Abstract

   The aim of this study was to identify miRNAs in plasma exosomes as
   noninvasive biomarkers for the early diagnosis of lung adenocarcinoma
   (LUAD). First, exosomal miRNA profiling of three patients with early
   LUAD and three patients with benign lung disease were screened by
   next‐generation sequencing (NGS) method. Sequencing results showed that
   154 exosomal miRNAs were differentially expressed in the plasma of LUAD
   patients, among which 68 miRNAs were up‐regulated and 86 miRNAs were
   down‐regulated. [36]GSE137140 is a GEO database containing serum miRNAs
   sequencing data from 1566 lung cancer patients and 1774 non‐cancer
   patients controls. When comparing the sequencing data, it was found
   that most miRNAs (37/68) up‐regulated in our LUAD group were also
   significantly up‐regulated in [37]GSE137140, suggesting that
   circulating miRNAs in lung cancer patients may be enriched in plasma
   exosomes. In [38]GSE137140, the AUC of the combination of hsa‐miR‐103b,
   hsa‐miR‐29c‐5p and hsa‐miR‐877‐5p was 0.873, showing great potential as
   new tumor markers. To our knowledge, these three exosomal miRNAs have
   not been reported in lung cancer research. Furthermore, bioinformatics
   tools were used to analyze the target genes of three candidate miRNAs,
   which were indeed closely related to the occurrence and development of
   lung cancer. Bioinformatics algorithms deduced a highly conserved
   sequence in the 3’‐UTR of SFRP4, FOXM1 and TMEM98 that could be bound
   with miR‐103b/877‐5p/29c‐5p. A luciferase assay indicated that
   miR‐103b/877‐5p/29c‐5p directly targeted the 3’‐UTR of SFRP4, FOXM1 and
   TMEM98, respectively. Finally, three candidate miRNAs were validated by
   qRT‐PCR in 17 early LUAD samples and 17 control plasma samples.
   Integration of bioinformatics analysis and experimental validation
   identifies, this study provides novel insights into miRNA‐related
   networks in LUAD. Hsa‐miR‐103b, hsa‐miR‐29c‐5p, and hsa‐miR‐877‐5p may
   be used as diagnostic biomarkers for early LUAD.

   Keywords: bioinformatics, exosome, lung adenocarcinoma, miRNA, qRT‐PCR
     __________________________________________________________________

   Integration of bioinformatics analysis and experimental validation,
   this study provides novel insights into miRNA related networks in LUAD.
   Plasma exosomal hsa‐miR‐103b, hsa‐miR‐29c‐5p and hsa‐miR‐877‐5p may be
   used as diagnostic biomarkers for early LUAD.

   graphic file with name CAM4-11-4411-g008.jpg

1. INTRODUCTION

   Although morbidity and mortality rates have been declining, lung cancer
   remains one of the most common malignancies in both the United States
   and China.[39] ^1 , [40]^2 Non‐small cell lung cancer (NSCLC) is the
   most common type of lung cancer, which can be further classified into
   lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC).
   Most lung cancer patients are diagnosed as advanced, with a 5‐year
   survival rate of less than 15% after surgery and anticancer
   treatment.[41] ^3 Currently, computed tomography (CT) has been used as
   a screening method for NSCLC, but its clinical application is limited
   by its lack of specificity and ionizing radiation.[42] ^4 Blood‐based
   testing is an ideal non‐invasive diagnostic method, such as serum
   markers carcinoembryonic antigen (CEA) and cytokeratin 19 fragments
   (CYFRA21‐1) can help diagnose early lung cancer. They are generally not
   recommended as biomarkers for lung cancer due to their lack of
   sensitivity and specificity.[43] ^5 Therefore, it is urgent to find
   specific biomarkers with high sensitivity to identify lung cancer.

