Introduction
Chronic obstructive pulmonary disease (COPD) is one of the top three causes of death worldwide [
1]. COPD is characterized by persistent respiratory symptoms and airflow limitations that are due to airway or alveolar abnormalities [
1]. High morbidity and mortality rates have affected more than 700 billion people, including nearly 100 million people in China [
2]. As the global population ages, morbidity and mortality rates are expected to increase. In the past few decades, pharmacological therapies for COPD have improved; examples include antibiotics, triple inhaled therapy, and alpaha-1 antitrypsin augmentation therapy. However, prognosis for COPD patients remains challenging due to high levels of heterogeneity of disease [
3]. Further complementary therapies are essential to improve the clinical outcomes of COPD patients.
Exercise is used to maintain and restore homeostasis at organismal, tissue, cellular, and molecular levels. It has the potential to prevent or inhibit a wide range of illnesses, including COPD [
4]. Exercise immunology research has revealed that both acute and ongoing exercise have a significant impact on the immune system, especially immune metabolism [
5]. Regular exercise mediates an anti-inflammatory and antioxidant state [
6,
7] and the benefits to COPD patients are considerable [
8]. Aerobic exercise is a cornerstone of pulmonary rehabilitation(PR) to improve health-related quality of life and exercise capacity, as well as reduce dyspnea, hospitalization, exacerbation, and mortality [
9‐
11]. Some studies suggest that exercises can reduce chronic inflammation, improve the diaphragm and cognitive function, and reverse airway remodeling [
12‐
15]. However, despite exercise’s profound benefits for treating COPD, knowledge of how exercise improves health and the molecular mechanism of immunometabolism response to exercise remains limited [
16,
17]. Furthermore, physiological responses to exercise vary between individuals because of the heterogeneous phenotype of COPD, exercise modalities, and levels of intensity [
18]. Additional studies on the molecular mechanisms of exercise intervention in COPD, coupled with advances in the characterization of the human genome, may improve personalized exercise interventions and offer new insight into treatment strategies.
Previous studies have shown molecular insights into the advantages of exercise for people with COPD. For example, nuclear receptor subfamily 4 group A member 3 (NR4A3) induced metabolic responses in skeletal muscle post-exercises [
19], and chemerin improved the diaphragm function by regulating inflammation and metabolism of COPD
5. Other metabolic diseases and secondary ageing [
20] were also precluded and ameliorated by exercise, and wide-scale use of.multi-omics approaches helped illuminate genomic regulation in response to exercise. Most public data referenced skeletal muscle transcriptomics and relevant phosphorylation cascades that activated metabolic enzymes such as AKT and AMPK [
21], as well as alterations in DNA structures [
22]. Further, although there are few predictive tools to access the exercise response for patients, MetaMEx (
https://metamex.eu/) provides the most extensive dataset of skeletal muscle transcription and an online interface to readily interrogate the database [
19]. Nevertheless, muscle samples were difficult to obtain from patients. Plasma samples are more convenient and more readily accepted by patients; however, no study has systematically evaluated the alteration of RNA expression in peripheral blood leucocyte in response to personalized exercise for COPD treatment. Therefore, we aimed to conduct a systems-level analysis of the therapeutic mechanism of personalized precise exercise training (PPET).
High-throughput molecular biological techniques, including a transcriptomics approach, have been used to explore complex biological processes and the role of exercise in systems biology. We used whole-transcriptome sequencing to explore responses to PPET [
23], which were more accepted by patients with exertional dyspnea. Subsequently, a differential expression analysis of mRNA, miRNA, lncRNA, and circRNA (DE-mRNAs, DE-miRNAs, DE-lncRNAs and DE-circRNAs) was performed between pre-and post-exercise groups followed by functional enrichment and interaction prediction analysis. In addition, the results were validated using GEO data. This study may shed light on a novel exercise program that is suitable for a number of COPD patients and may also identify potential biomarkers with various prognostic and therapeutic implications.
Methods
Patients and exercise training protocol
Four COPD patients who benefited from 12 weeks of PR were recruited from clinical cohort research (registration number: ChiCTR2100053232) in Pudong New Area Gongli Hospital. COPD was diagnosed according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria. The GOLD pulmonary function criterion for COPD diagnosis was post-bronchodilator forced expiratory volume in 1 s (FEV1)/ forced vital capacity (FVC) ratio < 0.7. All subjects had no significant cardiac dysfunction, active infection (e.g., hepatitis, tuberculosis), or exercise contraindications such as neurological or psychiatric disorders. The study was approved by the ethics committee of Shanghai Pudong New Area Gongli Hospital and all participants were provided with written informed consent for use of their blood samples for scientific purposes.
