Introduction
Obstructive sleep apnea (OSA) is a highly prevalent sleep disorder characterized by intermittent partial or complete collapse, causing chronic intermittent hypoxia (CIH), sleep fragmentation, and increased inspiratory efforts. These can induce a complex series of pathophysiological changes, leading to the damage of multi-organ and multi-system. OSA has been recognized as an independent risk factor for incident cardiovascular disease [
1], cognitive impairment [
2], and metabolic disease [
3]. Accumulating evidence shows that OSA is also independently associated with increased cancer incidence and mortality [
4,
5].
Lung cancer is reported to be the most common cause of cancer mortality worldwide, with an estimated 1.8 million deaths [
6]. The associations between OSA and occurrence and mortality of lung cancer were supported by more and more studies [
7‐
11]. For instance, a cross-sectional study of 302 subjects demonstrated that sleep apnea and its related nocturnal hypoxia were independently related with an increased prevalence of lung cancer [
7]. In addition, Huang et al. [
8] found that severe OSA was an risk factor of cancer mortality in stage III-IV lung cancer patients. However, the mechanisms whereby OSA results in increased risks of lung cancer incidence and mortality remain unclear.
In recent decades, RNA sequencing has become an established and powerful method for analyzing differential gene expression and differential splicing of mRNAs. RNA sequencing has been used to uncover molecular mechanisms and explore potential diagnostic and therapeutic targets for many kinds of diseases. Model of CIH has generally been used so as to mimic human OSA [
12]. The present study was carried out to identify differentially expressed genes (DEGs) and explore their function in Lewis lung carcinoma (LLC)-bearing mice exposed to CIH by transcriptome sequencing. In addition, the connections between different expression levels of DEGs and prognosis in lung adenocarcinoma were also investigated.
Methods
Animal
Seven-week-old male C57BL/6 mice were obtained from Chinese Academy of Science Laboratory Animals Center located in Shanghai. Mice were allocated to either the normoxia control (NC) (n = 12) or the CIH group (n = 12). All mice were housed in standard cages with a 12-h-day/12-h-night cycle and kept in an animal room. They were free access to water and food. This protocol was approved by the IACUC and IBC Committee in Zhongshan Hospital, Xiamen University (approved number: 2017–015) and performed according to the Guide for the Care and Use of Laboratory Animal. This study is reported in accordance with ARRIVE guidelines.
Intermittent hypoxia (IH) exposure
Mice in CIH group were exposed to IH for 5 consecutive weeks. IH exposure was conducted in the light-on period, from 9 AM to 5 PM. The protocol of IH has been reported in our previous study [
13]. The one cycle of IH consisted of the 50 s of nitrogen, lasting for 10 s, oxygen for 10 s, and subsequent compressed air for 50 s. The range of oxygen concentration was between 21% ± 1% to 6% ± 1%.
Lung cancer cell culture and tumor induction
LLC cells were obtained from CoBioer Biosciences Co., Ltd. China. The cells were cultured in high-glucose Dulbecco's Modification of Eagle's Medium in combination with 10% fetal bovine serum (GIBCO, USA). The right flank of each mouse was injected with the LLC cells at density of 1 × 106 LLC/100 μL PBS after 7 days of IH treatment. The tumor volume was recorded every 5 days after a tumor became palpable. The tumor width (W) and length (L) were collected and used for calculating the tumor volume (mm3) (V = W2 × L/2).
Tissue preparation
All the mice were euthanized with intraperitoneal injection of pentobarbital (150 mg/kg) after 5 weeks of the IH treatment. The tumors were excised, weighted and subsequently stored in RNA locker. 6 lung cancer tissue samples were randomly selected from each group and then subjected for sequencing.
RNA isolation, library preparation, and transcriptome sequencing
The total RNA was extracted from the lung cancer tissue using TRIzol® Reagent (Magen). Oligo(dT) magnetic beads was used for mRNA purification. Then the fragmentation was carried out with divalent cations. Subsequently, the double stranded cDNAs were synthesized. PCR amplification was performed with adaptor-ligated cDNA. Furthermore, we purified PCR products. Finally, the library preparations were sequenced on an Illumina Novaseq 6000 platform at Shanghai Applied Protein Technology Co., Ltd. Raw sequence data was deposited in the in the NCBI SRA database repository, accession number PRJNA948556.
