Exosomal miRNAs as circulating biomarkers for prediction of development of haematogenous metastasis after surgery for stage II/III gastric cancer

Abstract Exosomes secreted by living cancer cells can regulate metastasis. Exosomal miRNAs can reflect pathological conditions of the original cancer cells. Therefore, we aim to identify exosomal miRNAs as circulating biomarkers for haematogenous metastasis of gastric cancer. Pre‐treatment serum samples of eighty‐nine patients with stage II/III gastric cancer were collected. Thirty‐four of them developed haematogenous metastasis after surgery and the other fifty‐five did not. Extraction of exosomes was validated by western blot, transmission electron microscopy and nanoparticle tracking analysis. MiRNA qPCR array was performed in three matched pairs of samples. Internal control was selected from PCR array and validated in the remaining samples. Expressions of exosomal miRNAs were evaluated in the remaining samples by RT‐qPCR, as well as in gastric cancer tissue samples and cell culture medium. Expression levels of exosomal miRNAs were analysed with clinical characteristics. The results indicated thirteen up‐regulated and six down‐regulated miRNAs were found after normalization. MiR‐379‐5p and miR‐410‐3p were significantly up‐regulated in metastatic patients (P < .01). Higher expression of exosomal miR‐379‐5p or miR‐410‐3p showed shorter progression‐free survival of the patients (P < .05). It was also found that miR‐379‐5p and miR‐410‐3p were down‐regulated in gastric cancer tissue samples, while they were significantly up‐regulated in gastric cancer cell culture medium compared with cancer cells. In conclusion, exosomal miRNAs are promising circulating biomarkers for prediction of development of haematogenous metastasis after surgery for stage II/III gastric cancer.

advanced cancer may develop distant metastasis in the sequent 5 years.
Currently, most of these metastatic patients are diagnosed on clinical examination and/or diagnostic images. Therapy for such patients is limited, making the prognosis of patients with metastatic recurrence very poor. 2 It is therefore important to identify biomarkers which could predict the subsequent development of metastatic disease. The ability to do so may help select patients at high risk of developing metastasis for aggressive neoadjuvant therapy before surgery or adjuvant therapy after surgery. In addition, such patients should be followed more closely after surgery.
Studies have indicated that metastasis is an early event in cancer development. Primary tumours create a favourable microenvironment in secondary organs and tissue sites for subsequent metastases. 3 This is called the "seed" (the pre-metastatic niche) and "soil" (secondary sites) theory. To transfer the "seed" to its appropriate "soil", primary tumours secret extracellular vesicles. 4 Exosomes are an important type of these vesicles. Intriguingly, exosomes released from primary cancer cells have a distinct genetic and epigenetic makeup, which allows them to undertake their tumorigenic function. These unique cancer-specific fingerprints present in exosomes can be detected in circulation. 5 Therefore, molecules in exosomes are promising circulating biomarkers for predicting metastasis of cancer, even before the metastasis could be detected clinically.
Recently, attention has been paid to exosomes as functional transporters in various pathological conditions including cancer. [6][7][8] Importantly, molecules contained in exosomes can be transferred into cells of secondary sites, where these molecules affect gene expressions of the target cells, further to alter cell functions. 9 Hence, exosomes play significant roles in intercellular communication to promote cancer progression. 10 Exosomes range from 30 to 150 nm in diameter. They have been found in many body fluids including blood, saliva, urine, amniotic fluid and bile. 11 Besides, deep-sequencing experiments have shown that exosomes contain a very diverse RNA cargo. 12 Most of the RNA cargo is 20-200 nucleotides in length, including full-length molecules such as miRNAs. 13,14 Exosomes are involved in the pathogenesis of cancer. 15 For gastric cancer, studies indicated that exosomes were released by gastric cancer cells to perform certain functions in extracellular environment. For instance, it has been reported that oncogenic let-7 miRNA family is abundant in the culture medium of a metastatic gastric cancer cell line AZ-P7a. 16 Another study suggested that exosomes derived from culture medium of gastric cancer cells stimulated the activation of NF-κB pathway in macrophages to promote cancer progression. 17 Exosomes extracted from gastric cancer cell culture medium promote the expression of adhesion molecules in mesothelial cells, which might play a critical role in the development of peritoneal metastasis of gastric cancer. 18 Three studies used exosomal molecules as potential biomarkers for the pathological status of gastric cancer. [19][20][21] However, these exosomal molecules were of large size and none of these studies concerned the detection of gastric cancer metastasis.
Studies have shown that dysregulation of miRNAs was involved in initiation and progression of various malignancies, including gastric cancer. 22,23 MiRNAs belong to a group of small non-coding RNAs around 18-25 nucleotides in length. They bind to complementary sequences in the 3′-untranslated regions (3′-UTR) of target mRNAs to induce degradation or translational repression. 24,25 MiRNAs are tissue specific and even cells specific within those tissues. They are potentially useful for diagnosis, predicting clinical outcome or acting as therapeutic targets in patients with cancer. The unique pattern of miRNAs in gastric cancer provides the possibility of applying miR-NAs as biomarkers for monitoring gastric cancer.
MiRNAs are suitable for circulating biomarkers, as (a) the relatively smaller number of candidate sequences and thus relative ease of analysis and (b) stability due to small size (18-25 nt) and hence robustness of detection. In addition, miRNAs from whole serum or plasma may include endogenous cellular miRNAs derived from apoptotic or necrotic cells. Serum/plasma miRNAs, which are not associated with vesicles, were reported to show differential stability to treatment by RNase. Hence, the unique nature of the exosomal bilayer allows miRNA to be protected from degradation, making it an ideal source for biomarkers.
In this study, we compared the expression profiles of exosomal miRNAs in the preoperative serum of the patients who developed haematogenous metastasis to those without metastasis in at least 5-year follow-up period. Dysregulated serum exosomal miRNAs from profiling were evaluated in another set of serum samples.
Expressions of exosomal miRNAs were studied for the associated clinical characteristics and progression-free survival. Expressions were also evaluated in gastric cancer tissue samples and cell line culture medium. This is the first study to identify novel circulating exosomal miR-NAs for prediction of development of haematogenous metastasis of gastric cancer. Serum exosomal miRNAs may act as circulating biomarkers for evaluation of metastatic status of patients with gastric cancer. It will be helpful for development of cost-effective and mini-invasive circulating biomarkers for monitoring gastric cancer.