Patients were trained on Cycle Ergometer (Qianjing 20,003, China) for three days with different adaptive loads, based on the results of the cardiopulmonary exercise test (CPET) and continuous functional tests. Exercise intensity was individualized moderate intensity, and objectively and quantitatively formulated with CPET (Δ50% load ± 10 Watt) [
24], with Δ50% load = (load at anaerobic threshold – increasing load per minute × 0.75 / 2 + (peak load – increasing load per minute × 0.75) / 2. After the adaptive process, exercise began on the fourth day and lasted for 12 weeks. Exercise frequency was determined according to the individualized response: 1 ~ 4 times / day, 5 ~ 7 days / week. Patients warmed up at a load of zero watts for five minutes and then at a personalized load sustained for 30 min of effective exercise (if revolutions per minute (RPM) < 60, patients could rest then continue the exercise).
Sample preparation
Three milliliters of fresh whole blood were harvested from pre- and post-exercise COPD patients. Peripheral blood leucocytes were isolated from 3 ml of fresh whole blood within two hours of collection, by Pancoll gradient centrifugation of one collected Vacutainer EDTA-tube, then frozen in liquid nitrogen and stored at – 80 °C for further studies.
RNA isolation and library preparation
Total RNA was extracted using the TRIzol reagent according to the manufacturer’s protocol. RNA purity and quantification were evaluated using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The libraries were constructed using TruSeq Stranded Total RNA with Ribo-Zero Gold (illumina, Cat.No. RS-122-2301) according to the manufacturer’s instructions.
RNA sequencing and differentially expressed RNAs analysis
The libraries were sequenced on an Illumina HiSeq X Ten platform, and 150 bp paired-end reads were generated. Approximately 95 million raw reads for each sample were generated. Raw data (raw reads) of fastq format were first processed using the Trimmomatic software [
25]. In this step, clean data (clean reads) were obtained by removing reads containing adapter and ploy-N or low quality reads from raw data. Approximately 93 million clean reads for each sample were retained for subsequent analyses.
Sequencing reads were mapped to the human genome (GRCh38) using HISAT2 [
26]. For mRNAs, FPKM [
27] of each gene was calculated using Cufflinks [
28], and the read counts of each gene were obtained by HTSeq-count [
29]. Differential expression analysis was performed using the DESeq (2012) R package [
30]. P-value < 0.05 was set as the threshold for significant differential expression. For lncRNAs, the transcriptome from each dataset was assembled independently using the Cufflinks 2.0 program [
28]. All transcriptomes were pooled and merged to generate a final transcriptome using Cuffmerge (Cufflinks 2.0). All transcripts that overlapped with known mRNAs, other non-coding RNA, and non-lncRNA were discarded. Next, the transcripts longer than 200 bp and the number of exons > 2 were selected, and the CPC (v 0.9-r2) [
31], PLEK (v 1.2) [
32], CNCI (v 1.0) [
33], Pfam (v 30) [
34] were used to predict transcripts with coding potential. The novel predicted lncRNAs were obtained through these processes. The characteristics (including length, type, number of exons) of lncRNA were analyzed after screening. Then, the novel predicted lncRNAs and known lncRNAs (from NCBI and Ensemble database) were used for expression calculations and differential screening. circRNAs were identified using CIRI (v2.0.3) [
35] and the expression of circRNAs were calculated using RPM (spliced reads per millon mapping) [
9]. Differential expression analysis was completed using the DESeq (2012) R package. All sequencing processes and analyses were performed by OE Biotech Co., Ltd. (Shanghai, China).
Gene Ontology (GO) term and KEGG pathway analysis
The gene list of DE-mRNAs were uploaded to the Database for Annotation, Visualization, and Integrated Discovery (DAVID,
https://david.ncifcrf.gov/), which is a comprehensive set of functional annotation tools for researchers to understand biological meaning behind large sets of genes The official gene symbol was selected as an identifier, and homo sapiens was selected as the species. Finally, Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis enrichment results were obtained [
36,
37]. With the enriched gene count ≥ 2 and p < 0.05 significance threshold, GO terms and pathways were considered significant. The top five results in ascending order were displayed in this study.
