Extrachromosomal circular DNA containing DTX1 promotes cell growth in hydroquinone-induced malignantly transformed cells by regulating the transcription of DTX1

Abstract

Background

Extrachromosomal circular DNA (eccDNA), a novel class of DNA with a circular topological structure, is present in a variety of cancer cells and tissues and may play broad roles in processes ranging from aging to cancer cell heterogeneity through multiple mechanisms. EccDNA has been characterized by profile, structure and function in several prominent studies but its effect on hydroquinone (HQ)-induced malignantly transformed cells (TK6-HQ) is still elusive.

Methods

Circle-seq was applied to determine the eccDNA counts and characteristics of TK6-HQ cells. DNA-fluorescence in situ hybridization was used to measure the abundance of eccDNA_DTX1. Differential gene expression analysis was carried out by RNA-seq. Gene expression was quantified by wertern blot and qPCR. Decircularization of eccDNA_DTX1 was achieved by CRISPR/Cas9. Tumorigenicity was evaluated by xenograft assay in BALB/c nude mice.

Results

In this study, we characterized the structure of eccDNAs and the function of DTX1-containing eccDNA (eccDNA_DTX1) in TK6-HQ cells. A total of 669,179 eccDNAs were identified, including 901 eccDNAs with different counts. Most of the eccDNAs were < 1000 bp in length and were enriched in four periodic peaks starting at 186 bp with an interval of ~ 180 bp. The genomic distribution of eccDNAs confirmed that eccDNAs could be observed across all chromosomes and had greater enrichment on chromosomes 17, 19, 20, and 22, with abundant Alu repeat elements, introns and CpG islands. By combining the results of the integrated circle-seq analysis of eccDNAs with those from the RNA-seq analysis (differentially expressed genes, 1064 upregulated and 427 downregulated), the authors showed that the transcription of 20 potential coding genes might be driven by eccDNAs. Finally, the knockdown of eccDNA_DTX1 by CRISPR/Cas9 inhibited the growth of TK6-HQ cells in vitro and in vivo by inhibiting the transcription of DTX1 and promoting ferroptosis, and ferroptosis inhibior, Ferrostatin-1, abrogated the proliferation inhibition of eccDNA_DTX1 knockdown.

Conclusions

EccDNA_DTX1 promotes cell growth in hydroquinone-induced malignantly transformed cells by regulating the transcription of DTX1 and ferroptosis. This study profiles eccDNA characteristics and defines the role and mechanism of eccDNA_DTX1 for the first time, shedding new light on the relationship between eccDNAs and carcinogenesis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-024-13177-7.

Background

Extrachromosomal circular DNAs (eccDNAs), which reside in nuclei, were first discovered by Alix Bassel and Yasuo Hoota in 1964 [1] and are referred to as Double Minutes. EccDNAs have no 5’ or 3’ end, which facilitates their resistance to degradation by DNA exonucleases. The function of eccDNAs has not been studied until recently because of technological limitations. Currently, eccDNAs have emerged as promising small molecules in many fields because of their special topological structure and several identified functions, such as driving tumor development, and genetic and cellular heterogeneity [2–4]. EccDNA has been found to drive tumor evolution and genetic heterogeneity by promoting the expression of the oncogenes MYC and EGFR in 17 different cancer types [5]. The size of eccDNAs ranges from tens of thousands of bases, most of which are smaller than 1,000 bases [6–9], and 99% of the eccDNAs are shorter than 25 kb; the smaller the eccDNAs are, the more flexible they are. More than 50% of eccDNAs originate from different chromosomal sites, including genic, introgenic, and pseudogenic regions [10, 11], and from transcriptional elements [9], although some studies have shown that they primarily originate from the 5’ untranslated region (5’UTR) and CpG island regions [6]. Thus, some eccDNAs are long enough to replicate independently of cell mitosis and are present in tumor tissue, where they most frequently harbor genes, such as oncogenes or genes related to drug resistance in tumors, and transcriptional elements, including promoters and enhancers [12, 13].

EccDNAs can be generated by many different pathways through different mechanisms. Several models for eccDNA biogenesis, including the breakage-fusion-bridge (BFB) model, the chromothripsis model, the episome model, the translocation-deletion-amplification model, and the DNA replication and transcription model, have been established [3, 6, 14, 15], during which DNA breakage, DNA recombination, and DNA rearrangement might play critical roles [14, 16–18]. The circularization of DNA fragments from chromosomal breaks results in the formation of eccDNAs, and the apoptosis of lymphoblastoid cells induced by chemotherapy significantly promotes eccDNA production [16]; apoptotic eccDNAs can play pivotal roles in immunoregulation [19], but eccDNAs are not byproducts of apoptosis-driven fragmentation [9]. However, a recent study revealed that retrotransposons hijack alternative end-joining methods for DNA replication and eccDNA biogenesis, indicating that eccDNA research has promising potential [20]. Most carcinogens are mutagens that lead to DNA damage. The accumulation of eccDNA has been observed after exposure to the carcinogen, 7,1-dimethylbenz[a]anthracene, and hydroxyurea [21], which are DNA replication inhibitors. Chromothripsis is a consequence of severe DNA damage and can be induced by exogenous stimuli such as environmental mutagens [18].