| Human samples
The collection and storage of samples had been approved by the Ethics Committee of The University of Hong Kong. This project was performed in accordance with relevant guidelines/regulations. All of the samples were obtained with participants' informed consent.
Human gastric cancer serum samples and paired tumour/adjacent non-tumour tissue samples were collected from the patients with gastric cancer in Queen Mary Hospital, Hong Kong. None of the patients received preoperative treatment. All samples were immediately processed and stored at −70°C. Blood will be centrifuged at 1600 g for 10 minutes at 4°C, and serum will be transferred to new tubes followed by further centrifugation at 16 000 g for 10 minutes at 4°C. The supernatant will be transferred to new tubes. From this process, only the serum but not the blood cells, dead cells or cell debris will be collected for following study.

| Gastric cancer cell lines
Human gastric cancer cell lines AGS and SNU1 (ATCC) were used in this study. Both cell lines were established from gastric cancer tissue samples of original site (stomach). Cells were cultivated in RPMI1640 medium (Gibco BRL) supplemented with 10% exosome-depleted foetal bovine serum (FBS; SBI, System Biosciences). All cells were incubated at 37°C in a humidified incubator which contains 5% CO2.

| Extraction of exosomes and exosomal miRNAs
Exosomes were extracted from serum samples using the ExoQuick Exosome Isolation Kit (SBI, System Biosciences). 500 µL serum samples were incubated with 120 µL ExoQuick™ for 60 minutes at So the measurements among different groups were comparable.

| Western blot
In this study, CD9 and HSP70 were markers for validation of exosomes. Western blot was performed as previous described. 26 Briefly, exosomes were extracted and lysed by RIPA Buffer (Sigma Chemical Co.). The concentration of proteins in each sample was evaluated by BCA Protein Assay. This assay produces a linear response curve, which allows accurate determination of unknown protein concentrations and provides a high dynamic range. With concentration of each sample, equal amount of protein was used for Western blot. Samples containing equal amount of protein were separated by SDS-PAGE and electro blotted onto Immobilon-P Transfer Membrane (Applied Biosystems). The membrane was blocked with 5% no-fat milk, followed by incubation with antibodies specific for anti-CD9 and anti-HSP70 (1:1000; SBI) and anti-β-actin (1:20 000; Cell Signaling Technology), respectively. Blots were then incubated with anti-rabbit or antimouse secondary antibody conjugated to horseradish peroxidase (Amersham Pharmacia) accordingly. The signals were captured by FUJI Medical X-Ray Film and developed by the FUJI system.