   MicroRNAs (miRNAs) are highly conserved noncoding RNAs, approximately
   22 nucleotides in length, and play a central role in regulating gene
   expression.[44] ^6 Currently, an increasing number of abnormally
   expressed miRNAs have been found in various tissues and body fluids,
   and have been reported as biomarkers for diagnosis, prognosis, or
   monitoring of NSCLC.[45] ^7 , [46]^8 In human blood, miRNAs from donor
   cells can be incorporated into exosomes and transported through the
   blood to recipient cells, where they play important roles in new
   locations.[47] ^9 , [48]^10

   Exosomes are lipid bilayer vesicles released by various cells that
   mediate intercellular communication and are involved in the occurrence,
   development, and progression of cancer.[49] ^11 Exosomes are rich in
   miRNAs and enhance miRNAs tolerance through lipid bilayers that protect
   miRNAs from RNA enzyme degradation.[50] ^12 , [51]^13 Exosomal miRNAs
   should be the starting point for early biomarker studies, reducing the
   possibility of false‐negative results for miRNAs with low
   abundance.[52] ^12 Several studies have explored the potential
   application of exosomal miRNAs in the clinical diagnosis and prognosis
   of lung cancer.[53] ^14 , [54]^15

   In this study, exosomes were isolated from plasma of patients with
   early LUAD and benign lung disease. Exosomal miRNA profiling of three
   early LUAD patients and three patients with benign lung disease was
   performed using NGS method to identify LUAD‐specific miRNAs. Then,
   bioinformatics methods were used to analyze candidate miRNAs.
   Luciferase assay results showed that miR‐103b / 877‐5p / 29c‐5p
   directly targeted the 3’‐UTR of SFRP4, FOXM1 and TMEM98, respectively.
   Finally, three upregulated exosomal miRNAs were confirmed by qRT‐PCR.
   Our work suggests that the altered expressed exosomal miRNAs in plasma
   may be used as a noninvasive biomarkers for the diagnosis and treatment
   of LUAD at an early stage.

2. MATERIALS AND METHODS

2.1. Collection of patient samples

   Twenty patients with early LUAD and twenty patients with benign lung
   disease were included in the sequencing and validation studies.
   Detailed clinical data are summarized in Table [55]S1–S2. Three early
   LUAD patients and three controls were randomly selected for NGS method
   (Table [56]1), and the remaining samples were further verified by
   qRT‐PCR. Peripheral blood samples were collected with EDTA
   anticoagulant tube. Peripheral blood samples were collected from all
   patients before treatment procedures (e.g., surgery, preoperative
   radiotherapy, and chemotherapy). All patients with lung cancer were
   confirmed to be lung adenocarcinoma by histopathology. According to the
   eighth edition of the American Joint Committee on Cancer (AJCC),
   information on TNM staging, and differentiation of the tumor was
   retrospectively obtained from the patient's medical records. Ethical
   approval was waived off.

TABLE 1.

   Clinical characteristic of the patients
                       Testing set (N = 6) Validation set (N = 34) p value
   Categories
   Tumor               3                   17
   Control             3                   17
   Gender                                                          0.534
   Male                4                   17
   Female              2                   17
   Age (years), median 58.83 ± 7.57        54.09 ± 15.86           0.863
   TNM stage
   Ia                  2                   16
   Ib                  1                   1
   [57]Open in a new tab

2.2. Isolation of plasma exosome

   According to the manufacturer’s specification, ExoQuick™ (system
   Biosciences, CA, USA) is a commercial kit for the separation of
   exosomes from EDTA anticoagulated plasma. In brief, 200 μl of collected
   plasma was lightly mixed with 50 μl EQ reagent and placed at 4°C for
   30 min. After centrifugation at 1500g for 30 min, the exosomes are
   precipitated at the bottom of the centrifuge tube.

2.3. Characterization and quantification of exosomes

   Negative staining of exosomes suspension was performed and transmission
   electron microscopy was performed to identify the isolated exosomes. A
   total of 10 μl exosome suspension was deposited on a copper network,
   and phosphotungstate was coated on the grid. After drying for 2 min, 80
   KV electron microscope was used for detection and imaging, and the
   final transmission electron microscope (HT‐7800) imaging results were
   obtained. Exosome suspension was treated with 200 μL RIPA lysis buffer
   (Beyotime, Shanghai, China) to extract protein. Electrophoresis was
   performed on 10% SDS‐PAGE gel, imprinted on PVDF membrane, and
   incubated with primary antibodies (CD63 (AB68418), TSG101
   ([58]AB125011), and calnexin (10427‐2‐AP, Promega)) at 4°C overnight.
   The PVDF membrane was then rinsed three times with a TBST‐containing
   buffer and incubated with secondary antibodies for 1.5 h at room
   temperature. The immune response bands were detected using a Western
   ECL kit (Beyotime). The exosome precipitates were resuspended with PBS
   and mixed evenly with a vortex agitator. Diluted samples were taken and
   exosome size was measured using a ZetaView device (Particle Metrix).

