). Furthermore, the previously reported ATF3 binding sites in the promoter of ATF3,¹⁸ PPP1R15A,¹⁹ and ID1 ²⁰ were captured, and successfully verified by ChIP-quantitative PCR (qPCR) (Supplementary Fig. S7B, C).

By performing RNA-seq, we identified 478 upregulated and 713 downregulated differentially expressed genes (DEGs) in cells with ATF3 overexpression when compared with the control cells (Fig. 4E). Importantly, there were 290 DEGs overlapped with the ATF3 target genes identified by ChIP-seq and CUT&Tag assays (Fig. 4F). Gene enrichment analysis revealed that the overlap genes were enriched in several cancer-related signaling pathways (Fig. 4G), including the MAPK and PI3K/AKT pathways, which are regarded as key pathways in thyroid cancer progression.¹ As shown in Figure 4H and I, there were 17 and 14 ATF3 target genes enriched in the MAPK and PI3K/AKT pathway, respectively.

Exploration of ATF3 targets that associated with thyroid cancer prognosis

To identify the key mediator in the progress of thyroid cancer induced by ATF3, we next analyzed the expression and prognostic value of the 17 and 14 genes in the two signaling pathways. Among the 17 MAPK pathway-related genes, dual specificity phosphatase 10 (DUSP10), which was negatively regulated by ATF3, was associated with poor PFS of thyroid cancer. While the filamin C (FLNC) gene, which was positively regulated by ATF3, was associated with favorable PFS of thyroid cancer (Fig. 5A; Supplementary Fig. S8A–C).

FIG. 5.

Two prognosis-related genes, DUSP10 and FLNC, in the MAPK pathway were directly regulated by ATF3. (A) Kaplan–Meier analysis of the association between DUSP10 or FLNC expression and PFS of thyroid cancer patients in TCGA–THCA samples. (B) Representative genomic overlay tracks show the intensity of ATF3 signals binding at DUSP10 and FLNC locus by ChIP-seq and CUT&Tag assays. The signals of ATF3 in ChIP-seq and CUT&Tag were indicated by pink peaks, while the input and IgG control are indicated as blue peaks. The direction of transcription is indicated by black arrows. (C) ChIP-qPCR analysis of ATF3 occupancy at DUSP10 and FLNC promoter in vector control and ATF3-overexpressing CAL62 C643, and KHM-5M cells, respectively. Data are presented as mean ± SD. **p < 0.01, and ***p < 0.001, by two-tailed Student's t-test. The log-rank tests were applied to compare the difference between the survival curves.

ChIP-seq and CUT&Tag assays showed that ATF3 was capable of binding to the promoters of DUSP10 and FLNC (Fig. 5B), which were further confirmed by independent ChIP-qPCR assays in three cell lines (Fig. 5C). Among the 14 PI3K/AKT pathway-related genes, we found that 3 genes, fibronectin-1 (FN1), tenascin C (TNC), and CREB5, were negatively regulated by ATF3 (Fig. 4I; Supplementary Fig. S8D–F) and associated with poor prognosis of thyroid cancer (Fig. 6A; Supplementary Fig. S8A). There were remarkable enrichments of ATF3 in the promoter of these three genes (Fig. 6B, C). Taken together, these data suggested that the tumor suppressing effects of ATF3 were mediated by downregulating DUSP10, FN1, TNC, and CREB5 and upregulating FLNC in thyroid cancer.

FIG. 6.

ATF3 targeted FN1, TNC, and CREB5 in the PI3K/AKT pathway. (A) Kaplan–Meier analysis of PFS of thyroid cancer patients stratified by FN1, TNC, or CREB5 expression in TCGA–THCA samples. (B) Representative genomic overlay tracks show the intensity of ATF3 signals binding at FN1, TNC, and CREB5 locus. The signals of ATF3 in ChIP-seq and CUT&Tag are indicated by pink peaks, while the input and IgG control are indicated as blue peaks. The direction of transcription is indicated by black arrows. (C) ChIP-qPCR analysis of ATF3 occupancy at FN1, TNC, and CREB5 promoter in vector control and ATF3-overexpressing CAL62, C643, and KHM-5M cells, respectively. Data are presented as mean ± SD. ***p < 0.001, by two-tailed Student's t-test. The log-rank tests were used to compare the difference between the survival curves. All the experiments were conducted three times and similar results were obtained.