| Transmission electronic microscopy
Exosomes were visualized using transmission electron microscopy (TEM) to validate their morphology and general size, according to the method described before. 27 Generally, exosome suspension was fixed in 2% paraformaldehyde. A small amount of this mix was transferred onto each of 2 formvar-carbon-coated electron microscopy grids. PBS was placed on a sheet of parafilm and grids transferred with the sample membrane side facing down using clean forceps. The grids were kept wet on the side of the membrane during all steps, but dry on the opposite side. The grids were transferred to 1% glutaraldehyde before transferring to distilled water.
This was repeated several times. To contrast the samples, grids were transferred to uranyl-oxalate solution pH7, before transferring to methyl-cellulose-UA, placing the grids on a glass dish covered with parafilm on ice. The grids were removed with stainless steel loops and excess fluid blotted gently on Whatman no.1 filter paper. Grids were left to dry and stored in appropriate grid storage boxes. Grids were observed with Thermo Fisher Tecnai transmission electron microscope.

| Nanoparticle tracking analysis
Nanoparticle tracking analysis (NTA) was performed to quantify exosomes in serum samples. 28 The number and size of exosomes were directly tracked using the NanoSight NS 300 system (NanoSight Technology). Exosomes were resuspended in PBS at a concentration of 5 μg/mL, then diluted 100-to 500-fold, to achieve between 20 and 200 objects per frame. Samples were manually injected into the sample chamber at ambient temperature. Each sample was configured with a 488 nm laser and a highsensitivity scientific complementary metal-oxide-semiconductor (sCMOS) camera and was measured in triplicate at camera setting 13 with an acquisition time of 30 seconds and a detection threshold setting of 7. At least 100 completed tracks will be analysed per video. Finally, data were analysed using the NTA analytical software (version 2.3).

| miRCURY LNA™ miRNA miRNome PCR Array
MiRCURY LNA™ miRNA miRNome PCR panels were applied for miRNA profiling of exosomal miRNAs from three pairs of matched metastatic and non-metastatic serum samples. The panels con-

| Reverse transcription-quantitative polymerase chain reaction
Total exosomal miRNAs were reverse transcribed to cDNA using miRCURY LNA™ RT Kit (Exiqon) according to the manufacturer's instructions. Quantitative PCR was performed using miRCURY LNA™ SYBR Green Mix (Exiqon) in Vii7A real-time PCR system (Applied Biosystems). The miRNA-specific primer sequences were provided by Exiqon based on the miRNA sequences obtained from the miR-Base database. At the end of the PCR cycles, melting curve analyses were performed. Fold changes in expression of each miRNA were calculated by a comparative threshold cycle (C t ) method using the Currently, there is no standard internal control for exosomal miRNAs. MiRNAs with relatively abundant and consistent expressions released from profiling were applied as candidate internal controls.

| Statistics
Statistical analysis was carried out using Statistical Package for Social Sciences (SPSS) 24.0 for Windows (SPSS Inc). Mann-Whitney test was applied for analysis of expression differences of exosomal miRNAs between two groups. Log-rank test was applied for progression-free survival analysis. Chi-square test or Fisher's exact test was used to analyse clinic-pathological parameters. Wilcoxon's sign-rank test was applied for analysis of expression differences in paired tissue samples. Student's t test was used to analyse the results expressed as mean ± SD. Differences were considered significant when P ≤ .05.

| Patient recruitment and study design
In our cohort of patients, none of the patients in stage I developed metastasis in 5-year follow-up after operation. Patients in stage IV already had distant metastasis when they were diagnosed. Therefore, patients in stage II and stage III were included in this study. Among the patients in stage II/III, Chinese patients were included to eliminate racial differences. Only the patients with pre-treatment serum samples were included. Finally, 89 patients were included. All these patients had been followed up for at least 5 years (or until the date of death for metastatic patients). Definition of haematogenous metastasis in this study included distant metastasis to liver, lung, bone, brain, or skin and soft tissue (excluding peritoneum). The diagnostic criteria of haematogenous metastasis were stated in Table S1.
Patients who developed metastasis within 5 years after operation were categorized as metastatic patients, while patients without metastasis during 5-year follow-up were categorized as non-metastatic ones. Among these patients, 34 of them developed haematogenous metastasis within 5 years after operation ( Figure S1).
The characteristics of these patients, including age, gender, comorbidity, presenting symptoms and signs, operative findings and staging had been collected into our standard electronic database.
These serum samples were subjected to miRNA profiling and RT-qPCR. Analysis of clinical characteristics of the patients with metastasis and without metastasis was indicated in Table 1. The results suggested there was no significant difference between these two groups in gender and age. The difference in stages among II, IIIA and IIIB was consistent with higher risk of metastasis in more advanced stages. The samples were then applied for following study.