2.4. Libraries preparation and RNA‐seq analysis

   High‐throughput transcriptome sequencing was carried out by the BGISEQ
   sequencing platform. In short, linear DNA amplification sequencing was
   performed using DNA nanospheres (DNB). According to the manufacturer's
   instructions, Unique molecular identifier (UMI) was introduced during
   the library construction, which was connected to the cDNA molecule in
   the early stage of library construction, and tagged each molecule in
   the original sample. RNA libraries were quantified using Agilent 2100
   BioAnalyzer (Agilent Technologies). Finally, the sequencing was
   performed using the combined probe‐anchored polymerization (cPAS)
   technique according to the amplification procedure.

2.5. Prediction of miR‐103b/877‐5p/29c‐5p target genes

   MiRWalk 2.0
   ([59]http://zmf.umm.uni‐heidelberg.de/apps/zmf/mirwalk2/miRretsys‐self.
   html), is an easy to access open database, which contains the
   prediction data obtained by machine learning algorithm.[60] ^16 This
   database has a clear structure, intuitive interface, and user‐friendly
   design. The data in the database are free and constantly updated. The
   target genes of miR‐103b/877‐5p/29c‐5p were predicted using this
   database in conjunction with three other effective and promising target
   gene prediction tools for microRNAs(miRanda:
   [61]http://www.microrna.org/microrna/getDownloads.do, RNA 22:
   [62]https://cm.jefferson.edu/rna22v2/, Targetscan 6.2:
   [63]http://www.targetscan.org/cgi‐bin/targetscan/data_download.cgi?db=v
   ert_61).[64] ^17 , [65]^18 , [66]^19

2.6. Functional enrichment analysis and transcription factor prediction

   FunRich database([67]http://www.funrich.org/), an online annotation and
   analysis tool to understand bioinformatics information.[68] ^20 The
   FunRich database was used to study the GO annotation of target genes
   and predict biological signaling pathways and transcription factors.

2.7. 3’‐UTR luciferase assay

   Bioinformatics method was used to analyze the binding of
   miR‐103b/877‐5p/29c‐5p to the 3’‐UTR region of SFR4, FOXM1 and TMEM98,
   respectively. According to the analysis results, Wild‐type (WT) and
   Mutant‐type (MUT) 3’‐UTR of SFRP4, FOXM1 and TMEM98 were synthesized by
   Hefei qinheng. Then, 293T cells were co transfected with miR‐NC and
   miR‐103b/877‐5p/29c‐5p mimics and WT and MUT 3’‐UTR vectors of SFRP4,
   FOXM1 and TMEM98 for 24 h, respectively. The luciferase activity was
   carried out using Luciferase Assay System (Promega, E1910).

2.8. Quantification of miRNA expression with qRT‐PCR

   Total RNA was extracted from the plasma isolated exosomes using TRIzol
   reagent (Invitrogen, Carlsbad) according to the manufacturer's
   instructions. A similar amount of Caenorhabditis elegans cel‐miR‐39
   (RiboBio) was incorporated into exosome sample for external calibration
   of RNA extraction, reverse transcription, and miRNA amplification.
   According to the manufacturer's instructions, the PrimeScriptTM RT
   Master Mix kit (Takara) was used to reverse transcribes the extracted
   total exosomal RNA to obtain cDNA. PCR amplification of cDNA was
   performed using TB Green® Premix Ex TaqTM II (Takara) at a total
   reaction volume of 20 μl. The qRT‐PCR was run on ABI QuantStudio 5
   (Applied Biosystems) to amplify and detect miRNAs with TB Green dyes.
   The qRT‐PCR mixture was incubated at 95°C for 30 s, followed by
   40 cycles of 5 s at 95°C and 60°C for 30 s. All qRT‐PCR reactions were
   performed in triplicates, and miRNAs expression was calculated using
   the comparative cycle threshold (Ct) method. Using 2^−ΔΔCt method
   compared with the foreign reference of microRNAs (cel‐miR‐39), ΔCt = Ct
   miRNAs ‐ Ct cel‐miR‐39, calculate the relative expression of miRNAs.
   Specific primers designed for miRNAs are listed in Table [69]S2.