These new findings may help to elucidate the role of eccDNAs in the stress response and cancer development. Carcinogenesis is the process of cancer formation and is often referred to as an evolutionary process as it requires the mutation and selection of mutated cells at every step of the multistage process [22]. Cell pressure induced by carcinogens may continuously occur, and exogenous stimuli can drive the formation of eccDNA, which remodels the genome [2, 13] with high efficacy and provides survival advantages while facing challenges by encoding all of the currently defined transcripts, including those involved in the genome maintenance network [4, 11, 23, 24]. Hydroquinone (HQ), a human mutagen, is a key component of benzene-derived metabolites [25, 26]. Studies from our laboratory and other groups have shown that HQ can affect numerous biological processes in vitro and in vivo, including reactive oxygen species (ROS) levels, gene expression [27–29] and DNA damage [30]. Cells have evolved various mechanisms to survive when they are experiencing unfavorable conditions, alert neighboring cells to the imminent danger, and set in motion an orderly sequence of events that combats these danger signals. Many studies have shown that organisms develop an integrated signaling and genome maintenance network [31] that consists of DNA damage sensors, mediators, transducers and effectors. In previous studies, we found that HQ treatment could result in DNA damage and increase the expression of the DNA sensor protein PARP-1 and the mediator AATF. Recent research has revealed that chromothripsis may drive the evolution of gene amplification in cancer [18]. We hypothesized that eccDNAs have important functions in HQ-induced carcinogenesis, and it is necessary, though still a large challenge, to clarify the role of eccDNAs in tumorigenesis in detail.

Based on our previous research on eccDNAs [11], genomic and transcriptomic approaches were applied to determine the landscape of eccDNA in HQ-induced malignantly transformed TK6 cells (HQ-TK6). We identified the characteristics of eccDNAs and characterized 20 eccDNA-promoting genes, among which we focused on eccDNA_DTX1, which promotes the growth of HQ-TK6 cells by regulating the transcription of DTX1 and ferroptosis.

Methods

Cell culture and chemical treatment

The HQ-induced malignantly transformed TK6 cells (HQ-TK6) were derived from TK6 cells treated with 10 µM HQ for 20 weeks and termed HQ-TK6 cells. TK6 cells treated with PBS were the control cells and termed PBS-TK6 cells. The procedure for establishing HQ-TK6 cells was described in our previous report [32]. The TK6 cell line was kindly provided by Prof. Lishi Zhang from Sichuan University. All of the cells were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium (Gibco, Grand Island, NY, USA) supplemented with 10% horse serum and kept in an incubator with 100% humidity and 5% CO₂ at 37 °C. Ferrostatin-1 (Fer-1, 1 µM) was used to treat the cells for 120 h to explore the impact of ferroptosis on cell proliferation.

DNA extraction, eccDNA purification, and sequencing

The procedure is illustrated in Fig. 1. Cells were collected and lysed to obtain purified high-molecular-weight (HMW) DNA using a MagAttract HMW DNA Kit (Qiagen, 67563), and DNA quality was determined with a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies Inc., Wilmington, DE, USA). To eliminate linear DNA and enrich the eccDNA, 10 µg of DNA was subjected to digestion with Plasmid-Safe ATP-Dependent DNase (20 units) (Lucigen, E3110K) in a 100 µL reaction mixture at 37 °C for 24 h. The efficacy and quality of eccDNA enrichment were determined by qPCR using primers targeting the GAPDH gene. To confirm whether the linear DNA was removed cleanly according to qPCR, we investigated whether the target band could be amplified from HMW DNA. The target band can be amplified from HMW DNA before digestion, but after linear DNA digestion is complete, the band cannot be amplified, indicating that the linear DNA digestion is complete and that the remainder of the DNA is eccDNA.

Fig. 1.

Workflow for circle-seq including purification of eccDNA via sponge column, ATP-dependent exonuclease digestion, rolling amplification, paired-end sequencing, mapping, and detection of eccDNA from paired-end reads

The eccDNA-enriched DNA samples were amplified using phi29 polymerase with random hexamer oligos according to the provider’s instructions. The phi29-amplified DNA products were subsequently sheared with a focused ultrasonicator (Bioruptor, Diagenode) to generate fragments of 300–500 base pairs, followed by purification using a MinElute PCR Purification Kit (Qiagen, 28004).

The fragmented DNA samples were subjected to sequencing library construction (VAHTS Universal DNA Library Prep Kit for Illumina V3, VAHTS, ND607). The library was purified by purification beads, and the size distribution of the fragments was analyzed on an Agilent 2100 Bioanalyzer (Agilent). Next-generation sequencing was performed on an Illumina NovaSeq 6000 platform using the constructed libraries. All of the libraries were sequenced as 2 × 150-bp paired-end reads.

Sequencing data analysis and identification of eccDNA by Circle-MAP

FastQC software was used to evaluate the quality of the original data. Low-quality bases (Q < 10) and adapter sequences were trimmed off by Cutadapt [33], and Burrows–Wheeler Aligner (BWA-MEM) software [34] was subsequently used to compare and map the original data to the reference genome (hg19). SAMtools was applied to process the SAM file to fit the format required by Circle-MAP (https://github.com/iprada/Circle-Map) [35]. The identified eccDNAs were subjected to annotation.