Gene set variation analysis (GSVA)
The gene list for each biological function was obtained from the AmiGO2 portal (
http://amigo.geneontology.org). The biological functional enrichment score of each patient was calculated by Gene Set Variation Analysis (GSVA) analysis, using GSVA package (R environment) under default parameters [
38]. Diverging bars of the enrichment results was drawn with the package (R environment).
PPI network and module analysis
The interaction between DE-mRNA encoded proteins was analyzed by STRING (version 10.0,
https://string-db.org/cgi/input.pl) database. We input all DE-mRNA sets, and the species was set as human. The PPI network was built by Cytoscape software (version 3.9.1;
https://cytoscape.org/). The Cytoscape’s plug-in MCODE [
39] was used to examine the PPI network's most important clustering modules (version2.0.0). We set the PPI score parameter to 0.7 to obtain the interaction pairs that were most closely related. The threshold for the significant clustering module gene was score ≥ 2. GO enrichment analysis was conducted for the top 10 clustering module genes. The GO terms with enriched gene count ≥ 2 and the significance threshold p < 0.05 were considered significant.
Co-expression mRNAs of DE-cirRNA and DE-LncRNA
The Pearson's correlation coefficients of each DE-mRNA and DE-lncRNA and each DE-mRNA and DE-circRNA were calculated. The cor function in R software was used to calculate these correlation coefficients. A screen of |R|> 0.9 and p < 0.05 was used for co-expression relationships.
ceRNA network construction
Competing endogenous RNAs(ceRNAs) are the lncRNAs, circRNAs, and mRNAs that competitively bind miRNAs and act as miRNA sponges. The lncRNA, mRNAs, and circRNA regulatory relationships with DE-miRNA were predicted using the StarBase (
http://starbase.sysu.edu.cn/). The lncRNA, circRNAs and mRNA that were substantially differently expressed and regulated by the same miRNA were screened, using DE-lncRNAs, miRNAs, and mRNAs as well as regulatory relationships of DE-miRNA that were predicted using the StarBase. The lncRNA-miRNA-mRNA and circRNA-miRNA-mRNA networks were constructed with Cystoscope software v3.8.0 (San Diego, CA, USA) to investigate the role and interactions between ncRNAs and mRNAs after rehabilitation treatment.
Data validation
GEO data were used for RNA data validation (GSE76705 for mRNAs validation, GSE24709 and GSE 61741are for miRNAs validation). MiRNAs and mRNAs expression matrices and annotation information were downloaded from GEO database separately. The matrices were submitted to a differential expression analysis in COPDs against normal controls using the limma R package [
40]. A criteria for substantial DE-mRNA and DE-miRNA was defined as p-value 0.05. Expression levels of each genes were performed using GraphPad 8.
RT-PCR for DE-RNAs
cDNA was synthesized using a TransScript All-in-One First-Strand cDNA Synthesis SuperMix (Transgen Biotech, Beijing, China), was performed. PCR was performed using a Bio-Rad PCR instrument (Bio-Rad, Hercules, CA, U.S.A.) with 2 × Taq PCR Master Mix (Solarbio, Beijing, China) following the manufacturer’s instructions. The fold changes were calculated by means of relative quantification (2 − △△Ct method). PCR primers are described as below: mirRNA 144: forward 5′-UUCAAUCAACUUUACUGUAA-3′and reverse 5′-UCAUGUAGUAUAUGACAU-3′; CCL23: forward 5′-CATCTCCTACACCCCACGAAG-3′and reverse 5′-GGGTTGGCACAGAAACGTC-3′; CPA3: forward 5′-GGGTTTGATTGCTACCACTCTT-3′and reverse 5′-GCCAAGTCCTTTATGATGTCTGC-3′; PLCB4: forward 5′- TTGACAGATACGAGGAGGAATCC-3′and reverse 5′GAGGGAGCATTCTAGCACCTG-3′; IGF2R: forward 5′- GCTTTGACAGCGAGAATCCC-3′and reverse 5′-TCCTACAGCAAGTGGTCAGC-3′.
