circIARS: a potential plasma biomarker for diagnosing non-small cell lung cancer

Abstract

Non-small cell lung cancer (NSCLC) is one of the most prevalent cancers in the world, and early diagnosis can effectively improve patient survival. Here, differentially expressed circIARS genes are screened from the sequencing results, and their molecular characteristics are examined by Sanger sequencing, RNase R assay, agarose gel electrophoresis (AGE), and fluorescence in situ hybridization (FISH). Real-time fluorescence quantitative polymerase chain reaction (qRT-PCR) is performed to detect the expression level of circIARS. The diagnostic value of the signature is analyzed using a subject operating characteristic (ROC) curve. Moreover, plasma is collected from postsurgical, chemotherapy, and relapse patients to investigate the prognostic value of circIARS in NSCLC. The expression of circIARS is greater in both the plasma and tissues of NSCLC patients than in those of healthy individuals, and could be used to distinguish NSCLC patients from patients with benign pulmonary disease (BPD), small cell lung cancer (SCLC) patients, and healthy individuals. The expression level of circIARS relatively decreases after antitumor therapy, such as chemotherapy, and relatively increases after recurrence. ROC analysis reveals that circIARS has better detection efficiency than traditional markers. In addition, circIARS expression level is strongly correlated with several clinicopathological parameters. Finally, we tentatively predict the downstream miRNAs or RBP that might bind to circIARS. Plasma circIARS is significantly greater in NSCLC patients and has good stability and specificity as a diagnostic marker, which could aid in the adjuvant diagnosis and dynamic monitoring of NSCLC.

Background

Lung cancer is the leading cause of cancer death in men and women. The burden of treatment for non-small cell lung cancer (NSCLC), the most common type of lung cancer, has become a growing concern [1]. Despite the rapidly evolving prospects for NSCLC treatment, the disease is often diagnosed by local progression or metastasis or is already advanced. In contrast, early detection and treatment of NSCLC can significantly improve survival rates [1]. Among the ancillary diagnostic methods for detecting NSCLC, chest X-ray and sputum cytology often fail to detect NSCLC early; although bronchoscopy is excellent at detecting lung tumors, it is invasive. In addition, although CT is noninvasive and highly sensitive, it is often accompanied by overdiagnosis [ 2, 3]. Several studies have confirmed the effects of noncoding RNAs (ncRNAs) and mRNAs on the early screening and biological progression of NSCLC, such as the inhibition of growth, ING5 overexpression, miR-34c-5p/Snail1 inhibition of EMT and lung cancer cell invasion [4], and on the pseudogene DUXAP10 which contributes to gefitinib resistance in NSCLC by inhibiting OAS2 expression [5]. Therefore, the screening of simple and efficient diagnostic markers and the development of minimally invasive techniques for the diagnosis of early-stage NSCLC are of great clinical importance.

Circular RNA (circRNA) is a covalently closed loop that lacks the 5′ cap and 3′ tail and is usually formed by reverse shearing of the precursor mRNA (premRNA) [6]. circRNAs are evolutionarily conserved, highly abundant in eukaryotes, and tissue-, cell type-, or developmental stage specific. It is also more stable than other linear RNAs due to the lack of exposed ends that are easily degraded, which allows circRNAs to be detected noninvasively in body fluids [7]. This finding illustrates the natural advantages of circRNAs over other noncoding RNAs (ncRNAs) that act as molecular markers.

Advances in high-throughput sequencing and related bioinformatics algorithms have facilitated not only the discovery of more types of circRNAs in multiple model organisms but also the emergence of practical applications of circRNAs, including their utilization as molecular markers for a variety of diseases [8]. In addition, circRNAs have been found to regulate tumor cell proliferation, migration, and apoptosis through multiple mechanisms [ 9, 10]. For instance, circHIPK3 increases WEE1 expression via miR-124 in glioma and promotes cell proliferation and invasion [11]. cIARS physically interacts with ALKBH5 and positively regulates sorafenib (SF)-induced iron death by inhibiting the ALKBH5-mediated inhibition of autophagy [12]. As discussed above, circRNAs have great potential as new biomarkers and for promoting the biological progression of tumors.

This study further explored the role of circRNAs in the adjuvant diagnosis and dynamic monitoring of NSCLC. We screened circIARS, a molecule specifically expressed in NSCLC tissues, for the first time by high-throughput sequencing. We then analyzed its correlation with clinicopathological parameters and its diagnostic value by ROC curve analysis. Finally, we concluded that circIARS has potential as a novel marker for the diagnosis of NSCLC.