DNA-fluorescence in situ hybridization (DNA-FISH)

Cells in metaphase were prepared by treatment with 0.1 µg/ml colchicine for 12 h. The cells were collected, fixed with 4% paraformaldehyde solution, permeabilized with frozen methanol and soaked in 70% formamide/2× SSC at 70–75°C for 2–3 minutes. The samples were dehydrated with ethanol three times for three minutes each. For hybridization, 2 µl of a fluorescently labeled DNA probe (eccDNA_DTX1, 5’-cy3-GCTTTTGATAGTTAAGCCCTTAATAGGAAGTTAGAGTGGTGACA-3’) was added to the sample, which was subsequently placed in a wet box at 42 ℃ away from light overnight. Elution: The samples were washed in 50% formamide/2× SSC at 42–50 ℃ 3 times for 5 min. DAPI was added, and the cells were subsequently observed under a confocal microscope. The DNA probe was designed and evaluated by Oligo 7 using the junction seqeunce of eccDNA_DTX1, and synthesized by Umine Biotechnology Co., Ltd (Guangzhou, China). The procedure was performed according to previous methods, with some modifications [17].

RNA-seq

The cells were collected, and the total RNA was extracted from the cell samples with TRIzol (Invitrogen, CA, USA) according to the manufacturer’s protocol. After assessing the quality and quantity of the RNA, first-strand cDNA was synthesized, followed by immediate second-strand cDNA synthesis. The purification of double-stranded cDNA was performed using clean DNA beads (VAHTS DNA Clean Beads, VAHTS, N411). Sequencing libraries were established using a VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina (VAHTS, NR604) for Illumina^® Nova Seq (NEB, Ipswich, MA, USA). All of the libraries were sequenced as 2 × 150-bp paired-end reads.

The differential gene expression analysis was carried out with the DESeq2 R package (1.16.1). Normalized reads/fragments per kilobase of exon model per million and log₂FC were used to present the expression levels of the genes. The molecular functions, cellular components and biological processes associated with the differentially expressed genes were analyzed via Gene Ontology (GO) functional analysis. To determine the pathways enriched in the significantly differentially expressed genes, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed (http://www.genome.jp/kegg/).

CRISPR/Cas9-mediated decircularization of eccDNA

CRISPR/Cas9 was applied to inhibit eccDNA_DTX1 in HQ-TK6 cells according to our previous study [12]. We first transduced the virus package (GENE Corporation, Shanghai, China) with the CAS9 vector and obtained stable cell lines expressing CAS9 by puromycin screening. Then, we transduced lentiviral particles of the GV708 vector packaged with small guide RNA (sgRNA) targeting the junction of eccDNA_DTX1. The sgRNA sequence used was 5’ACCACTCTAACTTCCTATTA3’. The knockout efficacy was validated by FISH.

Genomic coverage analysis to determine eccDNA origins

The normalized genomic coverage was determined to evaluate the primary origin of the eccDNAs. The normalized genomic coverage was calculated as follows: percentage of eccDNA mapped to that class of element/percentage of the genome covered by that class of element according to the methods of a previous study [6].

Enrichment analysis of genes homologous to eccDNA

The functional classification and identification of important pathways associated with the differentially expressed eccDNA homologous genes were carried out using Gene Ontology (www.geneontology.org) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) databases (www.genome.jp/kegg). GO analysis focused on biological process (BP), cellular component (CC) and molecular function (MF), and KEGG analysis provided information on the molecular interaction networks of the eccDNA homologous genes.

Cell proliferation and tumorigenicity

Cell growth was evaluated both in vitro and in vivo by a CCK-8 kit and xenograft assay, respectively, as described in our previous study [27]. For the xenograft assay, 4-week-old BALB/c nude female mice were purchased from Southern Medical University. Twelve mice were randomly arranged into two groups (6 mice in each group). A total of 2 × 10⁶ cells transfected with the eccDNA_DTX1 knockdown vector or negative control vector were injected subcutaneously into the necks of the mice. The length (L) and width (W) of the tumors were measured every four days. The mice were humanely euthanized by phenobarbital sodium (CAS. 57-30-7) four weeks later, the tumor tissues were removed, weighed, and photographed. Tumor volumes were determined by the Formula V (mm³) = 0.5×LW². The animal study was approved by the Animal Welfare and Ethics Committee of Guangdong Medical University.

Western blotting

The protocol of Western blotting has been described in our previous study [27]. Antibodies of CD98 (#47213, 1:2000), FTH1 (#4393, 1:2000), GPX4 (#52455T, 1:2000), NCOA4 (#66849T, 1:2000), xCT/SLC7A11(#12691, 1:3000), GAPDH (#2118, 1:10,000), and α-tubulin (#3873S, 1:10,000) were purchased from Cell Signaling Technology (Massachusetts, USA).

Statistical analysis

The data are expressed as the mean ± standard deviation (SD) of three independent experiments. A two-sided Student’s t test was used to analyze the differences between the samples. Matched samples were tested with a paired, double-sided t test and rank correlation. The statistical analysis was performed using SPSS software (version 25.0). For all of the statistical comparisons, P < 0.05 was considered to indicate statistical significance.

