Circulating pancreatic cancer exosomal RNAs for detection of pancreatic cancer

Diagnostic biomarkers for the early diagnosis of pancreatic cancer are needed to improve prognosis for this disease. The aim of this study was to investigate differences in the expression of four messenger RNAs (mRNAs: CCDC88A,ARF6, Vav3, and WASF2) and five small nucleolar RNAs (snoRNAs: SNORA14B,SNORA18,SNORA25,SNORA74A, and SNORD22) in serum of patients with pancreatic cancer and control participants for use in the diagnosis of pancreatic cancer. Results were compared with the expression of sialylated Lewis (a) blood group antigen CA19‐9, the standard clinical tumor biomarker. Reverse transcription quantitative real‐time PCR showed that all of the mRNAs and snoRNAs, except CCDC88A, were encapsulated in exosomes and secreted from cultured pancreatic cancer cells, and present in cell culture medium. In a discovery‐stage clinical study involving 27 pancreatic cancer patients and 13 controls, the area under the receiver operating characteristic curve (AUC) of two mRNAs (WASF2 and ARF6) and two snoRNAs (SNORA74A and SNORA25) was > 0.9 for distinguishing pancreatic cancer patients from controls; the AUC of CA19‐9 was 0.897. Comparing serum levels of WASF2,ARF6,SNORA74A,SNORA25, and CA19‐9 revealed that levels of WASF2 were the most highly correlated with the risk of pancreatic cancer. The AUCs of WASF2,ARF6,SNORA74A, and SNORA25 in serum from patients in the early stages of pancreatic cancer (stages 0, I, and IIA) were > 0.9, compared with an AUC of 0.93 for the level of CA19‐9. The results of this study suggest that WASF2,ARF6,SNORA74A, and SNORA25 may be useful tools for the early detection of pancreatic cancer. Monitoring serum levels of WASF2 mRNA may be particularly useful, as it was the most highly correlated with pancreatic cancer risk.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) is the fourth leading cause of cancer-related mortality in the Western world (Siegel et al., 2013). However, no biomarkers that identify patients with PDAC in the early stages are available (Locker et al., 2006), and the absence of reliable serum biomarkers for PDAC reduces the potential effectiveness of screening strategies in at-risk populations, such as patients with diabetes mellitus and chronic pancreatitis. Therefore, better diagnostic markers of PDAC could improve the early diagnosis of this disease and enable more patients to undergo curative surgical resection.
Exosomes are extracellular vesicles (40-150 nm in size) secreted from all cell types (Thery et al., 2002). Endosomes formed during the inward budding of late endosomes can develop into intracellular vesicular endosomes containing messenger RNA (mRNA), microRNA, DNA, long noncoding RNA (lncRNA), RNA-binding proteins, and lipids (Mathivanan and Simpson, 2009). Recent studies have demonstrated that RNAs, including mRNAs, microRNAs, and lncRNAs, are secreted from tumor cancer cells into body fluids such as blood, urine, milk, and saliva via exosomes (Laurent et al., 2015). Exosome concentrations are increased in the systemic circulation of PDAC patients (Nuzhat et al., 2017). Recent publications have revealed that exosomes can play diagnostic and prognostic roles in a variety of cancers (Liu et al., 2016;Rodr ıguez et al., 2015).
Small nucleolar RNAs (snoRNAs) are noncoding regulatory RNAs of approximately 60-300 nucleotides localized primarily in the nucleolus, where they function in pre-ribosomal RNA modification and processing (Lafontaine, 2015). Two types of snoRNAs have been described: C/D-box and H/ACA-box (Falaleeva and Stamm, 2013). As snoRNAs exhibit tissue-specific expression (Castle et al., 2010), they could serve as novel cancer biomarkers (Su, 2016). The SNORD50A/ SNORD50B locus is deleted in 10-40% of PDAC patients, and the loss of this locus is associated with shorter survival time (Siprashvili et al., 2016). Levels of SNORD33, SNORD66, and SNORD76 are significantly elevated in the serum of lung cancer patients compared with cancer-free controls, suggesting they could serve as biomarkers for the early detection of this disease (Liao et al., 2010). SNORA42 is frequently overexpressed in lung cancer and colorectal cancer, and knockdown of SNORA42 slows the growth of cancer cells, indicating that SNORA42 is a putative oncogene (Mei et al., 2012).
The aim of the present study was to assess the utility of several serum mRNAs (CCDC88A, ARF6, Vav3, and WASF2) and snoRNAs (SNORA14B, SNORA18, SNORA25, SNORA74A, and SNORD22) as diagnostic markers for differentiating PDAC patients from control patients without pancreatic disease. We also describe the differential expression of these mRNAs and snoRNAs in serum samples from patients with early-stage (0, I, and IIA) and late-stage (IIB, III, and IV) PDAC.