Materials and Methods

Plasma and tissue sample collection

For this study, plasma samples were obtained from the clinical laboratory of Nantong University Hospital, all patients were diagnosed by clinicopathology, and those with previous tumor histories were excluded. Fresh peripheral blood was collected and mixed using ethylenediaminetetraacetic acid (EDTA) anticoagulation tubes. Peripheral blood was centrifuged at 1500 g for 15 min at 20°C to obtain plasma, which was subsequently aliquoted and frozen at ‒80°C. The dataset contains plasma from 100 patients with NSCLC, 50 patients with benign pulmonary disease (BP), 36 patients with SCLC, and 100 healthy donors. Sixteen pairs of cancerous tissues and adjacent nontumor tissues were also collected. The distance between the paracarcinoma tissue and the tumor tissue of NSCLC patients was 3 cm, and H&E staining confirmed that there was no tumor infiltration in the paracarcinoma tissue. All tissues were placed in liquid nitrogen immediately after excision. We screened patients who were diagnosed with NSCLC at the Affiliated Hospital of Nantong University and were older than 40 years old. Healthy individuals aged 40 years or older who had no abnormalities in any of the biochemical indices were screened by medical examiners. This study was approved by the Ethics Committee of Nantong University Hospital (Ethics Review Report No. 2018-L055). Informed consent was obtained from all participants, who agreed to publication prior to the clinical trial.

High-throughput sequencing

Total RNA minikit (Magen, Guangzhou, China) was used to extract total RNA from three pairs of NSCLC tissues and adjacent nontumor tissues. Subsequently, a Qubit 3.0 fluorometer (Invitrogen, Carlsbad, USA) and an Agilent 2100 bioanalyzer (Agilent, Santa Clara, USA) were used to assess the concentration and purity of the RNA, respectively. RNA-seq libraries were then constructed using TruSeq RNA sample preparation kits (Illumina, San Diegoa, USA). The purified cDNA libraries were sequenced on an Illumina HiSeq Xten platform with a bipartite 2×150 bp read length (PE150). After obtaining the raw data, low-quality reads and adapters were first removed. Subsequently, whether these genes are characterized by circRNAs ( e. g., back splicing identification) was confirmed, and after their identification as circRNAs, quantitative analysis was performed to obtain expression profiles of the circRNAs.

RNA extraction and analysis

RNA was extracted from plasma, tissues, and cells using TRIzol reagent (Invitrogen). Before RNA extraction, the experimental equipment was autoclaved, and the surface of the equipment was wiped with an RNase scavenger to remove exogenous RNase and avoid RNA degradation. We used a NanoDrop™ One (Thermo Fisher Scientific, Waltham, USA) for the determination of RNA concentration and purity, the Agilent 2100 Bioanalyzer and agarose gel electrophoresis (AGE) for the determination of RNA integrity and genomic DNA (gDNA) contamination.

cDNA synthesis and real-time fluorescence quantitative polymerase chain reaction (qRT-PCR)

The total RNA was reverse transcribed to cDNA (Thermo Fisher Scientific). The procedure was 25°C for 5 min, 42°C for 60 min and 70°C for 5 min. circIARS expression levels in different samples were measured by qRT-PCR using an ABI QuantStudio 5 (Thermo Fisher Scientific), and the fluorescent signal was generated by a fluorescent DNA binding dye (SYBR Green). The total volume of the qRT-PCR mixture was 20 μL, which included 10 μL of ChamQ Universal SYBR qRT-PCR Master Mix (Vazyme Biotech, Nanjing, China), 5 μL of cDNA, 1 μL of primers, and 4 μL of enzyme-free water. The PCR program was set to activate the enzyme at 95°C for 10 min, denature at 95°C for 10 s, anneal at 60°C for 30 s, collect fluorescence information at 80°C, and cycle 40 times. All operations were carried out in an enzyme-free environment. The primers used were synthesized by Rui Bo Biological (Suzhou, China). The forward primer sequence of circIARS was 5′-TGAGAACACGGTTAGAGAAAGC-3′, and the reverse primer sequence was 5′-AAATTAGAGGAGTGTCTGATCTGC-3′.The forward primer sequence of IARS was h-IARS-173-F1 5′-GCCACCACCTTCACTCACA-3′, and the reverse primer sequence was h-IARS-173-R1 5′-TTTCCAAATCGTTTTTCTCGTA-3′. The forward primer sequence for 18S was 5′-GTAACCCGTTGAACCCCATT-3′, and the reverse primer sequence was 5′-CCATCCAATCGGTAGTAGCG-3′. The forward primer sequence of GAPDH was 5′-AGAAGGCTGGGGCTCATTTG-3′, and the reverse primer sequence was 5′-GCAGGAGGCATTGCTGATGAT-3′. The PCR products were further identified by Sanger sequencing.