Discussion

As key players in the process of gene expression, TFs are commonly dysregulated in human cancers through genetic or epigenetic mechanisms.²¹ In this study, we identified a member of the ATF/CREB TF family, ATF3, as a tumor suppressor and revealed that DNA hypermethylation disrupted the binding of SP1 and MYC-MAX to the promoter region of ATF3 and thus decreased its expression in thyroid cancer. Moreover, we found that ATF3 regulated the expression of a number of prognosis-related genes in the MAPK and PI3K/AKT pathways by binding to their promoters.

ATF3 was found to play contradictory roles in different types of human cancer. In our study, ATF3 was significantly downregulated in thyroid cancer and patients with low ATF3 expression had shorter PFS. Moreover, ATF3 could inhibit the growth, migration, and invasion of thyroid cancer cells. These findings are consistent with previous reports in lung and prostate cancer cells,^14
–16 suggesting a tumor suppressor role of ATF3 in these cancer types.

Although a recent study indicated that DNA methylation might involve in ATF3 regulation in breast cancer,²² the mechanism of how methylation regulates ATF3 was unknown. Here, we determined that DNA methylation of 2 CGIs in ATF3 promoter played a key role in regulating its expression in thyroid cancer, since DAC treatment significantly decreased the methylation level of the 2 CGIs and transcriptionally restored the level of ATF3. Since CpG methylation was found to directly repress transcription by preventing the binding of TFs to their recognition motifs,²³ we conducted a DNA pull-down assay and discovered that SP1 and MYC-MAX bound to the unmethylated CGI-1 and CGI-2 of ATF3, respectively.

SP1 is a ubiquitous transcriptional activator and involves in a variety of biological processes.²⁴ Previous studies reported that the transactivation ability of SP1 was suppressed by the DNA methylation of its target genes, such as KIBRA and ZNF132.^25,26 MYC generally exerted physiological functions by heterodimerizing with MAX through the HLHZip domain. The heterodimer complex of MYC-MAX bound to E-box motif CACGTG in the corresponding target genes, thereby regulating the transcription of a large number of specific genes.^27,28

In fact, the potential binding sites of MYC-MAX in the ATF3 promoter have been predicted,²⁹ and MYC was reported to transcriptionally activate ATF3 in preconditioning peripheral nerve injury³⁰ and serum-induced cell proliferation.³¹ We found that the binding of MYC-MAX to ATF3 promoter was abolished by hypermethylation of the promoter region, which was in accordance with the concept that DNA methylation of CpG site within the E-box may dramatically reduce MYC-MAX binding affinity to its target genes.^32,33

Furthermore, RNA-seq, ChIP-seq, and CUT&Tag assays were combined to demonstrate ATF3 direct targets. Interestingly, the enrichment analysis showed that the direct targets of ATF3 were enriched in the MAPK and PI3K/AKT pathways, both of which are classic oncogenic pathways in thyroid cancer tumorigenesis and progression. Importantly, among multiple targets of ATF3 in the 2 signaling pathways, we identified 5 genes associated with the prognosis of thyroid cancer, including DUSP10, FLNC, FN1, TNC, and CREB5.

