In this study, we aimed to investigate the potential biomarkers for early-stage lung adenocarcinoma by analyzing saliva samples. We collected saliva samples from 18 early-stage (I–IIIA) lung adenocarcinoma patients and 18 non-cancer controls, aged between 50 and 75 years, at Harbin Medical University Cancer Hospital, according to the approved protocols by the Ethics Committee of Harbin Medical University (Supplementary Table S1). In general, having oral diseases will affect the expression of miRNA in saliva (Safari et al., 2023). In order to reduce the interference, the participants who did not have oral diseases were included in the study; meanwhile, age, gender, and smoking and drinking status of the participants in the control group were made as consistent as possible with those in the experimental group. In addition, all subjects provided written informed consent. During the collection period, the participants were seated straight up and instructed to refrain from swallowing or speaking, and approximately 5 mL saliva sample was obtained from each individual.

For each set, six samples were mixed to create one biological replicate and centrifuged at 2,000 g for 5 min at 4°C to discard insoluble materials and cell debris. The samples were stored at −80°C for subsequent use.

Saliva samples were centrifuged at 2,500 g for 15 min at 4°C, followed by ultracentrifugation at 120,000 g for 2 h at 4°C to pellet exosomes. Exosome pellets were washed in filtered phosphate-buffered saline (PBS) and re-centrifuged at 120,000 g, the supernatant was removed, and the final pellet was re-suspended in 150 µL of PBS.

Fixation of exosomes was carried out with 2.5% glutaraldehyde for 2 h. Subsequently, the exosomes were purified and suspended in 100 µL serum albumin. Approximately 20 µL of exosomes was overloaded onto a formvar/carbon-coated grid and stained with 3% aqueous phosphor-tungstic acid for 60 s. Exosomes were then examined with transmission electron microscopy (TEM), which indicated their spherical shape.

The supernatants containing particles were analyzed using a NanoSight LM20 instrument equipped with a 640 nm laser (NanoSight, Amesbury, United Kingdom) at 25°C. The movement of particles was tracked using NTA software (version 2.2, NanoSight) with a low refractive index corresponding to cell-derived vesicles. The mean, mode, median particle size, and particle concentration (in millions) for each size were analyzed in each track.

Total RNA was extracted from salivary exosomes with TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and purified using two phenol–chloroform extractions followed by treatment with RQ1 DNase (Promega, Madison, WI, United States) to remove DNA. The SmartSpec Plus spectrophotometer (Bio-Rad, Hercules, CA, United States) was used to assess the quality and quantity of the purified RNA. The integrity of RNA was further verified by 1.4% agarose gel electrophoresis.

Total RNAs (3 μg) from each sample were used for small RNA cDNA library preparation using a Balancer NGS Library Preparation kit (Gnomegen, San Diego, CA, United States) according to the manufacturer’s protocol prior to directional RNA-seq library preparation. The whole library was subjected to 10% native polyacrylamide gel electrophoresis. Then, the bands corresponding to miRNA insertion were cut and eluted. Ultimately, the purified libraries were quantified using the Qubit Fluorometer (Invitrogen; Thermo Fisher Scientific, Inc.) after ethanol precipitation and washing. The RNA libraries were subjected to 150 bp pair-end sequencing on the HiSeq 2500 platform (Illumina, Inc., San Diego, CA, United States).

The FASTX-Toolkit (version 0.0.13) (http://hannonlab.cshl.edu/fastx_toolkit/) was used to process raw reads for obtaining reliable clean reads. RNAs <16 or >30 nt in length were also removed from the subsequent analysis according to the length of the adapter and mature miRNA lengths. Subsequently, the clean reads were searched against the Rfam database (version 12.0) using Bowtie (Langmead et al., 2009). Matches to transfer RNAs and ribosomal RNAs were discarded. The remaining unique sequences were aligned using Bowtie against the miRBase database (Kozomara and Griffiths-Jones, 2014), with one mismatch allowed. The aligned small RNA sequences were convinced by conserved miRNAs.

To investigate the expression profiles of identified miRNAs, the frequency of miRNA counts was normalized to transcripts per million (TPM) using the following formula: normalized expression = actual read count/total read count × 10⁶. The edgeR (v3.22) package of Bioconductor software (Robinson et al., 2010) was used to explore the DE miRNAs, and fold change (FC) ≥ 2 or ≤0.5 and p ≤ 0.05 and false discovery rates (FDRs) ≤ 0.05 indicated a statistically significant DE miRNA. The human miRNA and transcript sequences of miRBase were calculated by applying the miRanda algorithm (Krek et al., 2005). Subsequently, TargetScan was used for predicting putative miRNA targets (Lewis et al., 2005). To explore the gene function and analyze the frequency distribution of functional categories, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were employed using the R package “clusterProfiler” using the DAVID bioinformatics database (Huang et al., 2009).