Cell culture

Human bronchial-like epithelial cells (HBEs) and NSCLC cell lines (NCI-H1299 and A549) were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The cell lines were cultured in RPMI 1640 (Gibco, Grand Island, USA) supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin and streptomycin (HyClone, Logan, USA). Then, the cells were cultured in a humidified incubator (37°C, 5% CO ₂).

Actinomycin D and RNase R experiments

The NCI-H1299 cell line was plated in 6-well plates and cultured at 37°C and 5% CO ₂. After the addition of 2.5 μg/mL actinomycin D, 2 mL of cell suspension was added to each well and incubated for a certain time. Follow-up assays were performed to detect circIARS expression. Total RNA extracted from the cells was treated with RNase R (3‒4 U/mg; Epicentre). The relative expression of circIARS and its parent gene IARS was then examined separately by qRT-PCR.

Fluorescence in situ hybridization (FISH)

Cells were inoculated in 24-well plates containing crawlers. A FISH kit (RiboBio, Suzhou, China) was used for subsequent manipulation. Hybridization solutions containing circIARS, 18S, and U6 probes (RiboBio) were added to the different wells. A final drop of Antifade Mounting Medium supplemented with DAPI (Beyotime, Shanghai, China) was added, and the cells were examined with a fluorescence microscope (Olympus, Tokyo, Japan).

Functional enrichment analysis

The functional enrichment analysis included Gene Ontology (GO) functional enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The GO analysis included three levels: biological process (BP), molecular function (MF), and cellular component (CC). In the functional annotation and enrichment analysis, based on the principle that proteins with the same or similar sequences have similar functions, the GO term and KEGG pathway corresponding to the submitted protein sequences were obtained for annotation through the Diamond program of eggNOG-mapper software. We used Perl (v5.32.1) for annotation and the R package ggplot2 for plotting. The GO database is available at https://geneontology.org/, and the KEGG database is available at https://www.kegg.jp/.

Statistical analysis

All data in this study were analyzed using SPSS 20.0 (IBM SPSS Statistics, Chicago, USA) software and plotted using GraphPad Prism V.8.00 (GraphPad Software Inc., La Jolla, USA). The relative expression of circIARS was calculated using the 2 ^‒ΔΔCt method. Data are expressed as the mean±standard deviation (SD) of the values obtained from three independent experiments. Paired t tests were used to assess differences in the circIARS between preoperative and postoperative specimens and between NSCLC and paracarcinoma tissues. Two-sided unpaired t tests were used for comparisons of two independent samples, and one-way analysis of variance (ANOVA) was used for comparisons of multiple independent samples. Bivariate logistic regression was used to analyze the diagnostic value of circIARS, carcino-embryonic antigen (CEA), squamous cell carcinoma antigen (SCC), and Cyfra21-1. The subject work characteristic (ROC) curve and area under the ROC curve (AUC) were obtained by nonparametric analysis. The Uden index was used as the threshold value. The cardinality test was used to analyze the correlation between the expression of circIARS and clinicopathological parameters. P<0.05 indicates that the difference was statistically significant.

Results

circIARS expression profile in NSCLC tissues

To identify circRNAs differentially expressed in NSCLC, 3 pairs of cancer and paracancerous tissues were selected for high-throughput sequencing [13]. As a result, 2319 circRNA targets were discovered. Molecules with differential fold change (logFC)>2.0 and P<0.05 were identified. The sequencing results were visualized by a heatmap and a volcano map. The clustered heatmap shows the differentially expressed circRNAs. In the volcano map, red dots represent circRNAs that are differentially upregulated, while green dots represent circRNAs that are differentially downregulated [13]. From the sequencing results, we selected hsa_circ_0077656, hsa_circ_0006702, hsa_circ_0002037, hsa_circ_0046580, and hsa_circ_0007355 based on their fold changes. All these genes, except hsa_circ_0006702, were gray molecules and were either not expressed or were underexpressed in plasma. Therefore, circIARS (hsa_circ_0006702, which is derived from IARS and designated circIARS) was selected for further in-depth study. First, the overall level of circIARS was assessed using 50 pairs of NSCLC tissues collected. The findings indicated that the expression level of circIARS was greater in cancer tissues than in corresponding paracarcinoma tissues ( Figure 1A). Second, circIARS was expressed at significantly greater levels in the plasma of 50 NSCLC patients than in that of 50 healthy subjects ( Figure 1B). Furthermore, we performed time-dependent experiments. To determine whether the circIARS in the supernatant was released passively, we initially cultured the same number of cells with sufficient nutrients and space. We found that the level of circIARS was greater in the supernatants of A549 and NCI-H1299 cells than in those of HBE cells with increasing culture time ( Figure 1C). This finding implies that circIARS may be released from NSCLC cells. In summary, the expression of circIARS in tissues and plasma is consistent with the sequencing results and can be confirmed through in-depth studies.