FLNC is a dimeric actin-binding protein that can regulate remodeling of the actin cytoskeleton.³⁴ It was reported that FLNC can act as a scaffold to interact with stress signaling kinases MKK4 and MKK7.³⁵ In this study, we found that ATF3 directly targeted and upregulated the expression of FLNC, and high expression of FLNC was associated with a more favorable PFS of thyroid cancer. These data suggested the antitumor role of FLNC in thyroid cancer, which was consistent with previous findings in gastric cancer.^36,37

In contrast to FLNC, four target genes (DUSP10, FN1, TNC, and CREB5) were suppressed by ATF3 and high expression of each of these genes was associated with poorer PFS of thyroid cancer, suggesting a potential oncogenic role in thyroid cancer. DUSP10, as one of the MAPK phosphatase members, was involved in cell proliferation, differentiation, and migration in cancers.³⁸ The oncogenic role of DUSP10 has been reported in esophageal squamous cell carcinoma and p53 wild-type breast cancer.^39,40 A previous study showed that CREB5 significantly activated PI3K/AKT pathway in hepatocellular carcinoma and microRNA-206 inhibited tumor progression by suppressing the PI3K/AKT pathway through CREB5.⁴¹

FN1 was also reported to be a positive regulator of PI3K/AKT pathway in many diseases.^42
–44 TNC was also reported to activate the PI3K/AKT pathway through interacting with integrins.^45
–47 Jang and Chung⁴⁸ reported that TNC-induced AKT activation promoted cell survival and was significantly repressed by the PI3K inhibitor LY294002, indicating the PI3K-dependent of TNC-mediated AKT activation. Luo et al.⁴⁹ reported that ATF3 inhibited TNC-induced macrophage foam cell formation by repressing TLR-4. However, ATF3 was reported to promote macrophage migration and reverse macrophage polarization toward M2 phenotype by upregulating TNC.⁵⁰ These data suggest that the function of ATF3 is mediated, at least partially, by activating FLNC and suppressing DUSP10, FN1, TNC, and CREB5.

This study is subject to several limitations. First, there might be selection bias in the patient samples included from our institution. Second, the sample size was relatively small and the clinical data from the patients were retrospectively collected. Our findings need to be validated by further studies with large cohorts. Despite the above limitations, to the best of our knowledge, this is the first study to systematically explore the regulation and function of ATF3 in thyroid cancer.

In conclusion, we demonstrated that DNA hypermethylation in the CGIs of ATF3 promoter suppressed the expression of ATF3 by preventing the binding of SP1 and MYC-MAX to its promoter. The resulting decreased ATF3 promoted thyroid cancer progression, at least partially, by directly regulating prognostic genes in the MAPK and PI3K/AKT pathways. These findings may inform the pursuit of a novel therapeutic strategy for thyroid cancer.

Footnotes

Acknowledgments

The authors thank Shuai He and Kai Yu at Sun Yat-sen University Cancer Center for their excellent technical assistance.

Authors' Contributions

H.X. and R.L. designed the research. X.X., M.C., Y.C., J.X., K.J., Y.S., and L.Z. performed experiments. W.L., Y.L., and S.Y. collected tissue samples. X.X. and M.C. analyzed the data. X.X., R.L., and H.X. drafted the article. H.X. and R.L. had the primary responsibility for the whole content, and all authors read and approved the final version of the article.

Data Availability

The raw sequencing data of RNA-seq, ChIP-seq, and CUT&Tag in this study are available in Sequence Read Archive with an accession number PRJNA983555.

Author Disclosure Statement

The authors declare that they have no competing financial interests.

Funding Information

The study was supported by grants from the National Natural Science Foundation of China (Nos. 82072952 and 82271776), and the “100 Top Talents Program” of Sun Yat-sen University (No. Y61227).

Supplementary Material

Supplementary Data

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6

Supplementary Figure S7

Supplementary Figure S8

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

References

Xing

. Molecular pathogenesis and mechanisms of thyroid cancer. Nat Rev Cancer, 2013; 13(3):184–199; doi: 10.1038/nrc3431.

Nikiforov

, Nikiforova

. Molecular genetics and diagnosis of thyroid cancer. Nat Rev Endocrinol, 2011; 7(10):569–580; doi: 10.1038/nrendo.2011.142.

Zhang

, Duan

, Wang

, et al. Epigenetic signature associated with thyroid cancer progression and metastasis. Semin Cancer Biol, 2022; 83:261–268; doi: 10.1016/j.semcancer.2021.03.026.