Results

Identification of eccDNAs by circle-seq in TK6-HQ and TK6-PBS cells

EccDNAs were detected in TK6-HQ and TK6-PBS cells by electron microscopy (Fig. 2A). Then, the circle-MAP [17] approach was applied to obtain information about the abundance of eccDNAs in the TK6-HQ and TK6-PBS cells. The eccDNA amplified by phi29 was sequenced as 2 × 150-bp paired-end reads on an Illumina NovaSeq platform. We obtained 200–300 million reads from each sample. Virtually all of the reads were mapped to the hg19 genome, suggesting that the eccDNAs were derived from the genome. Circle-seq identified a total of 629,179 eccDNAs, accounting for 7.3% (TK6-PBS) and 3.4% (TK6-HQ) of the genome; moreover, the normalized eccDNA counts in TK6-HQ cells were greater than those in TK6-PBS cells (Fig. 2B), and the normalized eccDNA counts and chromosomal distribution are shown in Fig. 2C. There were 12,987 eccDNAs with different counts and 901 eccDNAs with significantly different counts; 235 were downregulated, and 666 were upregulated. These results were visualized via a volcano plot (Fig. 2D). Chromosomal distribution analysis of the eccDNAs containing coding genes (Fig. 2E), and cluster analysis were visualized via a heatmap (Fig. 2F). For the volcano plot, significantly different eccDNAs were filtered by FDR (P < 0.05) and |log₂FC| > 2. For the heatmap, the FDR (P < 0.05), |log2FC| > 2, and length of eccDNA > 2000 bp were used. The top 20 eccDNAs based on the fold change from the heatmap are listed in Table 1, and 12 genes were included in these 20 eccDNAs. The full list is shown in Supplementary Table 1.

Fig. 2.

Identification of eccDNAs by circle-seq in TK6-HQ cells. (A) EccDNA was observed in both TK6-HQ and TK6-PBS cells by electron microscopy. The red square was enlarged 4-fold from the dark square to zoom in and view the eccDNA; scale bar = 2.0 μm. (B) The normalized counts of eccDNAs by chromosome length (million) were lower in TK6-HQ cells than in TK6-PBS cells. (C) The hotspot distribution of eccDNAs on chromosomes. D-F. Volcano plot (D, size > 2000 bp), chromosomal distribution analysis of the eccDNAs containing coding genes (E), and cluster analysis via a heatmap (F) based on the 902 eccDNAs with different counts. For E, the outer layer represents the chromosomes and the coding gene names, and the inner layer represents the counts of the different eccDNAs: blue, downregulated; red, upregulated

Table 1.

Top 20 different eccDNAs

Chromosome	Length (bp)	log2FC	P value	Significant	Gene	Functions of related gene
chrX	475	8.28	3.06E-7	upregulated	BGN	Inflammation and innate immunity
chr10	411	8.15	1.81E-7	upregulated	LINC01264	NA
chr1	312	7.31	6.14E-5	upregulated	PRDM16	pathogenesis of myelodysplastic syndrome and acute myeloid leukemia
chr13	354	7.08	1.71E-6	upregulated	MCF2L	Rho/Rac signaling pathways, osteoarthritis
chr2	1732	7.07	3.86E-5	upregulated	NA	-
chr17	1579	7.00	2.80E-5	upregulated	LINC00673	Cell proliferation
chr7	756	6.84	0.00011	upregulated	ADCY1	Brain development
chr7	581	6.81	9.99E-5	upregulated	SDK1	Gene regulation
chr12	170	6.60	0.0002	upregulated	VWF	Von Willebrand disease
chr11	665	6.58	1.61E-5	upregulated	NA
chr12	1373	6.54	1.43E-4	upregulated	TMEM132C	Adhesion
chr13	1742	6.53	1.99E-06	upregulated	COL4A2	Angiogenesis and tumor growth
chr3	3787	6.46	6.19E-05	upregulated	LINC02016	NA
chr16	752	6.38	0.00017	upregulated	NA	-
chr12	1263	6.35	0.0001	upregulated	NA	-
chr4	3706	6.31	0.00018	upregulated	NA	-
chr8	1797	6.30	4.41E-05	upregulated	NA	-
chr22	764	6.19	0.00022	upregulated	NA	-
chr17	609	6.17	9.32E-06	upregulated	RPTOR	Cell growth
chr8	1163	6.08	0.00018	upregulated	PXDNL	Hydrogen peroxide catabolic process

NA, not available

Characteristics of the eccDNA from TK6-HQ and TK6-PBS cells

The eccDNAs that were significantly different between the TK6-HQ and TK6-PBS cells covered 1.05% of the genome, indicating that a considerable fraction of the TK6 genome may be related to the process of carcinogenesis. The average sizes of the eccDNAs in TK6-HQ and TK6-PBS cells were 521 bp and 512 bp, respectively, which are similar. Most of the different eccDNAs (12194/12897) were less than 5 kb, although the longest eccDNA was almost 45 kb in length. Interestingly, four periodic peaks with an interval of ~ 180 bp were observed, and the length of the most abundant peak was approximately 366 bp (Fig. 3A). The normalized genomic coverage indicated that eccDNAs were enriched in repeat areas, introns, and CpG islands, with relatively low enrichment in exons and 5’UTRs (Fig. 3B). No significant difference in genomic enrichment was found between TK6-HQ and TK6-PBS cells. The identified eccDNAs based on genomic origin are shown in pie charts (Fig. 3C). The proportion of eccDNAs in different genomic regions is illustrated in Fig. 3D, showing no differences between TK6-HQ and TK6-PBS cells. Interestingly, the eccDNA per million (EPM) was positively associated with gene abundance (Fig. 3E), which was consistent with the findings of previous reports implying that eccDNAs are enriched in gene-rich chromosomes, such as chr17 [9].