Figure 1 .

Expression levels of circIARS in NSCLC samples

(A) Sixteen pairs of tissue samples were used to confirm the upregulation of circIARS. (B) Fifty pairs of plasma samples were used to confirm the upregulation of circIARS. (C) circIARS was secreted into the culture medium of NCI-H1299 and A549 cells in a time-dependent manner compared to that of HBE cells.

Basic information on circIARS

Analysis of the Genome Database revealed that circIARS is located on chromosome 9 and is back-spliced by exons ( Figure 2A). We constructed a divergent primer that can be reverse amplified based on the cyclization site of circIARS. qRT-PCR was followed by Sanger sequencing of the resulting product. This verified that the cyclization sites were consistent with the sequences in the CircBank database ( Figure 2B). Subsequently, 2% AGE revealed that the electrophoretic band size of the PCR product was 115 bp, and the products were clearly striped ( Figure 2C). At the same time, the Agilent 2200 assay of the RNA sample revealed a flat peak plot at baseline, an RNA integrity number (RIN) >5, a 28S/18S >1, and an OD260/280 of 2.0, indicating that the RNA integrity requirements were met ( Figure 2D). The above results illustrate the accuracy of the amplification product. Next, the loop properties of circIARS were verified using the actinomycin D test and the nucleic acid exonuclease digestion test. The results indicated that the expression level of circIARS decreased to a lesser extent than that of its linear parent gene IARS ( Figure 2E,F). Furthermore, we designed polymerized primers and reverse primers for the circIARS cyclization site, with GAPDH serving as a negative control, and performed qRT-PCR. cDNA-based PCR products amplified circIARS, while gDNA-based PCR products did not amplify circIARS ( Figure 2G).

Figure 2 .

Characteristics of circIARS

(A) Location and origin of circIARS. (B) Identification of the cyclization site by Sanger sequencing. (C) The length of the circIARS qRT-PCR product was verified by agarose gel electrophoresis (115 bp). (D) Verification of the integrity of the RNA sample by an RNA integrity number (RIN) experiment. (E) Actinomycin D assay confirmed the half-life of circIARS in NCI-H1299 cells. (F) Stability of circIARS confirmed by an RNase R digestion assay. (G) Verification of the ring structure of circIARS.

Comprehensive evaluation of the circIARS assay method

Next, we comprehensively evaluated whether the method for detecting plasma circIARS (qRT-PCR) could be used in clinical studies. As there is no standard internal reference gene for detecting plasma circRNA expression, we compared the expression stability of several commonly used internal reference genes ( 18S, GAPDH, β-actin, β2 M, Tub and RPII) in NSCLC. The results indicate that 18S has the smallest coefficient of variation (CV) ( Supplementary Table S1) and has good stability. First, the precision of qRT-PCR was determined using mixed plasma. The intra- and interbatch CVs of both circIARS and 18S rRNA were less than 5% ( Table 1), verifying the reproducibility of the method. Then, we diluted the cDNA of NSCLC cells at a fold ratio. The regression equation of the circIARS standard curve was y=–2.993 x+18.34 with an R ² of 0.9666; the regression equation of the 18S standard curve was y =–3.502 x+2.477 with an R ² of 0.9741 ( Figure 3A), which indicated good linearity and proved that qRT-PCR is a reliable method for detecting different concentrations of circIARS. Second, to verify its stability, we repeatedly freeze-thawed the patient’s mixed plasma 0, 1, 3, 5, and 10 times; the other group was left at room temperature for 6, 8, 12, 18, and 24 h, respectively. As shown in Figure 3B,C, the cycle threshold (CT) value of circIARS did not change significantly even when the external conditions changed, indicating that the circIARS assay has good stability. In addition, the single peak specificity of the qRT-PCR melting curve ( Figure 3D) verified the specificity of the method. In general, the above results indicated that the assay met the requirements.

Table 1 Intra- and inter-assay reproducibility of circIARS and 18S rRNA

	circIARS	18S rRNA
Intra-assay CV, %	2.87%	2.64%
Inter-assay CV, %	3.72%	3.99%

Figure 3 .

Methodology evaluation of circIARS in NSCLC plasma samples

(A) Standard curves of plasma circIARS and 18S in a tenfold serial dilution. (B,C) Stability of circIARS after incubation at room temperature and after multiple freeze-thaw cycles. (D) The specificity of the PCR products determined by melting curve analysis.