Hai

, Liu

, Coukos

, et al. Transcription factor ATF cDNA clones: An extensive family of leucine zipper proteins able to selectively form DNA-binding heterodimers. Genes Dev, 1989; 3(12b):2083–2090; doi: 10.1101/gad.3.12b.2083.

Deutsch

, Hoeffler

, Jameson

, et al. Structural determinants for transcriptional activation by cAMP-responsive DNA elements. J Biol Chem, 1988; 263(34):18466–18472.

Chen

, Liang

, Whelan

, et al. ATF3 and ATF3 delta Zip. Transcriptional repression versus activation by alternatively spliced isoforms. J Biol Chem, 1994; 269(22):15819–15826.

Hai

, Wolfgang

, Marsee

, et al. ATF3 and stress responses. Gene Expr, 1999; 7(4–6):321–335.

Thompson

, Xu

, Williams

. ATF3 transcription factor and its emerging roles in immunity and cancer. J Mol Med (Berl), 2009; 87(11):1053–1060; doi: 10.1007/s00109-009-0520-x.

Gilchrist

, Thorsson

, Li

, et al. Systems biology approaches identify ATF3 as a negative regulator of Toll-like receptor 4. Nature, 2006; 441(7090):173–178; doi: 10.1038/nature04768.

10.

Yan

, Lu

, Hai

, et al. Activating transcription factor 3, a stress sensor, activates p53 by blocking its ubiquitination. Embo J, 2005; 24(13):2425–2435; doi: 10.1038/sj.emboj.7600712.

11.

, Nguyen

, Dziunycz

, et al. Opposing roles for calcineurin and ATF3 in squamous skin cancer. Nature, 2010; 465(7296):368–372; doi: 10.1038/nature08996.

12.

Yin

, Wolford

, Chang

, et al. ATF3, an adaptive-response gene, enhances TGF{beta} signaling and cancer-initiating cell features in breast cancer cells. J Cell Sci, 2010; 123(Pt 20):3558–3565; doi: 10.1242/jcs.064915.

13.

Yin

, Dewille

, Hai

. A potential dichotomous role of ATF3, an adaptive-response gene, in cancer development. Oncogene, 2008; 27(15):2118–2127; doi: 10.1038/sj.onc.1210861.

14.

Wang

, Xu

, Ding

, et al. Loss of ATF3 promotes Akt activation and prostate cancer development in a Pten knockout mouse model. Oncogene, 2015; 34(38):4975–4984; doi: 10.1038/onc.2014.426.

15.

Jan

, Tsai

, Yang

, et al. Adenylate kinase-4 is a marker of poor clinical outcomes that promotes metastasis of lung cancer by downregulating the transcription factor ATF3. Cancer Res, 2012; 72(19):5119–5129; doi: 10.1158/0008-5472.Can-12-1842.

16.

, Wolfgang

, Hai

. Activating transcription factor 3, a stress-inducible gene, suppresses Ras-stimulated tumorigenesis. J Biol Chem, 2006; 281(15):10473–10481; doi: 10.1074/jbc.M509278200.

17.

Lloyd

, Osamura

, Klöppel

, et al. WHO Classification of Tumours of Endocrine Organs, 4th ed. International Agency for Research on Cancer, Lyon: France; 2017.

18.

Wolfgang

, Liang

, Okamoto

, et al. Transcriptional autorepression of the stress-inducible gene ATF3. J Biol Chem, 2000; 275(22):16865–16870; doi: 10.1074/jbc.M909637199.

19.

Yoo

, Kim

, Oh

, et al. Transcriptome analysis upon potassium usnate exposure reveals ATF3-induced apoptosis in human gastric and colon cancer cells. Phytomedicine, 2021; 91:153655; doi: 10.1016/j.phymed.2021.153655.

20.

Kang

, Chen

, Massagué

. A self-enabling TGFbeta response coupled to stress signaling: Smad engages stress response factor ATF3 for Id1 repression in epithelial cells. Mol Cell, 2003; 11(4):915–926; doi: 10.1016/s1097-2765(03)00109-6.

21.

Bushweller

. Targeting transcription factors in cancer—From undruggable to reality. Nat Rev Cancer, 2019; 19(11):611–624; doi: 10.1038/s41568-019-0196-7.