2.9. Statistical analysis

   Statistical analysis was performed using Social Science Statistical
   Program (SPSS) 22.0 software and graphs were generated using GraphPad
   Prism 5.0 (GraphPad Software). The qRT‐PCR results of the LUAD group
   and the control group were analyzed by t‐test. The binary logic model
   is constructed by SPSS software. The diagnostic accuracy of candidate
   miRNAs and their combinations was evaluated by the receiver operating
   characteristic (ROC) curve, and the area under the curve (AUC) was
   calculated. p < 0.05 was considered statistically significant.

3. RESULTS

3.1. Characterization of isolated plasma exosomes

   TEM, western blotting, and NTA were performed to identify the
   morphology and concentration of exosome particles. TEM analysis showed
   that the exosomes isolated were elliptical in size and ranged from 50
   to 150 nm (Figure [70]1A). NTA results showed that the concentration of
   nanoparticles was 10^7–10^11 Particles/ml, and the particle size was
   30–200 nm (Figure [71]1B). Western blotting showed that exosome markers
   CD63 and TSG101 were detected in exosomes isolated from plasma.
   Conversely, there was no negative marker Calnexin for exosomes
   (Figure [72]1C). The above results indicate that the plasma vesicle
   structure of isolated clinical patients was identified as exosomes.

FIGURE 1.

   FIGURE 1
   [73]Open in a new tab

   (A) TEM was used to analyze exosomes, which exhibited a cupshaped
   membrane morphology with a diameter of 60–150 nm. (B) The NTA results
   suggested that the majority of vesicle particles were mainly between 60
   and 150 nm in diameter. (C) Exosomal markers CD63 and TSG101 were all
   detected in the exosomes enriched fractions isolated from the plasma,
   and Calnexin, a negative marker of exosome was absent in our isolated
   exosomes enriched fraction samples.

3.2. Plasma exosomal miRNA expression profiles

   In order to obtain the general profile of exosomal miRNA in plasma of
   LUAD patients, six samples (three LUAD patients and three non‐tumor
   controls) were tested by NGS method (bioproject ID: PRJNA699702). In
   the analysis of differentially expressed miRNAs (DEMs) (|log2
   [FC]| > 1, p < 0.001), 68 miRNAs were upregulated, and 86 were
   downregulated in LUAD patients. When analyzing disorder miRNAs of LUAD,
   Venn diagram was used to identify overlapped and nonoverlapped miRNAs.
   Compared with the differentially expressed miRNAs in the GEO database,
   119 miRNAs were differentially expressed in these two analyses (77.3%)
   (Figure [74]2A). Among the 68 upregulated miRNAs, 37 overlapped with
   the database (Figure [75]2B). Researchers have studied some miRNAs with
   significant differences in sequencing data. For example,
   hsa‐miR‐200a‐3p is upregulated in peripheral blood and exosomes of LUAD
   patients, and its expression level is significantly higher than that of
   patients with tubercular pulmonary disease and other benign pulmonary
   lesions.[76] ^21 Compared with the control group, the plasma and
   exosomal miR‐660‐5p of NSCLC patients were significantly increased, and
   miR‐660‐5p may inhibit the proliferation, migration, and invasion of
   NSCLC by targeting KLF9, providing a potential therapeutic target for
   NSCLC.[77] ^15

FIGURE 2.

   FIGURE 2
   [78]Open in a new tab

   (A) A Venn diagram shows the shared differential expression of miRNAs
   between the TCGA dataset and our plasma exosomal miRNA dataset. (B)
   Upregulated miRNAs sharing differential expression between plasma
   exosomes and data sets.

3.3. Validation of the miRNA‐seq data

   Among all the differentially expressed candidate exosomal miRNAs, three
   new miRNAs with high expression in lung adenocarcinoma were selected
   for validation. So far, the miRNAs have not been reported in the
   literature after a PubMed search. In [79]GSE137140, hsa‐miR‐103b,
   hsa‐miR‐29c‐5p, and hsa‐miR‐877‐5p achieved an AUC of 0.798(95%
   confidence interval [CI]: 0.784–0.813), 0.719 (95% CI: 0.703–0.736),
   and 0.770 (95% CI: 0.752–0.788) (Figure [80]3), respectively. In brief,
   all three miRNAs provided meaningful AUC values, which were of great
   significance in distinguishing patients with lung cancer from the
   control group.

FIGURE 3.

   FIGURE 3
   [81]Open in a new tab

   Validation of miRNA as a biomarker for lung cancer. (A) miR‐29c‐5p. (B)
   miR‐103b. (C) miR‐877‐5p. The ROC curve of the microarray of
   [82]GSE137140 is shown in blue.