| Validation of extraction of exosomes
To confirm the extracellular vesicles were mostly exosomes,  Metastasis on follow-up (N = 34) vs no metastasis (N = 55).

| Exosomal miRNAs profiling in three pairs of serum samples
To identify dysregulated exosomal miRNAs in metastatic patients, And UniSp3 (triplicated) was applied for inter-plate calibrator.
Expressions of 752 miRNAs were evaluated in each sample (N = 6, 3 matched pairs; Table S2). The expression of every miRNA was presented as a C t value. If the C t value was larger than 40 (C t > 40), it was

| Normalization of miRNA expressions
To clarify whether the dysregulated miRNAs were up-regulated or down-regulated, internal control was needed to normalize the expressions. Currently, there is no standard internal control for exosomal miRNAs. From the qPCR array, we found C t values of miR-16-5p, miR-93-5p, miR-486-3p and SNORD38B were more consistent among the six samples. So we chose these four small  (Table S3).
Thirteen up-regulated and six down-regulated miRNAs were found to be overlapped in both normalization analysis Table 3.

| Sensitivity, specificity and clinical characteristics
With the above optimal cut-off points, expression values larger than cut-off points were categorized as positive, while expression values smaller than cut-off points were categorized as negative. Sensitivity and specificity of exosomal miR-379-5p and miR-410-3p were indicated in Figure 5. and miR-410-3p were associated with gastric cancer progression. Association of expressions of these two miRNAs and distant metastatic sites was summarized in Table 4. It indicated that higher expressions of exosomal miR-379-5p were associated with distant metastasis, especially to lung and bone, whereas higher expressions of exosomal miR-410-3p were associated with distant metastasis especially to liver and lung.

| Expressions of miRNAs in gastric cancer tissue samples and cell culture medium
To investigate the mechanism of higher expressions of exosomal miR-379-5p and miR-410-3p, the expressions of these two miRNAs    suppressor in these sites. It revealed that tumour metastasis is a complicated and comprehensive process. Other miRNAs and molecules (DNAs, long non-coding RNAs, proteins, etc) were also involved in development of metastasis.
It has been reported exosomal miRNAs may be applied as circulating biomarkers for gastric cancer. For example, a study suggested that serum exosomal miR-19b-3p and miR-106a-5 could be potential biomarkers for detection of gastric cancer. 41 Another study investigated six serum miRNAs as diagnostic biomarkers for gastric cancer and further indicated four serum exosomal miR-10b-5p, miR-195-5p, miR-20a-3p and miR-296-5p were significantly increased in patients with gastric cancer. 42 Additionally, exosomal miRNA profiling in peritoneum lavage fluid showed that exosomal miR-21 and miR-1225-5p might serve as prognostic biomarkers for peritoneal metastasis after curative gastric cancer resection. 43 In this study, we focused on exosomal miRNAs as circulating biomarkers for prediction of development of haematogenous metastasis in stage II/III gastric cancer. We found exosomal miRNAs had been dysregulated in haematogenous metastatic gastric cancer patients, even before those patients presented with clinical signs and/ or abnormal imaging. Exosomal miR-379-5p and miR-410-3p were significantly up-regulated in the serum of gastric cancer patients who would develop haematogenous metastasis. Higher exosomal miR-379-5p or miR-410-3p was associated with worse progression-free survival of the patients. This will be helpful for a selection of patients for aggressive neoadjuvant therapy before surgery, aggressive adjuvant therapy after surgery or closer follow-up after surgery. In conclusion, serum exosomal miR-379-5p and miR-410-3p may be applied as circulating biomarkers for prediction of development of haematogenous metastasis after surgery for patients with stage II/III gastric cancer.

ACK N OWLED G EM ENT
This project was funded by The University of Hong Kong and General Fund for Gastric Cancer.

CO N FLI C T S O F I NTE R E S T S
The authors confirm that there are no conflicts of interest.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.