Fig. 3.

Characteristics of eccDNAs from TK6-HQ and TK6-PBS cells. (A) The length distribution of eccDNAs from TK6-HQ and TK6-PBS cells. (B) The distribution of eccDNAs with different counts among different kinds of genomic elements. (C) The eccDNA fraction based on its genomic origin. TSS, transcription start site; UTR, untranslated region. (D) The normalized eccDNA count in 23 pairs of chromosomes. (E) The eccDNA abundance was associated with the frequency of genes, right for TK6-HQ cells, middle for TK6-PBS cells, and left for the average of TK6-HQ and TK6-PBS

Biological functions of the eccDNAs according to GO and KEGG analyses

To provide a comprehensive understanding of eccDNAs, the homologous genes of the corresponding eccDNAs were annotated and further evaluated by GO and KEGG analyses. GO analysis revealed that the homologous genes of eccDNAs with different counts were enriched in 33 BP terms (adjusted P < 0.05). Nine of these terms were related to eccDNA biogenesis and cancer development (Fig. 4A), developmental cell growth, blood vessel endothelial cell proliferation involved in sprouting angiogenesis, regulation of blood vessel endothelial cell proliferation involved in sprouting angiogenesis, regulation of reactive oxygen species metabolic process, reactive oxygen species metabolic process, regulation of superoxide metabolic process, regulation of superoxide anion generation, and positive regulation of reactive oxygen species metabolic process. There were were no significantly enriched MF or CC terms. Additionally, KEGG pathway enrichment analysis revealed that 1 pathway (mTOR signaling pathway), 1 function (growth hormone synthesis, secretion and action), and 4 diseases (hepatitis B, non-small cell lung cancer, glioma, chronic myeloid leukemia) were enriched (Fig. 4B). These findings implied that some of the homologous genes of eccDNAs with different counts may be involved in the induction of carcinogenesis by HQ.

Fig. 4.

KEGG (A) and GO (B) analyses based on the genes contained in eccDNAs with different counts; right section, the eccDNAs with upregulated counts; and left section, the downregulated counts. The top 10 significantly enriched functions or pathways are listed

Effect of DNA circularization on gene expression

Differential expression of mRNAs determined by RNA-seq

DNA circularization can drive gene expression in a special manner, including the structure, number, and function of diversity of eccDNA [11]. To determine the effects of DNA circularization on gene expression, we measured the expression of mRNAs via RNA-seq. mRNAs with a fold change > 2 and a P value < 0.05 were considered to be significantly expressed. Overall, 1491 mRNAs with significant differences in expression were identified between HQ-TK6 cells and PBS-TK6 cells (Fig. 5A). Of these mRNAs, 1064 were upregulated, and 427 were downregulated. A heatmap showed the mRNA expression profiles between HQ-TK6 and PBS-TK6 (Fig. 5B). The top 20 significantly differentially expressed mRNAs according to fold change are shown in Table 2, and the full list of significantly differentially expressed genes is included in Supplementary Table 2. These findings indicated that during the carcinogenesis induced by HQ, considerable alterations in mRNA expression occurred. There might be some associations with the eccDNA regulatory pathway.

Fig. 5.

RNA-seq. A total of 1491 DEGs according to the criteria of a p value < 0.05 and a fold change ≥ 2 were visualized in a volcano plot (A) and heatmap (B). KEGG (C) and GO (D) analyses of the differentially expressed genes; right panel shows the eccDNAs with upregulated expression, and left panel shows the downregulated expression. The top 10 significantly enriched functions or pathways are listed

Table 2.

Top 20 different expressed genes

Gene	log2FC	P value	Significant	Function
BANK1	8.92	3.53E-48	upregulated	B-cell receptor-induced calcium mobilization
FUT1	7.24	2.38E-07	upregulated	Cell growth, migration, cancer stemness
CPED1	6.21	2.69E-46	upregulated	Bone mineral density
GDAP1L1	6.00	3.62E-07	upregulated	Cell differentiation
LAMA5	5.26	1.45E-41	upregulated	Cell differentiation, growth, migration
SIPA1L2	5.25	5.52E-25	upregulated	Lipid regulation
DENND1B	5.25	3.89E-21	upregulated	Gene regulation, Asthma
INHBE	5.07	1.50E-05	upregulated	Cervical Cancer, rgulating the proliferation of pancreatic exocrine cells
CSRNP3	5.06	8.47E-12	upregulated	Gene regulation
HOMER2P1	5.06	1.99E-29	upregulated	NA
SLC6A12	5.03	1.58E-09	upregulated	Neurological disease, membrane transportation
VIL1	4.97	2.61E-38	upregulated	Microvillus inclusion disease, candidate molecular marker for Hepatocellular carcinoma
ACSBG1	4.96	4.66E-46	upregulated	Fatty acid metabolism
FGF13	4.91	6.19E-14	upregulated	Mmbryonic development, Cell growth, Morphogenesis, Tissue repair, Tumor growth
RP11-448G15.1	4.89	4.01E-10	upregulated	Retinitis pigmentosa
CTC-573N18.1	4.89	4.09E-06	upregulated	NA
RP11-284M14.1	4.69	6.98E-09	upregulated	Retinitis pigmentosa
CHAC1	4.62	2.02E-05	upregulated	Cell differentiation, apoptosis, ovarian and breast cancer
LL22NC03-75H12.2	4.51	6.58E-17	upregulated	NA
MYO7B	4.51	9.04E-26	upregulated	Brush border microvilli function