Statistical analysis
R 4.1.3 was used for bioinformatics analysis. Data processing and analysis were performed using GraphPad 8 (GraphPad Software, Inc., La Jolla, CA, USA, www. graphpad.com).We used paired Student t-test for clinical characters to analyze the differences between groups with double tail test. P < 0.05 was considered statistically significant.
Conclusion
COPD is a heterogeneous disease, although pharmacological therapies for COPD have improved, they produce insufficient results. Exercise training is a vital tool in the fight against the global epidemic of aging and metabolic disease
3, including COPD. Numerous studies have shown the effects of exercise on the immune system. Regular exercise performed at moderate intensity leads to an anti-inflammatory, anti-infection, and controlled immunological metabolic state [
42,
43]. However, the exact mechanism that underlies its effects on COPD remains unclear. Since the basis of exercise limitation in COPD patients is breathlessness [
16], we used personalized, precise exercise training to inhibit these respiratory symptoms to achieve the daily exercise volume. Our study found individualized aerobic exercise training improved peak VO2, CAT score, and 6MWT (see Fig.
1), compared to healthy cohorts in a meta-analysis study [
44]. According to the findings, personalized, precise exercise training is appropriate for COPD and has promising future prospects. But the mechanism by which exercise increases exercise tolerance and improves respiratory symptoms remains unknown.
We employed a whole-transcriptome sequencing technique to investigate the potential role of RNAs in COPD following exercise training and provided a comprehensive look at COPD patients' pre- and post-exercise caused RNA-level modifications. 570 lncRNAs, 2064 circRNAs, eight miRNAs, and 86 mRNAs were found to have significantly altered expression in COPD patients pre- and post-exercise using log
2FC ≥ 1 and p-value < 0.05 as the criteria (Fig.
2). Direct function enrichment analysis revealed that DE-genes were involved in brown fat cell differentiation, muscle contraction, G-protein coupled receptor signaling pathway, and detection of chemical stimulus, T cell chemotaxis, DNA replication, and antibacterial and antimicrobial humoral response. PPI network analysis of these DE-genes identified several hub gene: CDK1, TTK, HJURP, DLGAP5, PLCAF, GINS2, TYMS, DTL, KIFC1, and ELANE. Enrichment analysis of those genes related to the biologic process of DNA replication, chromosome segregation, telomere organization, and anti- inflammation. We also performed GSVA analysis with non-zero expressed mRNAs in each patient. Eleven hub genes (DTL, TTK, FPR3, CDK1, HJURP, GINS2, IGF2R, NDUFAB1, NDUFS8, PLCB4, and SCD) and some vital pathways linked to regular exercise training were discovered: chemokine receptor biding, tricarboxylic acid (TCA) cycle, fatty acid metabolism, and oxidative phosphorylation, which were known to be associated with the pathophysiology of COPD [
3,
45,
45]. In addition, enrichment analysis of co-expression genes linked to miRNA, lncRNA, and circRNA also identified the function of anti-infection and T-cell chemotaxis. According to the ceRNA complex network, we found hsa-miR-144-3p, hsa-miR-1277-5p, and hsa-miR-7c-5p were significantly enriched in. Moreover, majority of the DE-mRNAs and a small number of miRNAs were effectively confirmed using GEO data. Different threshold choices or sample variances may be the cause of the discrepancy between our sequencing results and GEO data.
Despite limited evidence, the key genes listed in DE-mRNAs from our results may be linked to RNA replication immune defense, anti-inflammation, mitochondrial functions, and ATP and protein degradation processes. Genes such as CDK1, TTK, HJURP, GINS2, DTL, ELANE, FPR3, PLCB4, and IGF2R may contribute to the pathology of of COPD. CDK1, a member of the cyclin-dependent kinase family which is up-regulated in several cancers by regulating cell cycle progression and activating of JAK/STAT3 signaling[
46,
47], participated in the pathogenesis of pulmonary arterial hypertension (PAH) by influencing mitochondrial dynamics and the cell cycle [
48]. Further studies are needed to validate them as a target for cancer therapy. TTK is a critical component of the spindle assembly checkpoint [
49]. It is a biomarker for non-small cell lung cancer prognosis, and its overexpression accelerates the tumor’s progression [
50,
51]. HJURP, a centromeric protein (chaperone), has been shown to increase in lung tumors and COPD and is essential for the insertion and maintenance of the histone H3-like variation CENPA at centromeres [
52,
53]. GINS2 promoted cell proliferation, migration, invasion, and EMT via modulating PI3K/Akt and MEK/ERK signaling pathways [
54], GINS2 knock-down stimulated inflammation and apoptosis in microglia [
55]. DTL, a homolog of E3 ubiquitin ligase that belongs to the DCAF protein family, was reported to enhance the motility, proliferation, and invasion of cancer cells [
56,
57], and also significantly decrease total glucose consumption and lactate production [
58]. ELANE, a factor that contributes to a protease-antiprotease imbalance and may cause inflammatory lung illnesses [
59], induces autophagy, which in turn induces lung epithelial cell apoptosis and pulmonary emphysema through the overexpression of PGF [
60]. Further research is needed to determine if RNAs with differential expression are involved in these biological activities.