   Furthermore, the integration of the two candidate miRNAs was verified
   by using a logical model, and better performance was obtained than that
   of a single miRNA. The formula for the logical models are as follows:
   hsa‐miR‐103b + hsa‐miR‐877‐5p = hsa‐miR ‐ 103b * 0.488 + hsa‐miR‐877‐5p
   * 0.459 ‐ 1.627, hsa‐miR‐29c‐5p + hsa‐miR‐103b = hsa‐miR‐103b * 0.445 +
   hsa‐miR‐29c‐5p * 0.314 ‐ 0.109, hsa‐miR‐877‐5p + hsa‐miR‐29c‐5p =
   hsa‐miR‐877‐5p * 0.458 + hsa‐miR‐29c‐5p * 0.363 ‐ 1.892, hsa‐miR‐103b +
   hsa‐miR‐29c‐5p + hsa‐miR‐877‐5p = hsa‐miR‐103b * 0.432 + hsa‐miR‐877‐5p
   * 0.423 + hsa‐miR‐29c‐5p * 0.28 ‐ 1.549. It is worth noting that
   hsa‐miR‐103b plus hsa‐miR‐877‐5p exhibited an AUC of 0.854 (95% CI:
   0.842‐0.866) (Figure [83]4A). The AUC of hsa‐miR‐29c‐5p plus
   hsa‐miR‐103b and hsa‐miR‐877‐5p plus hsa‐miR‐29c‐5p is 0.840 (95% CI:
   0.828‐0.853) and 0.807 (95% CI: 0.794‐0.821) respectively
   (Figure [84]4B, and 4C). After the integration of hsa‐miR‐103b plus
   hsa‐miR‐29c‐5p plus hsa‐miR‐877‐5p, the AUC value was increased to
   0.873 (95% CI: 0.862‐0.884) (Figure [85]4D).

FIGURE 4.

   FIGURE 4
   [86]Open in a new tab

   MiRNA panel validates the performance of biomarkers for lung cancer.
   (A) hsa‐miR‐877‐5p + hsa‐miR‐103b. (B) hsa‐miR‐29c‐5p + hsa‐miR‐103b.
   (C) hsa‐miR‐877‐5p + hsa‐miR‐29c‐5p. (D)
   hsa‐miR‐103b + hsa‐miR‐29c‐5p + hsa‐miR‐877‐5p. The ROC curve of the
   microarray of [87]GSE137140 is shown in blue.

3.4. Bioinformatics analysis of miR‐103b/877‐5p/29c‐5p target genes

   For hsa‐miR‐103b, hsa‐miR‐877‐5p, and hsa‐miR‐29c‐5p, the four tools
   found 280, 650, and 26 target genes, respectively (Figure [88]S1). In
   order to better understand the functions and regulatory patterns of
   hsa‐miR‐103b, hsa‐miR‐877‐5p, and hsa‐miR‐29c‐5p at the cellular level,
   280, 650, and 26 target genes of hsa‐miR‐103b, hsa‐miR‐877‐5p, and
   hsa‐miR‐29c‐5p were annotated by network analysis and the enrichment of
   biological signal pathway was analyzed. Regarding cellular components
   (CC), target genes of hsa‐miR‐103b were usually enriched in cytoplasm,
   plasma membrane, BRCA1‐a complex and myosin complex (Figure [89]S2A).
   The results show that terms such as signal transduction, transcription,
   reproduction and cell‐cell signaling were enriched in GO biological
   process (BP) (Figure [90]S2B). Regarding GO molecular function (MF),
   these 280 targets were mainly enriched in protein serine/threonine
   kinase activity and GTP binding. (Figure [91]S2C). Furthermore, C‐MYC
   transcriptional activation and transport of organic anions were both
   enriched by biological pathway enrichment (Figure [92]S2D). Terms such
   as cytoplasm, plasma membrane, and Golgi apparatus were enriched in GO
   CC (Figure [93]S3A). The results of hsa‐miR‐877‐5p showed that GO BP
   was enriched in terms of cell communication and signal transduction
   (Figure [94]S3B). Regarding GO MF, terms such as Auxiliary transport
   activity, ATPase activity, and phosphoric diester hydrolase were
   significantly enriched (Figure [95]S3C). In the biological pathway
   enrichment analysis, terms such as JNK signaling pathway and integrins
   in angiogenesis and other pathways were enriched (Figure [96]S3D).
   Regarding GO CC, target genes of hsa‐miR‐29c‐5p were usually enriched
   in endosome, early endosome, node of Ranvier and protein transporting
   two sector ATPase complex (Figure [97]S4A). The results show that terms
   such as cell communication and signal transduction were enriched in GO
   BP (Figure [98]S4B). Regarding GO MF, these 26 targets were mainly
   enriched in transporter activity, transmembrane receptor tyrosine
   kinase activity, and T‐cell receptor activity (Figure [99]S4C).
   Transmission across chemical synapses, adenylate cyclase activating
   pathway and rapid glucocorticoid signaling pathways were enriched by
   biological pathway enrichment (Figure [100]S4D).