Plasma circIARS expression level and its clinical diagnostic value

To elucidate the diagnostic value of plasma circIARS in NSCLC patients, a large number of clinical plasma samples were collected. This included 100 patients with NSCLC, 50 patients with BPD, 36 patients with SCLC, and 100 normal controls. The circIARS expression levels were measured by qRT-PCR. Surprisingly, circIARS expression was not only observably greater in the plasma of NSCLC patients than in that of healthy individuals ( P<0.0001) but also significantly different between NSCLC patients and patients with BPD ( P<0.0001) or SCLC ( P<0.05) ( Figure 4A). Markers such as CEA, SCC, and Cyfra21-1 have been widely used in the ancillary diagnosis of NSCLC. Although the plasma levels of CEA, SCC, and Cyfra21-1 were significantly greater in NSCLC patients than in healthy controls, the available findings suggest that they do not distinguish well between NSCLC patients and BPD patients or SCLC patients ( Figure 4B–D). As mentioned above, circIARS has shown clear superiority in differentiating NSCLC from BPD and SCLC compared to traditional detection markers.

Figure 4 .

Plasma circIARS, CEA, SCC and Cyfra21-1 expression in NSCLC and the diagnostic value of circIARS

(A) Detection of plasma circIARS levels in NSCLC patients (n=100), BPD patients (n=50), healthy donors (n=100), and SCLC patients (n=36). (B) Detection of plasma CEA levels. (C) Measurement of plasma SCC levels. (D) Detection of plasma Cyfra21-1 levels. (E) ROC analysis of the independent use of plasma circIARS, CEA, SCC and Cyfra21-1 for differentiating NSCLC patients (n=100) from healthy donors (n=100). (F) ROC analysis of the independent/joint diagnostic efficacy of plasma circIARS, CEA, SCC, and Cyfra21-1 for distinguishing NSCLC patients from healthy donors. (G) ROC analysis of plasma circIARS in differentiating NSCLC patients from BPD patients. (H) ROC analysis of plasma circIARS for differentiating NSCLC patients from SCLC patients. BPD, benign pulmonary disease; ROC, receiver operating characteristic. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. NS, no significant difference.

In the next section, we summarize the diagnostic validity of each biomarker. ROC curves were analyzed for NSCLC patients, BPD patients, SCLC patients, and healthy physical examiners. The results demonstrated that the AUC of plasma circIARS was 0.778, which was better than that of CEA, SCC, and Cyfra21-1 (0.589, 0.519, and 0.551, respectively) ( Figure 4E). The combination of circIARS with other markers resulted in a greater AUC, and the combination of the four markers provided the largest AUC ( Figure 4F). Moreover, the AUC of circIARS for distinguishing NSCLC from BPD was 0.739 ( Figure 4G), and the AUC of circIARS for distinguishing NSCLC from SCLC was 0.746 ( Figure 4H). This finding implies that circIARS has good diagnostic performance in distinguishing NSCLC from BPD and SCLC. Subsequently, we constructed single and combined diagnostic models to analyze the sensitivity (SEN), specificity (SPE), total accuracy (ACCU), positive predictive value (PPV), and negative predictive value (NPV) ( Table 2). circIARS was 0.78 for SEN patients, 0.73 for SPE patients, 0.76 for ACCU patients, 0.74 for PPV patients, and 0.77 for NPV patients for distinguishing NSCLC patients from healthy individuals, which was greater than the overall CEA, SCC, and Cyfra21-1 levels. The combination of the diagnosis with other tumor markers revealed different degrees of improvement in SEN and NPV. The above findings suggest that circIARS may be a potential biomarker for the diagnosis of NSCLC and that its combination with other tumor markers can improve diagnostic efficacy.

Table 2 The diagnostic performance of circIARS, CEA, SCC and Cyfra21-1 in differentiating NSCLC patients from healthy subjects