Cell culture
The S2-013 human PDAC cell line, a subline of SUIT-2, and HPNE immortalized normal pancreatic epithelial cells were cultured as previously described (Taniuchi et al., 2011).

Preparation of conditioned medium
S2-013 and HPNE cells grown to 70% confluence were incubated for 48 h in the presence of conditioned medium supplemented with 10% exosome-free FBS (Exo-FBSHI; System Biosciences, Palo Alto, CA, USA). Culture conditioned medium was collected and centrifuged using a Beckman Coulter Allegra X-15R centrifuge (Brea, CA, USA) at 300 g and 4°C for 10 min, and the supernatant was collected as the conditioned medium.
2.5. Isolation of exosomal RNAs from cell lysates and conditioned medium S2-013 and HPNE cells were lysed in lysis buffer [50 mM Tris (pH 7.4), 150 mM NaCl, 1 mM MgCl 2 , 0.5% NP-40, a protease inhibitor cocktail tablet (Roche Applied Science, Penzberg, Germany), and phosphatase inhibitor cocktail (Nacalai, Kyoto, Japan)]. The exosomes in the cell lysates were precipitated using an ExoCap Exosome Composite kit (JSR Life Sciences, Tokyo, Japan) according to the manufacturer's recommendations. Exosomes in conditioned medium harvested from S2-013 and HPNE cells were precipitated using ExoQuick-TC exosome precipitation solution (System Biosciences) according to the manufacturer's recommendations. RNA was extracted from exosomes isolated from either cell lysates or cell culture medium using a Plasma/Serum RNA Purification Mini kit (Norgen BIOTEK, Thorold, ON, Canada) according to the manufacturer's instructions. The exosomal RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Scientific, Fremont, CA, USA).

Serum samples for reverse transcription quantitative real-time PCR
Serum samples from patients undergoing resection for PDAC were prospectively obtained at the Department of Surgery of Kochi Health Sciences Center between April 2015 and March 2016. Serum samples from PDAC patients were selected according to the following criteria: (a) Patients were newly diagnosed and previously untreated, (b) tumors were pathologically diagnosed as PDAC, and (c) PDAC patients were not suffering from any kind of malignancies, including PDAC. Tumors were classified (stages I-IV) according to the system of the International Union against Cancer (Table 1; Sobin et al., 2009). Clinicopathologic parameters were classified according to pancreatic carcinoma criteria of the Japan Pancreas Society (2003). Control serum samples from individuals diagnosed with benign gastrointestinal diseases who were being evaluated for nonpancreatic diseases were prospectively obtained at the Department of Gastroenterology and Hepatology of Kochi Medical School Hospital. This study was approved by the ethical review boards of Kochi Medical School (approval number: ERB-101894) and Kochi Health Sciences Center (approval number: 151002) regarding patient recruitment. The study was carried out in accordance with the approved guidelines. Informed consent was obtained from each patient. The experiments were undertaken with the understanding and written informed consent of each subject. The study methodologies conformed to the standards set by the Declaration of Helsinki. Serum was obtained at the time of diagnosis and stored at À80°C.

Isolation of exosomal RNAs from serum samples
Exosomal RNA was extracted from all serum samples using a Plasma/Serum RNA Purification Mini kit (Norgen BIOTEK) according to the manufacturer's recommendations. The RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Scientific).