NA, not available

GO and KEGG analyses of differentially expressed genes

To determine the functions of the differentially expressed genes, we conducted GO and KEGG enrichment analyses. GO analysis revealed 11 MF terms, 191 BP items were enriched in the upregulated genes and downregulated genes (adjusted P < 0.05) (Fig. 5C). Notably, five pathways, the intrinsic apoptotic signaling pathway in response to endoplasmic reticulum stress, the intrinsic apoptotic signaling pathway, the response to topologically incorrect proteins, the regulation of Ras protein signal transduction, the cellular response to topologically incorrect proteins, and myeloid leukocyte cytokine production, were associated with the generation of eccDNA and carcinogenesis. Moreover, for the KEGG pathway analysis, 3 and 13 items were associated with the upregulated and downregulated differentially expressed genes, respectively, in addition to 3 cancer-related pathways (Fig. 5D). These results implied that a variety of genes were significantly differentially expressed during HQ-induced carcinogenesis.

DTX1-containing eccDNA (eccDNA_DTX1) promotes cell growth by promoting DTX1 expression

In human tumor cells, eccDNAs drive high level of oncogene products via several mechanisms [36–38], including high copy numbers of eccDNA. However, until now, no studies have provided evidence to support this concept in the process of HQ-induced transformation. We subsequently hypothesized that the eccDNA-driven gene expression changes could be found in HQ-induced transformation. By conducting a joint analysis of different eccDNA counts (by the standard of P < 0.05 and fold change ≥ 2) and differentially expressed transcripts (by the standard of P < 0.05 and fold change ≥ 2), we constructed a four-quadrant graph that included 21 pairs of eccDNAs and mRNAs. However, only 8 pairs (DTX1, IL21R, TBXAS1, JPH2, SHANK2, SVIL, NEDD9, and PTPRS) followed the pattern of eccDNA-driven transcription according to their expression and functions (Fig. 6A). The detailed information is listed in Table 3 and Supplementary Table 3. These genes were enriched in 4 CC and 5 BP items according to the GO analysis (Fig. 6B).

Fig. 6.

Knockdown of eccDNA_DTX1 promoted cell growth both in vitro and in vivo. (A) Integrated analysis based on the different counts of eccDNAs and their related genes revealed 8 potential genes affected by eccDNA. (B) GO analysis revealed the top 10 pathways associated with the 19 genes from A. (C) DNA-FISH showed that eccDNA_DTX1 was inhibited by CRISPR/Cas9, eccDNA_DTX1 was probe by Cy3, scale bar, 5 μm. eccDNA knockdown inhibited cell growth in vitro assessed by CCK-8 (D), in vivo assessed by xenograft assay in BALB/c nude female mice (E-F) and morphology (G, 40×, the black square is digitally enlarged at 4×), and inhibited DTX1 transcription (H-I) and promoted ferroptosis (J-K). L. Fer-1 abrogated the cell proliferation inhibition induced by eccDNA_DTX1 knockdown. *, P < 0.05 vs. ecc_NC or ecc_DTX1 + DMSO

Table 3.