Data processing
Initially, we processed raw data of fastq format through in-house perl scripts. In order to obtain mapped reads, the clean reads were separately aligned to reference genome with orientation mode using HISAT2 software (
http://daehwankimlab.github.io/hisat2/). We used FeatureCounts (
http://subread.sourceforge.net/) to count the reads numbers mapped to each gene. Then fragments per kilobase of transcript per million mapped fragments (FPKM) of each gene was calculated based on reads count mapped to the gene and the length of this gene. Differential expression analysis was performed using the DESeq2 (
http://bioconductor.org/packages/release/bioc/html/DESeq2.html). Genes with
p value < 0.05 and | log2FC |> 1 were judged to be DEGs.
qRT-PCR
Six genes were selected to validate the sequencing data with a real-time PCR technology. Real-time PCR reactions were conducted on an ABI 7500 thermocycler (Applied Biosystems, USA). The 2-ΔΔCt method was utilized to analyze the relative gene expressions. The primer sequences are presented in Supplementary Table S
1. Data were analyzed using GraphPad Prism 5.0. Data were expressed as mean ± SD. Unpaired Student’s t-test was used for comparisons of qRT-PCR data between the two groups. Statistical significance was determined as
p-value < 0.05.
Prognosis analysis
In order to evaluate the prognostic value of DEGs, the Kaplan–Meier survival analysis combined with Cox proportional hazard model were applied based on The Cancer Genome Atlas (TCGA) using R language packages (survival and survminer). R software (version 4.2.1) was used for all statistical analyses.
Discussion
This study established models of CIH and lung cancer in mice to explore the potential mechanisms whereby OSA promoted lung cancer progression. 388 genes were differentially expressed in lung cancer between the CIH-treated LLC-bearing mice and NC group. Further bioinformatics analysis found that these DEGs were related to various pathways involving chemokine signaling pathway, IL-17 signaling pathway, TGF-β signaling pathway, transcriptional misregulation in cancer, natural killer cell mediated cytotoxicity, PPAR signaling pathway. In addition, we identified 11 DEGs, which were associated with unfavorable prognosis in lung adenocarcinoma patients.
Growing evidence suggests a close relationship between OSA and cancer incidence and mortality [
4,
5]. A community-based cohort including 400 residents with 20 years follow-up reported that moderate-to-severe OSA was independently correlated with an increased risk of cancer incidence and mortality [
14]. Nieto et al. [
15] demonstrated a dose–response relationship between OSA and cancer mortality in a community-based sample. A multicenter study with 5427 patients showed that OSA severity was associated with increased cancer mortality after follow-up of 4.5 years [
16]. As for lung cancer type, a prospective study followed up 65,330 women and suggested that no independent association existed between OSA and overall cancer risk. However, significant associations were noted for the smoking-related cancers (e.g., lung cancer) [
11]. A meta-analysis revealed that OSA patients had an nearly 30% higher risk of lung cancer compared with non-OSA subjects after pooling data of four observational studies with 4,885,518 patients [
9]. Liu et al. [
10] showed that the recurrence rate, metastasis rate and mortality increased in lung cancer patients with OSA during the one-year follow-up period. Li et al. [
17] also demonstrated an association between OSA and increased risk for mortality in lung cancer patients. Furthermore, they reveal some molecular pathways involved in sustained hypoxia-induced lung cancer progression some based on Gene Expression Omnibus data base. However, the model of sustained hypoxia did not reflect the pathophysiology of OSA.