22.

, Sun

, Jiang

. ATF3 demethylation promotes the transcription of ARL4C, which acts as a tumor suppressor in human breast cancer. Onco Targets Ther, 2020; 13:3467–3476; doi: 10.2147/ott.S243632.

23.

Yin

, Morgunova

, Jolma

, et al. Impact of cytosine methylation on DNA binding specificities of human transcription factors. Science, 2017; 356:6337; doi: 10.1126/science.aaj2239.

24.

Solomon

, Majumdar

, Martinez-Hernandez

, et al. A critical role of Sp1 transcription factor in regulating gene expression in response to insulin and other hormones. Life Sci, 2008; 83(9–10):305–312; doi: 10.1016/j.lfs.2008.06.024.

25.

Jiang

, He

, Wang

, et al. Epigenetic silencing of ZNF132 mediated by methylation-sensitive Sp1 binding promotes cancer progression in esophageal squamous cell carcinoma. Cell Death Dis, 2018; 10(1):1; doi: 10.1038/s41419-018-1236-z.

26.

Schelleckes

, Schmitz

, Ciarimboli

, et al. Promoter methylation inhibits expression of tumor suppressor KIBRA in human clear cell renal cell carcinoma. Clin Epigenetics, 2017; 9:109; doi: 10.1186/s13148-017-0415-6.

27.

Cascón

, Robledo

. MAX and MYC: A heritable breakup. Cancer Res, 2012; 72(13):3119–3124; doi: 10.1158/0008-5472.Can-11-3891.

28.

Blackwood

, Eisenman

. Max: A helix-loop-helix zipper protein that forms a sequence-specific DNA-binding complex with Myc. Science, 1991; 251(4998):1211–1217; doi: 10.1126/science.2006410.

29.

Liang

, Wolfgang

, Chen

, et al. ATF3 gene. Genomic organization, promoter, and regulation. J Biol Chem, 1996; 271(3):1695–1701; doi: 10.1074/jbc.271.3.1695.

30.

Shin

, Kwon

, Lee

, et al. Role of Myc Proto-Oncogene as a transcriptional hub to regulate the expression of regeneration-associated genes following preconditioning peripheral nerve injury. J Neurosci, 2021; 41(3):446–460; doi: 10.1523/jneurosci.1745-20.2020.

31.

Tamura

, Hua

, Adachi

, et al. Stress response gene ATF3 is a target of c-myc in serum-induced cell proliferation. Embo J, 2005; 24(14):2590–2601; doi: 10.1038/sj.emboj.7600742.

32.

Wang

, Hashimoto

, Zhang

, et al. MAX is an epigenetic sensor of 5-carboxylcytosine and is altered in multiple myeloma. Nucleic Acids Res, 2017; 45(5):2396–2407; doi: 10.1093/nar/gkw1184.

33.

Prendergast

, Ziff

. Methylation-sensitive sequence-specific DNA binding by the c-Myc basic region. Science, 1991; 251(4990):186–189; doi: 10.1126/science.1987636.

34.

Feng

, Walsh

. The many faces of filamin: A versatile molecular scaffold for cell motility and signalling. Nat Cell Biol, 2004; 6(11):1034–1038; doi: 10.1038/ncb1104-1034.

35.

Nakagawa

, Sugahara

, Yamasaki

, et al. Filamin associates with stress signalling kinases MKK7 and MKK4 and regulates JNK activation. Biochem J, 2010; 427(2):237–245; doi: 10.1042/bj20091011.

36.

Cao

, Li

, Chen

, et al. Tripartite motif-containing 54 promotes gastric cancer progression by upregulating K63-linked ubiquitination of filamin C. Asia Pac J Clin Oncol, 2022; 18(6):669–677; doi: 10.1111/ajco.13747.

37.

Kokate

, Dixit

, Poirah

, et al. Testin and filamin-C downregulation by acetylated Siah2 increases invasiveness of Helicobacter pylori-infected gastric cancer cells. Int J Biochem Cell Biol, 2018; 103:14–24; doi: 10.1016/j.biocel.2018.07.012.