The integrated analysis of results from circle-seq and RNA-seq

Genes	Chr.	eccDNA_Start	eccDNA_End	eccDNA			RNA		Function
				P value	log2FC		P value	log2FC
CMTM4	chr16	66,727,887	66,732,308	0.0320	2.95	0.0095		-1.36	Inhibit cell proliferation
COL6A3	chr2	238,246,652	238,249,858	0.0303	-3.12	0.0063		1.30	Mutation, cell proliferation, cycle apoptosis
CROCC	chr1	17,281,268	17,282,826	0.0158	-4.32	0.0001		2.31	EMT and cell cycle
DTX1	chr12	113,528,106	113,532,168	0.0200	3.53	0.0143		1.28	Ubiquitin Proteasome Pathway
IL21R	chr16	27,452,674	27,453,787	0.0018	5.38	0.0036		1.05	Tumorigenesis
JPH2	chr20	42,816,123	42,818,125	0.0483	2.74	0.0024		2.54	Signal transduction
KLC4	chr6	43,042,296	43,043,686	0.0436	-3.07	0.0020		1.03	Cell cycle and DNA damage repair
MARCO	chr2	119,732,960	119,733,744	0.0045	4.06	0.0001		-2.56	Autophagy
MPP2	chr17	41,957,402	41,959,719	0.0110	-4.05	0.0222		2.35	Cell proliferation and signal transduction
NEDD9	chr6	11,188,673	11,191,273	0.0419	-2.87	0.0084		-1.95	DNA methylation
PTPRN2	chr7	157,947,328	157,948,064	0.0385	3.11	0.0084		-1.76	Tumor related pathway
PTPRS	chr19	5,310,200	5,312,767	0.0074	-2.93	0.0314		-1.61	Cell proliferation, differentiation and malignant transformation
SARDH	chr9	136,565,495	136,566,671	0.0476	-3.01	0.0084		3.00	Catalytic oxidative demethylation of sarcosine
SERINC2	chr1	31,902,143	31,903,136	0.0373	-3.30	0.0061		2.81	Autophagy
SHANK2	chr11	70,808,480	70,809,453	0.0166	3.47	0.0000		2.81	Scaffold protein
SLCO5A1	chr8	70,746,089	70,747,665	0.0354	-3.04	0.0011		1.09	Organic anion transport
STARD13	chr13	33,872,954	33,874,333	0.0355	-3.38	0.0001		1.76	Scaffold protein, cell proliferation, cell migration
SVIL	chr10	29,844,089	29,845,668	0.0192	3.75	0.0009		3.43	Cell migration
TBXAS1	chr7	139,697,765	139,700,210	0.0196	3.37	0.0005		2.33	Cell homeostasis, cell metabolism

DTX1 is a member of the deltex protein family and functions as a ubiquitin ligase. The DTX1 protein is involved in cell signal transduction, growth, differentiation and apoptosis and is related to the occurrence and development of a variety of tumors. We applied the CRISPR/Cas9 technique to target the junction sequence of eccDNA_DTX1. DNA-FISH confirmed that the abundance of eccDNA_DTX1 decreased in the targeting sequence group (eccDNA_DTX1_KD) compared with the scramble sequence group (eccDNA_NC) (Fig. 6C). Cell proliferation, a hallmark of cancer, was inhibited by the decircularization of eccDNA_DTX1 (Fig. 6D). Moreover, the same results were also found in the tumorigenicity assay: decircularization of eccDNA_DTX1 inhibited the growth of xenograft tumors in nude mice (Fig. 6E-G). To explore the mechanism by which eccDNA_DTX1 participates in cell growth, the expression of DTX1 was measured via RT‒PCR, and the results showed that both DTX1 mRNA and protein expression was inhibited (Fig. 6H‒I). Moreover, eccDNA_DTX1 knockdown induced ferroptosis (Fig. 6J-K). Importantly, ferroptosis inhibitor, Fer-1, abrogated the cell proliferation inhibition induced by eccDNA_DTX1 knockdown (Fig. 6L). These findings indicated that eccDNA_DTX1 facilitates tumorigenesis by increasing DTX1 expression and promoting ferroptosis.

Discussion

This study revealed the eccDNA profile of TK6-HQ cells by circle-seq, which is a high-throughput sequencing method that is used after DNA exonuclease digestion and rolling circle amplification. We found 629,179 eccDNAs in total and 901 eccDNAs with different counts, most of which mapped to the repeat sequence area and intronic region of the genome. For individual chromosomes, our findings are similar to previous findings in which the eccDNA regions were related to the noncoding gene area [39, 40], while some studies have shown that eccDNA-rich regions are located within gene-rich regions [9]. Genomic annotation of the overall population of eccDNA molecules revealed that these molecules are preferentially generated from 5′-UTRs, exonic regions, and CpG island regions [6]. However, we also found that the counts of eccDNA normalized by the length of the chromosomal location were positively correlated with the number of genes on the corresponding chromosome. Because the noncoding region accounts for most of the genome and seems to be a primary source of eccDNAs, it might not be unreasonable to suggest that the eccDNA regions are related to the noncoding gene region.

To explore the mechanism of eccDNA formation, motif analysis of the eccDNA junction sequence was conducted, and we found that repeat sequences were observed across the junction area. Several studies have shown that some eccDNA sequences are flanked at both ends by, on average, 9 ~ 11 bp of direct repeats [21, 41–43]. In humans and pigeons, most eccDNA molecules, which contain or are adjacent to short direct repeats, are derived from repetitive elements [7]. The fact that self-homology and DNA motifs are near circle breakpoints supported a model for circle generation through either a homology or microhomology-mediated process, and the circles exhibit homology near their breakpoint [44]. These findings indicated that the repeat sequence contributes to the biogenesis of eccDNAs via DNA repair pathways, such as homologous recombination and microhomology-mediated end joining. The underlying mechanism of eccDNA functions has not been elucidated, although several prominent studies have proposed some models of eccDNA formation, including the breakage-fusion-bridge (BFB) cycle, chromothripsis, episome model, and translocation-deletion-amplification model [3, 6, 14, 15]. DNA breakage, DNA recombination, and DNA rearrangement [14, 16–18], which can be induced by short-term or long-term exposure to HQ [45–47], play critical roles in these eccDNA biogenesis models. The mechanism of biogenesis varies under different circumstances, which contributes to the characteristics of eccDNAs, shedding new light on the application of eccDNA in disease diagnosis, monitoring, prognosis and treatment.