These findings have been further supported by animal experiments. In a mouse model of sleep apnea, Almendros and colleagues [
18] found that CIH induced a remarkable increase in melanoma lung metastasis when compared to NC group. Another study also demonstrated that CIH promoted lung epithelial TC1 cell tumor growth and invasion toward adjacent tissues in a mouse model of OSA [
19]. In addition, our previous animal study showed that CIH accelerated lung cancer development and enhanced the vascular endothelial growth factor expression. The mean SUVmax values assessed by micro-PET–CT were considerably higher in the CIH group than the NC group [
20]. Collectively, these results indicated that CIH was an important component of tumor malignant properties, facilitating cancer growth, invasion and migration.
The contributing mechanisms underlying this association are not yet well understood. To gain a better understanding of the mechanisms, Illumina high-throughput technology was utilized to sequence the transcriptome of LLC-bearing mice exposed to CIH. The present study revealed that the up-regulated DEGs were markedly related to chemokine signaling pathway, IL-17 signaling pathway, TGF-β signaling pathway, transcriptional misregulation in cancer. Considerable evidence linked IL-17 with lung cancer. Numasaki et al. [
21] demonstrated that IL-17 obviously increased angiogenesis and promoted the growth of lung cancer transplanted in severe combined immunodeficient mice. Another study reported that intraperitoneal injection of IL-17 resulted in remarkably larger tumors in a LLC mouse model when compared with the control group [
22]. Furthermore, It was reported that TGF-β could promote tumor progression through suppression of immune surveillance, angiopoiesis, and promotion of epithelial to mesenchymal transition [
23,
24]. An increasing body of evidence suggested that the TGF-β pathway activation contributed to poor prognosis in lung cancer patients [
25,
26]. The down-regulated DEGs were enriched in natural killer cell mediated cytotoxicity and PPAR signaling pathway. In fact, the killing mediated by natural killer cells and cytotoxic T lymphocytes represents a crucial mechanism in the immune defense against cancers. And the impairment of natural killer cell mediated cytotoxicity facilitates the growth of lung cancer. A clinical study found that severe OSA had considerably fewer invariant natural killer T cells (iNKT) compared to mild-moderate OSA or no OSA patients and 12 months of continuous positive airway pressure therapy increased the frequency of iNKT cells. Furthermore, they found that hypoxia resulted in impaired cytotoxicity [
27].
The present study also revealed several DEGs related to poorer prognosis in lung adenocarcinoma patients. Planque et al. [
28] analyzed KLK8 mRNAs in 60 NSCLC tissues and in paired unaffected tissues by PCR and found that KLK8-T4 alternative splice variant, alone or in combination was independent marker of poor prognosis in lung cancer. Another study demonstrated that MIR186 could inhibit lung cancer progression through targeting SIRT6 [
29]. In addition, a study showed that the deficiency of PRMT5 enhanced Klrg1
+ terminal CD8
+ T cell development and eliminated antitumor activity [
30]. So this indicated that the DEGs identified in this study could be promising candidates for future research in investigating the association between OSA and increased cancer mortality. The identified DEGs collectively suggest a molecular landscape favoring tumor growth, immune evasion, and inflammatory responses in the context of CIH-induced lung cancer. The dysregulation of pathways related to angiogenesis, immune modulation, and transcriptional control may contribute to the observed poorer prognosis in lung adenocarcinoma.
The current study has some strengths. Firstly, high-throughput sequencing was utilized to investigate the mechanisms linking OSA with progression of lung cancer. Secondly, the findings of the key signaling pathways and prognosis related genes provided a new direction for future study on this issue. Thirdly, the lung cancer prognosis related genes were validated in both animal model and human study, which made the result more reliable. Our study also has several limitations that warrant mention. Firstly, the CIH animal models only simulated one of the major features of OSA and lacked of other features (e.g., hypercapnia and sympathetic hyperactivity). Secondly, although the DEGs and their potential function were identified and analyzed, the functional and mechanistic study on cell line were not conducted. Finally, only LLC-bearing mouse model was established. Whether these results can also be successfully applied to other lung cancer type needed further confirmation. Recognizing the heterogeneity of lung cancer, future studies should incorporate different lung cancer subtypes, such as squamous cell carcinoma and small cell carcinoma.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.