Salivary exosome proteins were extracted and separated by denaturing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Subsequently, they were transferred to a polyvinylidene difluoride (PVDF) membrane and incubated with blocking buffer [5% skimmed milk powder in PBS–Tween-20 (PBST)] for 1 h. After washing with PBST, the membrane was incubated with primary antibodies (CD9 and syntenin; Abcam, MA, United States) according to the manufacturer’s instructions. Then, the PVDF membrane was incubated with the appropriate secondary antibody for 1 h followed by washing. Finally, the PVDF membrane was incubated for 3 min with SuperSignal West Pico Western HRP substrate at room temperature and imaged using a Tanon 5200 multi-imaging system (Tanon, Shanghai, China).

qPCR was carried out to detect the expression levels of miRNAs using primers designed for reverse transcription (RT) and qPCR. The primer sequences used in our study were as follows: miR-92b-5p: 5′-CTC AAC TGG TGT CGT GGA GTC GGC AAT TCA GTT GAG CAC TGC AC-3′ (RT), 5′-ACA CTC CAG CTG GGA GGG ACG GGA CGC GGT-3′ (forward), and 5′-TGG TGT CGT GGA GTC G-3′ (reverse); miR-532-3p: 5′-CTC AAC TGG TGT CGT GGA GTC GGC AAT TCA GTT GAG TGC AAG CC-3′ (RT), 5′-ACA CTC CAG CTG GGC CTC CCA CAC CCA AGG-3′ (forward), and 5′-TGG TGT CGT GGA GTC G-3′ (reverse); miR-135b-5p: 5′-CTC AAC TGG TGT CGT GGA GTC GGC AAT TCA GTT GAG TCA CAT AG-3′ (RT), 5′-ACA CTC CAG CTG GGT ATG GCT TTT CAT TCC T-3′ (forward), and 5′-TGG TGT CGT GGA GTC G-3′ (reverse); and U6: 5′-CTC GCT TCG GCA GCA CA-3′ (forward) and 5′-AAC GCT TCA CGA ATT GTG CGT-3′ (reverse). cDNAs were generated by standard procedures, and real-time PCR was performed on Bio-Rad S1000 (Bio-Rad, Hercules, CA, United States) with Bestar SYBR Green RT-PCR Master Mix (DBI Bioscience, Shanghai, China). The PCR cycling conditions were set as follows: 95°C for 10 min, 38 cycles at 95°C for 15 s, annealing, and extension at 60°C for 1 min. qPCR amplifications were performed in triplicate for each sample. Transcript levels for the genes analyzed were measured in comparison with the housekeeping gene U6 as an internal reference standard using the 2^−ΔΔCT method (Livak and Schmittgen, 2001).

The data were analyzed for statistical significance using Microsoft Excel (2012). All data are presented as the mean ± standard deviation (mean ± SD). Student’s t-test (paired) was used to determine the statistical significance when comparing the means of two datasets. A p-value < 0.05 was considered statistically significant.

In this study, a total of 567 samples, including 521 patients with lung cancer and 46 normal tissues, were obtained from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/).

The datasets obtained in the current study have been deposited in the National Center for Biotechnology Information under accession number PRJNA908267 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA908267).

Saliva samples from 18 patients with lung cancer (LC) and 18 individuals without lung cancer (NC) were collected (Figure 1A). Six samples for each group were mixed to create one biological replicate, and six samples of salivary exosomes (LC-1, LC-2, and LC-3; NC-1, NC-2, and NC-3) were obtained by a series of differential centrifugation, ultracentrifugation, and filtration steps (Figure 1B).

TEM examination indicated that exosomes from two groups showed a characteristic round- or cup-shaped morphology and dimension (Figure 1C). Nanoparticle tracking analysis (NTA) indicated that the average sizes of exosomes from LC and NC groups were 73.34 and 75.68 nm, respectively (Figure 1D). Furthermore, exosomes were validated by Western blotting. Exosomes can be characterized using exosome markers (CD9 and syntenin) expressed on their outer membrane. By analyzing the gray values of marker proteins, it was discovered that the expression level of the CD9 protein within the LC group was higher than that in the NC group. In contrast, the opposite was observed for the expression level of the syntenin protein (Figure 1E). These results indicated that salivary exosomes were isolated successfully and were suitable for further investigation.

Previous studies have indicated that exosomal miRNAs play significant roles in cancer biology and clinical applications (Sun et al., 2018a; Hu et al., 2020). The aim of this study was to investigate the miRNA cargo of salivary exosomes from two groups. Therefore, a total of six small RNA libraries (LC-1, LC-2, and LC-3; NC-1, NC-2, and NC-3) were constructed for miRNA-seq with three biological replicates for each group.