   The enriched transcription factors for the target genes of hsa‐miR‐103b
   were NFIC, PPARG, YY1, ARIDA, and HOXC9 (Figure [101]5A). The
   transcription factors of hsa‐miR‐29c‐5p target genes mainly include
   REST, NR1H3, NR4A2, CACD and TFAP2A (Figure [102]5B). SP4, TCF3,
   ZNF513, ZNF238 and HOXC9 were mainly enriched in hsa‐miR‐877‐5p target
   genes (Figure [103]5C).

FIGURE 5.

   FIGURE 5
   [104]Open in a new tab

   Enriched transcription factors by miRNA target genes. (A) The
   transcription factor for hsa‐miR‐103b target genes. (B) The
   transcription factor for hsa‐miR‐29c‐5p target genes. (C) The
   transcription factor for hsa‐miR‐877‐5p target genes.

3.5. MiR‐103b/877‐5p/29c‐5p directly targets the SFRP4, FOXM1 and TMEM98
3’‐UTR

   In order to investigate possible miRNA‐target gene interactions, we
   used various bioinformatics based methods described above to predict
   the target of miR‐103b/877‐5p/29c‐5p. In this study, the predicted
   binding sites of miR‐103b in SFRP4 mRNA 3’‐UTR, miR‐877‐5p in FOXM1
   mRNA 3’‐UTR and miR‐29c‐5p in TMEM98 mRNA 3’‐UTR were selected (Figure
   [105]S5). Subsequently, in order to evaluate whether
   miR‐103b/877‐5p/29c‐5p directly regulates the expression of SFRP4,
   FOXM1 and TMEM98, 3’‐UTRs carrying SFRP4, FOXM1 and TMEM98 mRNA and
   luciferase reporter plasmids (wild type and mutant) carrying
   miR‐103b/877‐5p/29c‐5p binding sites or mutant binding sites were
   cotransfected into 293T cells with miR‐103b/877‐5p/29c‐5p mimics and
   miRNA control, respectively Luciferase reporter gene analysis showed
   that 293T cells co‐transfected with miR‐103b/877‐5p/29c‐5p mimics and
   plasmids of SFRP4, FOXM1 and TMEM98‐WT‐3’‐UTR (wild‐type and mutant)
   showed significantly lower luciferase activity in wild‐type cells
   compared with mutant ones. These results indicated that SFRP4, FOXM1
   and TMEM98 were the direct targets of miR‐103b/877‐5p/29c‐5p,
   respectively (Figure [106]6).

FIGURE 6.

   FIGURE 6
   [107]Open in a new tab

   (A) Decrease activity of dual luciferase assay clarified that miR‐103b
   bind to the 3’‐UTR of SFRP4 in 239T cells while the mutant vector
   showed no changes. (B) Decrease activity of dual luciferase assay
   clarified that miR‐877‐5p bind to the 3’‐UTR of FOXM1 in 239T cells
   while the mutant vector showed no changes. (C) Decrease activity of
   dual luciferase assay clarified that miR‐29c‐5p bind to the 3’‐UTR of
   TMEM9 in 239T cells while the mutant vector showed no changes.

3.6. Expression of Exosomal Mir‐103b/877‐5p/29C‐5p

   The technology of qRT‐PCR was applied in the validation set to study
   the robustness of the sequencing data, including 17 additional LUAD
   plasma exosomes samples and 17 control samples. ΔCt was evaluated for
   the three miRNAs normalized to cel‐miR‐39. As shown in Figure [108]7,
   hsa‐miR‐103b, hsa‐miR‐877‐5p and hsa‐miR‐29c‐5p in plasma exosomes of
   LUAD patients were significantly up‐regulated compared with the control
   group (p < 0.05), which was consistent with the previous sequencing
   results. qRT‐PCR data validated the robustness of sequencing data
   analysis.