38.

Jiménez-Martínez

, Stamatakis

, Fresno

. The Dual-Specificity Phosphatase 10 (DUSP10): Its role in cancer, inflammation, and immunity. Int J Mol Sci, 2019; 20(7):1626; doi: 10.3390/ijms20071626.

39.

Chen

, Yao

, Di

, et al. MiR-450a-5p inhibits autophagy and enhances radiosensitivity by targeting dual-specificity phosphatase 10 in esophageal squamous cell carcinoma. Cancer Lett, 2020; 483:114–126; doi: 10.1016/j.canlet.2020.01.037.

40.

Hrstka

, Bouchalova

, Michalova

, et al. AGR2 oncoprotein inhibits p38 MAPK and p53 activation through a DUSP10-mediated regulatory pathway. Mol Oncol, 2016; 10(5):652–662; doi: 10.1016/j.molonc.2015.12.003.

41.

Chi

, Gong

, Xin

, et al. microRNA-206 prevents hepatocellular carcinoma growth and metastasis via down-regulating CREB5 and inhibiting the PI3K/AKT signaling pathway. Cell Cycle, 2022; 21(24):2651–2663; doi: 10.1080/15384101.2022.2108275.

42.

Yoshihara

, Kajiyama

, Yokoi

, et al. Ovarian cancer-associated mesothelial cells induce acquired platinum-resistance in peritoneal metastasis via the FN1/Akt signaling pathway. Int J Cancer, 2020; 146(8):2268–2280; doi: 10.1002/ijc.32854.

43.

Cho

, Horzempa

, Jones

, et al. The fibronectin III-1 domain activates a PI3-Kinase/Akt signaling pathway leading to αvβ5 integrin activation and TRAIL resistance in human lung cancer cells. BMC Cancer, 2016; 16:574; doi: 10.1186/s12885-016-2621-6.

44.

Han

, Khuri

, Roman

. Fibronectin stimulates non-small cell lung carcinoma cell growth through activation of Akt/mammalian target of rapamycin/S6 kinase and inactivation of LKB1/AMP-activated protein kinase signal pathways. Cancer Res, 2006; 66(1):315–323; doi: 10.1158/0008-5472.Can-05-2367.

45.

Zhang

, Xu

, Hu

, et al. Tenascin C promotes glioma cell malignant behavior and inhibits chemosensitivity to paclitaxel via activation of the PI3K/AKT signaling pathway. J Mol Neurosci, 2021; 71(8):1636–1647; doi: 10.1007/s12031-021-01832-8.

46.

Dang

, Zhang

, Tian

, et al. MicroRNA-218 inhibits the malignant phenotypes of glioma by modulating the TNC/AKT/AP-1/TGFβ1 feedback signaling loop. Int J Mol Med, 2021; 48(5):205; doi: 10.3892/ijmm.2021.5038.

47.

Ding

, Jin

, Liu

, et al. Tenascin-C promotes the migration of bone marrow stem cells via toll-like receptor 4-mediated signaling pathways: MAPK, AKT and Wnt. Mol Med Rep, 2018; 17(6):7603–7610; doi: 10.3892/mmr.2018.8855.

48.

Jang

, Chung

. Tenascin-C promotes cell survival by activation of Akt in human chondrosarcoma cell. Cancer Lett, 2005; 229(1):101–105; doi: 10.1016/j.canlet.2004.12.012.

49.

Luo

, Wang

, Qiao

, et al. ATF3 inhibits tenascin-C-induced foam cell formation in LPS-stimulated THP-1 macrophages by suppressing TLR-4. J Atheroscler Thromb, 2015; 22(11):1214–1223; doi: 10.5551/jat.28415.

50.

Sha

, Zhang

, et al. ATF3 promotes migration and M1/M2 polarization of macrophages by activating tenascin-C via Wnt/β-catenin pathway. Mol Med Rep, 2017; 16(3):3641–3647; doi: 10.3892/mmr.2017.6992.