Using Illumina HiSeq 2500, >433.1 million reads were generated, corresponding to ∼73.2 million sequence reads per library; the clean reads accounted for ∼87.7%. Hierarchical clustering heatmap analyses indicated that the global expression pattern of miRNAs differed obviously between the LC and NC groups (Figures 2A, B). The clean reads were matched to Rfam and a human reference genome (Ensembl release 102 GRCh38), and the filtered reads were mapped to miRBase. The read counts were normalized to TPM, and a total of 372 miRNAs were identified (Supplementary Table S2).

DE miRNAs were determined using edgeR. The threshold for screening differentially expressed miRNAs was set as follows: a “fold change (FC) ≥ 2 or ≤ 0.5 and p ≤ 0.05 and false discovery rates (FDR) ≤ 0.05.” A total of 15 DE miRNAs were identified in LC vs. NC, including eight upregulated miRNAs and seven downregulated miRNAs, respectively (Figures 2C, D). The highest and lowest FC values were 2^5.13 and 2^−4.89, respectively (Supplementary Table S3). These findings indicated that lung cancer is associated with the expression levels of numerous miRNAs in salivary exosomes.

Three miRNAs (miR-92b-5p, miR-532-3p, and miR-135b-5p) were selected from these identified miRNAs and verified by RT-qPCR (Figure 3A). The RT-qPCR expression results were highly correlated with the RNA-seq results (Figure 3B). Hence, the RNA-seq data were reliable and suitable for further analysis.

miRNAs commonly exert their functions by binding to complementary target sites in the mRNAs of their target genes. Using the miRanda algorithm on human miRNA and transcript sequences of miRBase and TargetScan, 488 and 161 overlapped target genes were identified for the upregulated and downregulated DE miRNAs in LC vs. NC (Supplementary Table S4), respectively. This suggests that the DE miRNAs of salivary exosomes have the potential to regulate the expression levels of multiple genes.

To identify the pathways associated with the target genes of the upregulated DE miRNAs, GO enrichment analysis was conducted and 54 biological pathways were identified (p < 0.01, FDR < 0.01) (Figure 4A; Supplementary Table S5). A large number of terms were associated with cell proliferation and cycle, including “mitotic cell cycle phase transition” (GO: 0044772), “regulation of cell cycle phase transition” (GO: 1901987), “regulation of mitotic cell cycle phase transition” (GO: 1901990), “G1/S transition of mitotic cell cycle” (GO: 0000082), “negative regulation of cell cycle” (GO: 0045786), “cell cycle G1/S phase transition” (GO: 0044843), and “negative regulation of mitotic cell cycle” (GO: 0045930). These results suggested that deregulated expression of miRNAs might be associated with a disorder of cell proliferation.

KEGG enrichment analysis revealed that 20 significant functional terms were associated with the target genes of the upregulated DE miRNAs (p < 0.01, FDR < 0.01) (Figure 4B; Supplementary Table S6). Multiple enriched pathways were associated with cancer, including “colorectal cancer” (ID: hsa05210), “hepatocellular carcinoma” (ID: hsa05225), “microRNAs in cancer” (ID: hsa05206), “proteoglycans in cancer” (ID: hsa05205), “gastric cancer” (ID: hsa05226), “endometrial cancer” (ID: hsa05213), “chronic myeloid leukemia” (ID: hsa05220), “chemical carcinogenesis-receptor activation” (ID: hsa05207), “prostate cancer” (ID: hsa05215), “breast cancer” (ID: hsa05224), and “bladder cancer” (ID: hsa05219).

Moreover, GO and KEGG enrichment analyses were carried out for downregulated DE miRNAs (Figures 4C, D; Supplementary Tables S7, S8), resulting in 64 and 16 terms, respectively (p < 0.01, FDR < 0.05). It was found that multiple enriched GO terms were associated with apoptosis of cell, including “intrinsic apoptotic signaling pathway” (GO: 0097193), “negative regulation of apoptotic signaling pathway” (GO: 2001234), “neuron apoptotic process” (GO: 0051402), “regulation of apoptotic signaling pathway” (GO: 2001233), “intrinsic apoptotic signaling pathway in response to DNA damage by p53 class mediator” (GO: 0042771), and “cellular response to abiotic stimulus” (GO: 0071214), and several KEGG terms were involved with cancer, including “viral carcinogenesis” (ID: hsa05203), “prostate cancer” (ID: hsa05215), and “glioma” (ID: hsa05214).

These results suggested that deregulated expression of miRNAs might be involved in the occurrence and development of cancer.