Group	SEN	SPE	ACCU	PPV	NPV
circIARS	0.78 (78/100)	0.73 (73/100)	0.76 (151/200)	0.74 (78/105)	0.77 (73/95)
CEA	0.38 (32/100)	0.78 (78/100)	0.55 (110/200)	0.59 (32/54)	0.53 (78/146)
SCC	0.21 (21/100)	0.91 (91/100)	0.56 (112/200)	0.70 (21/30)	0.54 (91/170)
Cyfra21-1	0.25 (25/100)	0.89 (89/100)	0.57 (114/200)	0.69 (25/36)	0.54 (89/164)
circIARS+CEA	0.87 (87/100)	0.54 (54/100)	0.71 (141/200)	0.65 (87/133)	0.81 (54/67)
circIARS+SCC	0.81 (81/100)	0.66 (66/100)	0.61 (147/200)	0.70 (81/115)	0.78 (66/85)
circIARS+Cyfra21-1	0.83 (83/100)	0.62 (62/100)	0.73 (145/200)	0.69 (83/121)	0.78 (62/79)
circIARS+CEA+SCC	0.88 (88/100)	0.50 (50/100)	0.69 (138/200)	0.64 (88/138)	0.81 (50/62)
circIARS+CEA+Cyfra21-1	0.90 (90/100)	0.48 (48/100)	0.69 (138/200)	0.63 (90/142)	0.83 (48/58)
circIARS+SCC+Cyfra21-1	0.85 (85/100)	0.56 (56/100)	0.71 (141/200)	0.66 (85/129)	0.79 (56/71)
circIARS+CEA+SCC+Cyfra21-1	0.90 (90/100)	0.45 (45/100)	0.68 (135/200)	0.62 (90/145)	0.82 (45/55)

SEN: sensitivity; SPE: specificity; ACCU: overall accuracy; PPV: positive predictive value; NPV: negative predictive value.

Correlation of the circIARS signature with clinicopathological parameters and prognostic monitoring

The correlation of circIARS expression with the clinicopathological parameters of NSCLC patients is another important indicator of whether circIARS can be used as a tumor marker. We summarized the clinical characteristics of 100 NSCLC patients. NSCLC patients were divided into a low-expression group ( n=50) and a high-expression group ( n=50) according to the median circIARS expression level. The chi-square analysis suggested that elevated circIARS expression was associated with the degree of NSCLC differentiation, TNM stage, tumor size, T stage, lymph node metastasis, distal metastasis, and pleural infiltration ( Table 3). Next, pathological parameters that were significantly correlated with each other were analyzed in depth, specifically to detect differences in circIARS expression levels between groups. When plasma circIARS expression was compared between patients with NSCLC stages I–IV and healthy subjects, plasma circIARS was found to be greater in NSCLC patients than in healthy subjects, and plasma circIARS levels also increased with disease progression ( Figure 5A). Moreover, the data showed that circIARS expression increased with increasing infiltration depth and distant metastasis ( Figure 5B,D). However, comparison of plasma circIARS levels in healthy individuals with lymph node metastases from NSCLC revealed that although plasma circIARS was significantly greater in patients with lymph node metastases than in healthy individuals, the available data have not revealed a correlation between circIARS levels and the extent of metastasis ( Figure 5C). In conclusion, plasma circIARS may play an important role in predicting tumor progression.

Table 3 The association between plasma circIARS expression and the clinicopathological parameters in NSCLC patients

Parameter	circIARS level		Total	P-value
Low	High
( n=50)	( n=50)
Age (years)					0.161
< 60	20	27		47
≥ 60	30	23		53
Gender					0.423
Male	22	26		48
Female	28	24		52
Histology					1.000
Adenocarcinoma	40	40		80
Squmous cell carcinoma	10	10		20
Differentiation					0.008**
Well or moderate	36	23		59
Poor or undifferentiation	14	27		41
TNM stage					0.001**
I+II	38	22		60
III+IV	12	28		40
Tumor size (cm)					0.027*
≤ 3	49	44		93
> 3	1	6		7
T stage					0.012*
T1–T2	47	38		85
T3–T4	3	12		15
Lymph node metastasis					0.043*
Negative	41	32		73
Positive	9	18		27
Distant metastasis					0.008**
Negative	42	30		72
Positive	8	20		28
Pleural invasion					0.023*
Negative	42	32		74
Positive	8	18		26
Perineural invasion					0.079
Negative	50	47		97
Positive	0	3		3
Airway spread					0.461
Negative	41	38		79
Positive	9	12		21

* P<0.05, ** P<0.01.

Figure 5 .

Correlation between circIARS expression levels and clinicopathologic parameters of NSCLC patients

(A) The expression levels of circIARS in the plasma of stage I–II NSCLC patients (n=60), stage III–IV patients (n=40), and healthy donors (n=100). (B) The expression levels of circIARS at different stages of tumor invasion and in healthy donors (T1–T2: n=85; T3–T4: n=15; healthy donors: n=100). (C) The expression levels of circIARS in the plasma of NSCLC patients with (n=27) or without lymph node metastasis (n=73). (D) The expression levels of circIARS in the plasma of NSCLC patients with (n=28) or without distant metastasis (n=72). (E) Detection of plasma circIARS expressions in NSCLC patients (n=100), patients after chemotherapy (n=35) and patients with tumor recurrence (n=35). (F) Changes in the plasma expression levels of circIARS in 40 GC patients before and after surgery (n=40).