FIGURE 7.

   FIGURE 7
   [109]Open in a new tab

   Expression levels of the three miRNAs in the plasma exosome of 17 LUAD
   patients and 17 non‐tumor patients. (A) miR‐29c‐5p; (B) miR‐877‐5p; (C)
   miR‐103b. N, non‐tumor patients; T, tumor patients. **p < 0.05.

4. DISCUSSION

   NSCLC accounts for about 85% of all lung cancers, and LUAD is one of
   the main histological subtypes.[110] ^22 MiRNAs are involved in the
   occurrence and development of a variety of malignant tumors, and may be
   a treasure house for cancer diagnosis and treatment, including lung
   cancer.[111] ^23 Further study of exosomal miRNAs may help to
   understand their value in lung cancer. To our knowledge, this study was
   the first to report that hsa‐miR‐29c‐5p, hsa‐miR‐103b, and
   hsa‐miR‐877‐5p were enriched in plasma exosomes from patients with
   early lung adenocarcinoma.

   Tumor cells exchange basic information between local and distant
   recipient cells by releasing exosomes that carry proteins, lipids, and
   nucleic acids. Exosomes can be used as prospective noninvasive
   biomarkers of lung cancer, which are widely distributed in body
   fluids.[112] ^24 The lipid membrane structure of exosomes protects
   their contents from degradation by ribonuclease.[113] ^25 This enhances
   the potential of exosomal miRNAs as biomarkers for cancer diagnosis and
   treatment.[114] ^26 , [115]^27

   The AUC values of hsa‐miR‐103b, hsa‐miR‐29c‐5p, and hsa‐miR‐877‐5p in
   the [116]GSE137140 were verified to be 0.798, 0.719, and 0.770,
   respectively. Many other studies have suggested that circulating miRNAs
   can be used as biomarkers for lung cancer, and the AUC of most studies
   is about 0.75.[117] ^28 , [118]^29 The effect of multiple miRNAs
   integration is better than that of single miRNA in differentiating
   early lung cancer patients.[119] ^30 , [120]^31 Here, the AUC obtained
   by combining only two miRNAs (hsa‐miR‐877‐5p and hsa‐miR‐29c‐5p) was
   0.807, while the combined AUC of three biomarkers (hsa‐miR‐103b,
   hsa‐miR‐29c‐5p and hsa‐miR‐877‐5p) was 0.873. There is evidence that
   hsa‐miR‐29c‐5p is an important regulator of CDC42 and DNMT3A genes in
   the pathogenesis of gastric cancer.[121] ^32 Yu et al. showed that
   hsa‐miR‐877‐5p can play a role of self‐protection or adaptation in
   liver injury induced by acetaminophen.[122] ^33 In addition, most
   miRNAs are tissue specific and capable of reflecting pathological
   characteristics. In addition, most miRNAs are tissue specific and
   capable of reflecting pathological features. For example, plasma
   exosomal miRNAs (let‐7b‐3p, miR‐139‐3p, miR‐150‐3p, miR‐145‐3p) can
   distinguish patients with early colorectal cancer from healthy
   controls.[123] ^34 Lai et al. believed that exosomal miRNAs have a
   better diagnostic value for pancreatic cancer.[124] ^35 It is necessary
   to further study these exosomal miRNAs in the plasma of a large number
   of LUAD patients to prove their potential as non‐invasive biomarkers of
   LUAD.

   More and more studies have found that miRNAs play a key role in tumor
   angiogenesis, proliferation, metabolism, migration, and invasion.[125]
   ^36 So far, the function of most miRNAs has been unknown. With the
   development of high‐throughput miRNA detection techniques, more and
   more new miRNAs have been discovered. MiRNA has been widely studied in
   the occurrence and development of lung cancer.[126] ^37 Although the
   relationship between hsa‐miR‐29c‐5p, hsa‐miR‐103b, and hsa‐miR‐877‐5p
   in lung cancer has not been reported, the results suggest that they may
   play an important role in the pathogenesis of LUAD.