Our findings are in agreement with previous studies demonstrating that eccDNAs can be of various lengths ranging from dozens of bases to hundreds of thousands of bases, and the majority of the eccDNAs are smaller than 1,000 bp [8, 9, 48]. Most related research has shown that the distribution of eccDNA length has several unique peaks and that the sizes of the peaks are significantly different between lung cancer tissues and adjacent normal tissues [49, 50] and between maternal and fetal plasma [6]; however, no apparent differences in length were found in our study, which is consistent with the findings of previous studies linking eccDNAs and esophageal squamous cell carcinoma [39]. Recent studies have demonstrated that these eccDNA molecules exhibit bimodal size distributions peaking at ~ 202 and ~ 338 bp, with distinct 10-bp periodicity observed throughout the size ranges within both peaks, suggesting their nucleosomal origin [6]. Together with the motif analysis and distribution of eccDNA length, we propose that the eccDNAs in HQ-TK6 cells and PBS-TK6 cells are derived from genomic DNA, which is consistent with the findings of all of the previous studies.

We found 901 eccDNAs with different counts between HQ-TK6 and PBS-TK6 cells by circle-seq, which is a sensitive and high-throughput method that has been verified by several studies. Currently, investigating the role and mechanism of action of eccDNA is the most challenging work in the eccDNA field. By combining circle-seq and RNA-seq, we screened eccDNA_DTX1, which is involved in cell proliferation and other functions. Destruction of the circular structure of eccDNA_DTX1 by CRISPR/Cas9, which targets the junction sequence, inhibited cell growth in the short-term and long-term both in vitro and in vivo. Notably, the repressive effect of one eccDNA on cell growth was not apparent. One recent study revealed that multiple eccDNAs function cooperatively as eccDNA hubs [51], which implied that eccDNA may be an individual genome and that one eccDNA has limited functions. A study by Zou et al. first revealed that microRNA-17-92-containing eccDNAs promote hepatocellular carcinoma by promoting microRNA-17-92 expression [4, 52], while eccDNA derived from PLCG2 contributes to the metastasis of non-small cell lung cancer by enhancing the transcription of PLCG2 [53]. To the best of our knowledge, our study is the first to profile the characteristics and functions of eccDNA in a malignant cell model induced by an environmental carcinogen. EccDNA has been found to be involved in aging [54–56], drug resistance [57, 58], gene regulation [59], evolution [5, 18], and environmental adaptation [17, 59–62], which involve different mechanisms, including those that modulate gene amplification [63]gene expression [64]. However, we still pay more attention to the role and mechanism of eccDNAs because several recent prominent studies have revealed that eccDNAs exert critical effects through different mechanisms [19, 65, 66]. The role and mechanism of eccDNAs remain elusive, but emerging bioinformatic tools (e.g., the ECCsplorer [67], ecc_finder [37], CircleBase [68], and eccDNAdb [69]) and methods [70] for eccDNA research will broaden the field and contribute more to outstanding works revealing the role and mechanism of eccDNA. We will focus on the mechanism of eccDNAs to determine their utilization in disease prevention, especially in the prevention of environmental carcinogen-induced carcinogenesis.

Conclusions

EccDNA_DTX1 promotes cell growth in hydroquinone-induced malignantly transformed cells by regulating the transcription of DTX1 and ferroptosis. This study profiles eccDNA characteristics and defines the role and mechanism of eccDNA_DTX1 for the first time, shedding new light on the relationship between eccDNAs and carcinogenesis.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

BFB

breakage-fusion bridge

BP

biological process

CC

cellular component

DTX1

deltex E3 ubiquitin ligase 1

eccDNA

extrachromosomal circular DNA

EPM

eccDNA per million

FISH

fluorescence in situ hybridization

GO

Gene Ontology

HQ

hydroquinone

HQ-TK6

hydroquinone-induced malignantly transformed TK6 cells

KEGG

Kyoto Encyclopedia of Genes and Genomes

MF

molecular function

ROS

reactive oxygen species

UTR

untranslated region

Author contributions

Xiaoxuan Ling designed, conceptualized, and wrote the article. Qunfang Jiao, Daifan Lin, Jialong Chen, Yali Han, Jinxue Meng, Bohuan Zhong, and He Zhang performed the experiments and reviewed the manuscript. Gongda Zhang, Fanglin Zhu, Jiheng Qin, and Yongdui Ruan validated the data, provided the images, and reviewed the manuscript. Linhua Liu designed, conceptualized, wrote, and revised the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (81202231, 82204134), the Natural Science Foundation of Guangdong Province (2022A1515140180, 2020A1515110614, 2023A1515140163), the Discipline Construction Project of Guangdong Medical University (4SG22260G, 4SG21021G).

Data availability

The datasets generated and/or analysed during the current study are available in the NCBI Sequence Read Archive under the accession number PRJNA1082506 (circle-seq) and PRJNA1082397 (RNA-seq).

Declarations

Ethics approval and consent to participate

Humans were not included in this study. The animal study was approved by the Animal Welfare and Ethics Committee of Guangdong Medical University (GDY2204004). The animal experiments were carried out according to the 3Rs principle.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.