Next, to explore the dynamic monitoring potential of circIARS, we collected plasma from 35 patients who had undergone chemotherapy and 35 patients who had relapsed. As shown in Figure 5E, plasma circIARS levels were relatively decreased in patients after chemotherapy, and plasma circIARS levels in patients after relapse were not significantly different from those at the initial diagnosis. We further studied preoperative and postoperative plasma samples from 40 paired NSCLC patients. The results suggested that the paired postoperative patients had significantly lower circIARS expression levels than the patients at initial diagnosis ( Figure 5F). This finding suggests that circIARS can be effective in tracking NSCLC postoperatively. As described above, circIARS may be a prognostic biomarker for NSCLC.

Target prediction and enrichment analysis of circIARS

After exploring the potential of circIARS as a molecular marker, we explored the biological mechanisms of circIARS. First, the subcellular localization of circIARS was detected. FISH was performed in NSCLC cells and tissues to visualize the localization of circIARS in clinical samples ( Figure 6A,B). Nucleoplasm separation assays further confirmed that circIARS is localized in the nucleus and cytoplasm, with slightly more circIARS in the cytoplasm ( Figure 6C). This finding implies that circIARS may both act as a competing endogenous RNA (ceRNA) and interact with RBPs to perform its biological functions. We predicted the circRNA-miRNA-mRNA axis ( Figure 6D,F) using databases for miRNA (circAltas, circBank, starBase, and TargetScan), miRNA-binding mRNA (TargetScan, MirPathDB, miRDB, and miRWalk), circRNA-binding mRNA (circAltas, RBPmap, catRAPID and RBPdb), and downstream of mRNA (STRING database) ( Figure 6E,G). All of the above data were plotted using Venny and Cytoscape software.

Figure 6 .

Downstream regulatory network prediction of circIARS

(A) FISH assay of circIARS in NCI-H1299 cells. (B) FISH assay of circIARS in NSCLC tissue. (C) Nuclear and cytoplasmic RNA separation assays were performed on NCI-H1299 and A549 cells to detect circIARS. (D) The potential target miRNAs of circIARS. (E) The potential target mRNAs of circIARS. (F) Prediction of the circRNA-miRNA-mRNA regulatory network of circIARS. The blue hexagon represents circIARS, the blue square represents miRNAs that may bind to circIARS, and the blue circle represents the target mRNA of the corresponding miRNA. (G) Prediction of the circRNA-RNA-binding protein regulatory network of circIARS. The blue hexagon represents circIARS, the blue square represents RBP that may bind to circIARS, and the blue circle represents the target mRNAs of the corresponding RBP. (H,I) GO functional enrichment analysis of the potential target genes. (J,K) Enrichment analysis of the KEGG biological pathways of the potential target genes.

Subsequently, mRNAs predicted by the ceRNA mechanism were intersected, and mRNAs that might interact with circIARS were intersected and analyzed separately via bioinformatics. GO analysis revealed that the target genes enriched by both ceRNA and RBP mechanisms were correlated with cellular processes, biological regulation, and metabolic processes at the BP level; in terms of MF and CC, they were clearly associated with cellular anatomical entities and binding ( Figure 6H,I). KEGG analysis revealed that in terms of environmental information processing and cellular processes, the ceRNA mechanism intersected with the target genes involved in signal transduction and transport, while the mRNAs associated with the RBP mechanism were mainly associated with transcription and neurodegenerative disease ( Figure 6J,K). These results suggest that circIARS may play an important role in the development of NSCLC by targeting downstream miRNAs or mRNAs, and provide a new research direction to further explore the molecular mechanism of circIARS in NSCLC.

Discussion

Lung cancer is the leading cause of cancer-related deaths worldwide [14]. Early diagnosis is particularly important for improving patient survival and postoperative quality of life [ 15, 16]. In contrast, conventional markers are not sensitive or specific enough for diagnosis when used alone [17]. For example, CEA is mostly used in the adjuvant diagnosis of lung adenocarcinoma, but its stage varies widely and has poor specificity according to different studies [ 18, 19], and plasma CEA, SCC, and Cyfra21-1 levels may be related to variables such as age, sex, and smoking habits [20]. Therefore, the identification and application of more effective biomarkers are urgently needed. It has been reported that circRNAs are stably expressed at high levels in body fluids such as blood [21], urine and exosomes [22] and are dynamically regulated [23]. Several studies have also demonstrated its potential as a molecular marker for a variety of tumors [24]. Therefore, it is essential to identify more efficient circRNAs for the diagnosis of NSCLC.