RNA-seq-based networks have proven to be a valuable tool for investigating functional noncoding RNAs and their functional mechanisms in many disease models. We discovered that has-miR-144-3p plays an important role in the ceRNA network. Evidence shows that miR-144-3p, which is improperly regulated, suppresses tumor growth in a variety of cancer types [
61]. In addition, miR-144-3p was downregulated in the peripheral blood of COPD patients compared to normal controls according to GSE24709 and GSE6141. However, our results indicated that miR-144-3p was down-regulated post-exercise when compared to pre-exercise COPD. Anti-mir-144-loaded extracellular vesicles was proven to protect against obstructive sleep apnea or chronic intermittent hypoxia-associated endothelial dysfunction [
62].The miR-144 family was reported to target NF-kB pathways and play a pro-inflammatory role in coronary artery disease [
63]. Thus, we speculated that exercise training may improve progression of COPD by regulating the expression of miR-144-3p.
Interestingly, up-regulated hsa-let-7c and down-regulated hsa-miR-1277 post-exercise in COPD, which were validated as expressed but without difference compared to normal controls in GEO data, play an important role in the ceRNA network of our study. Evidence suggests that overexpression hsa-miR-1277 could ameliorate IL-1β-induced CHON-001 cell injury and inhibit the progression of Parkinson’s disease [
64,
65], but these studies were all in vitro. Future mechanistic investigations are thus necessary to ascertain the impact of exercise on COPD and the function of miR-144-3p and other ncRNAs. However, because of the paucity of study in this field, a significant portion of the DE-lncRNAs and DE-circRNAs were not previously identified. In general, the discovery of RNAs changes following exercise improves our comprehension of the cardiopulmonary-regulation mechanisms of aerobic exercise in COPD, which may enhance the effectiveness of this non-pharmacological intervention and result in the discovery of novel alternative therapeutic targets for COPD patients.
In summary, while the influence of exercise on multiple organs is well documented, our knowledge of how this occurs at the cellular and molecular level is mostly limited to skeletal muscle. According to GO enrichment, KEGG, and GSVA analysis, this comprehensive study of noncoding RNAs and mRNAs uncovers regulatory pathways and key DE-genes involved in the effectiveness of aerobic exercise on COPD. Furthermore, co-expression networks (lncRNA–miRNA–mRNA and circRNA–miRNA–mRNA) were constructed to understand the regulatory roles of these mRNAs and ncRNAs. The observed DE-mRNAs and DE-ncRNAs may provide the foundation for understanding the genetic basis and ceRNA mechanism of exercise in COPD. The results provide molecular insights related to the effects of exercise on COPD and inform future therapeutic selection. Further research into the molecular mechanisms underlying the expression changes on differentially expressed mRNAs and ncRNAs may reveal more RNA therapeutic targets.
The results of this study offer implications for further investigation. First, more COPD patients who volunteer for 12 weeks of supervised exercise are expected to participate in RNA sequencing and bioinformatics analysis based on sample size estimation. This will enhance research methodology for highly confident differential expression identification. Second, future comprehensive investigations involving in vivo and in vitro trials are necessary because the RNA regulatory networks and data validation were solely based on bioinformatics predictions and GEO database, and lacking sufficient sample sizes for verification. Research on the potential functions and evolutionary conservation of RNAs can benefit from using COPD model exercises. Moreover, repeatability in different COPD phenotypes and RNA alterations in the current study are incomplete. Our team is researching these key RNAs’ repeatability and mechanisms post-exercise in COPD models, as well as expanding the COPD participant pool.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.