   Four promising miRNA target prediction tools were used, such as miRwalk
   2.0,[127] ^16 RNA 22,[128] ^18 miRanda,[129] ^17 and Targetscan
   6.2[130] ^19 to explore the target genes of hsa‐miR‐103b,
   hsa‐miR‐877‐5p, and hsa‐miR‐29c‐5p. Target genes of 280, 650, and 26
   were found in the predicted cross points of the four tools, indicating
   that they may be promising target genes of hsa‐miR‐103b,
   hsa‐miR‐877‐5p, and hsa‐miR‐29c‐5p. Some of these target genes, such as
   IL7R,[131] ^38 DDR1,[132] ^39 FZD3,[133] ^40 BCL9,[134] ^41 RAD518[135]
   ^42 etc., have been confirmed to be involved in the occurrence and
   development of lung cancer in previous studies, which also proved that
   miRNAs may play an important role in the occurrence and development of
   lung cancer. Go annotation and biosignaling pathway analysis were
   performed using 280, 650, and 26 target genes to reveal the main
   regulatory patterns of hsa‐miR‐103b, hsa‐miR‐877‐5p, and
   hsa‐miR‐29C‐5p. The results suggest a potential association between
   miRNAs and lung cancer. The down‐regulation of NFIC expression may be a
   key transcription factor in the development of LUSC by regulating genes
   involved in cell cycle and DNA replication.[136] ^43 HOXC9 promotes the
   invasion and proliferation of NSCLC cells, and miR‐495 binds to HOXC9
   3’‐UTR region to inhibit its expression. Hsa‐circ‐0020123 may play an
   antitumor role in NSCLC by regulating miR‐495/HOXC9 axis.[137] ^44 Many
   studies have shown that PPAR gamma activation can hinder the progress
   of lung cancer, and PPARG ligand can be used as a potential therapeutic
   agent for lung adenocarcinoma and PPARG can also effectively treat lung
   squamous cell carcinoma.[138] ^45 Liu's data shows that TFAP2A induced
   the high expression of KRT16 in LUAD, and KRT16 promoted the
   tumorigenicity of LUAD through EMT migration, invasion, and
   proliferation.[139] ^46 Huang’s study showed that YY1 was up‐regulated
   in lung cancer tissues and affected cell proliferation, migration and
   invasion, and suggested that YY1 played a key role in the progression
   of lung cancer partly through the activation of its downstream target
   PVT1.[140] ^47

   In addition, the robustness of miRNA‐Seq data was verified by qRT‐PCR
   analysis of three candidate miRNAs in the cohort. Therefore, the
   results indicate that hsa‐miR‐103b, hsa‐miR‐29c‐5p, and hsa‐miR‐877‐5p
   have great potential as markers of lung cancer. This provides a new way
   of thinking for finding new potential diagnostic biomarkers and also
   provides a direction for studying the mechanism of action of these
   miRNAs in LUAD. However, based on the potential interaction of target
   gene, more participants are required to participate in whether these
   miRNAs play a role in the occurrence and development of LUAD.

5. CONCLUSIONS

   In conclusion, next‐generation sequencing and bioinformatics methods
   were used to screen plasma exosomal miRNAs closely related to LUAD. Our
   work first provided a preliminary image of the differential expression
   of plasma and foreign secrete miRNAs in LUAD patients Further, dual
   luciferase assay confirmed that downstream target genes SFRP4, FOXM1
   and TMEM98 may directly bind to miR‐103b, miR‐877‐5p and miR‐29c‐5p,
   respectively. Compared with the control group, the expression level of
   miR‐103b/877‐5p/ 29C‐5p was upregulated in LUAD patients, suggesting
   that these three plasma exosomal miRNAs may be more specific,
   sensitive, and stable noninvasive diagnostic biomarkers in early LUAD
   patients.

AUTHOR CONTRIBUTIONS

   JW and ZF designed and performed the research; RW, AL, HW, XH, and ZS
   analyzed the data; JW wrote the manuscript. All authors read and
   approved the final manuscript.

CONFLICT OF INTEREST

   The authors declare that there is no conflict of interest.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

   Ethical approval was waived off. ID number: 2021‐RE‐012.

Supporting information

   Figure S1
   [141]Click here for additional data file.^ (3.2MB, tiff)

   Figure S2
   [142]Click here for additional data file.^ (438.3KB, tiff)

   Figure S3
   [143]Click here for additional data file.^ (480KB, tiff)

   Figure S4
   [144]Click here for additional data file.^ (451.4KB, tiff)

   Figure S5
   [145]Click here for additional data file.^ (503.9KB, tiff)

   Table S1–S2
   [146]Click here for additional data file.^ (48KB, doc)

ACKNOWLEDGMENTS