We conducted the first in-depth study of circIARS in NSCLC. It was found to be significantly upregulated in NSCLC tissues by high-throughput sequencing. circIARS is derived from IARS. First, actinomycin D assay, RNase R assay and Sanger sequencing were used to verify the cyclic properties of circIARS. After that, qRT-PCR of large plasma samples revealed that circIARS levels in NSCLC patients were significantly different from those in BPD patients, SCLC patients and healthy physical examiners. Its ability to discriminate NSCLC from BPD and SCLC was better than that of CEA, SCC and Cyfra21-1. ROC curves and single and combined diagnostic models suggested that the clinical diagnostic value of circIARS was greater than those of traditional plasma markers in distinguishing NSCLC patients from healthy individuals, while the combination of circIARS with other markers had different degrees of diagnostic value. Next, the clinicopathological characteristics of the patients were summarized, and the analysis showed that elevated levels of circIARS expression were associated with the degree of differentiation, tumor size, T stage, lymph node metastasis, distal metastasis, and pleural infiltration. In addition, circIARS levels were relatively downregulated in chemotherapy-treated NSCLC patients and relatively increased in relapsed patients, and paired postoperative patients showed a significant decrease in circIARS levels compared with the initial diagnosis, suggesting that circIARS has dynamic monitoring potential. All of the above studies provide a strong basis for the use of circIARS as a plasma marker for NSCLC. However, this study also has several limitations. The samples of all subjects were obtained from Nantong University Hospital, and the results may not be representative; moreover, we need to follow up with relevant NSCLC patients for a long time to understand their health status to obtain the survival curve of circIARS and verify the prognostic ability of circIARS.

We have made some progress in the study of circIARS diagnostic markers, but the mechanism of their differential expressions remains to be investigated. FISH and nuclear plasma separation assays showed that circIARS was localized in both the cytoplasm and nucleus. This finding suggested that circIARS may function both in cells as a miRNA sponge (miRNA sponge) and through the RBP mechanism. On this basis, by bioinformatics analysis, we predicted the circIARS-miRNA-mRNA axis and the downstream mRNAs on which circIARS may act. Among these targets, multiple molecules are involved in the progression of multiple types of cancer. For example, hsa-mir-1270 can regulate the proliferation, migration, invasion and apoptosis of osteosarcoma (OS) cells by targeting the linker protein CGN, thereby inhibiting the progression of OS [25]. miR-330-5p expression is significantly upregulated in thyroid cancer (TC) tissues and cell lines and promotes TC progression by targeting FOXE1 [26]. Dicer1 promotes colon cancer cell migration and invasion by regulating tRF-20-MEJB5Y13 levels under hypoxic conditions [27]. QKI is involved in various biological processes, such as embryogenesis, angiogenesis, glial differentiation, apoptosis and transcription [28].

GO and KEGG enrichment analyses suggested that circIARS may be involved in the progression of NSCLC through complex regulatory mechanisms. These include metabolic reprogramming. In recent years, metabolic reprogramming has been recognized as a new hallmark of cancer [29] and is one of the important features of tumor cells. Several studies have shown that circRNAs are involved in cancer metabolic reprogramming [ 30, 31]. For example, circRPN2 inhibits aerobic glycolysis and metastasis in HCC by accelerating enolase 1 (ENO1) degradation and regulating the miR-183-5p/FOXO1 axis and is a potential prognostic biomarker and therapeutic target for hepatocellular carcinoma [32]; circRIC8B acts as a sponge for miR-199b-5p, preventing it from reducing LPL mRNA level, which promotes altered lipid metabolism and facilitates the progression of chronic lymphocytic leukemia (CLL) [33]; and hypoxia-induced circWSB1 can interact with USP10 to attenuate USP10-mediated p53 stabilization and promote BC progression, providing an alternative breast cancer (BC) prognostic marker and therapeutic target for BC [34]. Whether circIARS functions in NSCLC by participating in metabolic reprogramming needs to be urgently explored. The above study provides great inspiration, and we will conduct further studies and validate our findings in the future.

In conclusion, early diagnosis of NSCLC is crucial for patient treatment, and the stable presence of circRNA in body fluids offers great possibilities for it to act as a marker. In this study, we identified circIARS, which is differentially expressed in NSCLC plasma, by high-throughput sequencing and was found to have great potential as a biomarker for the early diagnosis of NSCLC.

Supplementary Data

Supplementary data is available at Acta Biochimica et Biphysica Sinica online.

COMPETING INTERESTS

The authors declare that they have no conflict of interest.

Funding Statement

This work was supported by the grants from the National Natural Science Foundation of China (Nos. 82272411 and 82072363), the Jiangsu Provincial Medical Key Discipline (Laboratory) (No. ZDXK202240), and the Science and Technology Project of Jiangsu Province (No. BE2023741).