One-step SYBR Green I real-time RT-PCR assay
Total RNA from S2-013 was extracted using an RNeasy kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions, and the RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Scientific). One-step SYBR Green I realtime RT-PCR assay (SYBR Green I assay) was performed using a One-Step TB Green PrimeScript RT-PCR Kit II (Takara BIO, Shiga, Japan) in a 10-lL reaction consisting of 1 lL exosomal RNA or total RNA from S2-013, 0.4 lL each of 10 lM forward and reverse primers, 0.4 lL of PrimeScript One Step Enzyme Mix 2, 0.2 lL of 509 ROX Reference Dye, 5 lL of 29 One Step SYBR RT-PCR Buffer 4, and 2.6 lL of RNase-free H 2 O. The primers used for the SYBR Green I assay are summarized in Table 2. For the negative controls, exosomal RNA was substituted with actin beta (ACTB) and hypoxanthine phosphoribosyltransferase 1 (HPRT1). Forty cycles of amplification were performed using a thermal cycling profile of 94°C for 15 s, 58°C for 15 s, and 72°C for 1 min. Subsequently, a melting curve was recorded by holding at 95°C for 15 s, cooling to 60°C for 1 min, and then heating at 0.1°CÁs À1 to 95°C. The amplification and melting curve data were collected and analyzed using STEPONE software v2.2 (Applied Biosystems, Foster City, CA, USA).

Statistical analysis
For in vitro experiments, STATFLEX software (ver. 6; YUMIT, Osaka, Japan) and SAS software (ver. 9.1.3; SAS Institute, Cary, NC, USA) were used for statistical analyses. Student's t-test was used for comparing continuous variables. P-values < 0.05 were considered significant, and all tests were two-tailed.
In the discovery-stage clinical study, statistical analyses were performed using R software (ver. 3.3.3; The R Foundation, Vienna, Austria). The pROC package (The R Foundation) was used for receiver operating characteristic (ROC) curve analysis. The Wilcoxon rank-sum test was used to compare differences in serum levels of the snoRNAs, mRNAs, and CA19-9 between the PDAC and control groups. The predictive performance of the snoRNAs and mRNAs was evaluated using ROC analysis, area under the curve (AUC), and corresponding 95% confidence intervals (CIs; DeLong et al., 1988) and compared with CA19-9. Multivariate logistic regression was used to establish the diagnostic mathematical model, and Pearson's correlation coefficients were used to analyze the multicollinearity between the mRNAs and snoRNAs and CA19-9.

Intracellular distribution of exosomal RNAs in cultured PDAC cells
We used RNA fluorescence in situ hybridization to determine the subcellular localization of mRNAs for CCDC88A, ARF6, Vav3, and WASF2 and snoRNAs for SNORA14B, SNORA22, SNORA25, SNORA74A, and SNORD22 in moderately differentiated S2-013 PDAC cells. All of these mRNAs and snoRNAs were concentrated in S2-013 cells positive for the cytoplasmic exosome marker CD63 (Fig. 1A).
We validated the expression of these mRNAs and snoRNAs in the intracellular exosomes of S2-013 cells. Intracellular CD63-positive exosomes were isolated from S2-013 cell lysates using an ExoCap Exosome composite kit. Intracellular exosomal RNAs were purified from the isolated CD63-positive exosomes, and a SYBR Green I assay showed that all of these mRNAs and snoRNAs were significantly increased in CD63positive exosomes of S2-013 cells compared to reference RNAs for ACTB and HPRT1 (Fig. 1B).

Extracellular localization of exosomal RNAs from cultured PDAC cells
We utilized a SYBR Green I assay to examine the presence of mRNAs for CCDC88A, ARF6, Vav3, and WASF2 and snoRNAs for SNORA14B, SNORA18, SNORA25, SNORA74A, and SNORD22 in CD63- Table 2. Primer sequences for the SYBR Green I assay.

Gene
Forward primer sequence positive exosomes isolated from the culture medium of S2-013 and HPNE cells. Western blotting showed the presence of CD63 in the total lysate and culture medium of both S2-013 and HPNE cells ( Fig. 2A). The expression level of CD63 was higher in the culture medium of S2-013 cells compared with the medium of HPNE cells ( Fig. 2A). The SYBR Green I assay revealed the presence of all of the mRNAs and snoRNAs except CCDC88A in CD63-positive exosomes in S2-013 cell culture medium (Fig. 2B). We found that levels of these RNAs in the culture medium of HPNE cells were lower than in the culture medium of S2-013 cells (Fig. 2C). These results indicate that compared with HPNE cells, the expression of these RNAs was upregulated in exosomes secreted from S2-013 cells.

Serum levels of exosomal RNAs
The prospective clinical study included a total of 40 serological samples from patients with PDAC and control patients without pancreatic disease. The clinical characteristics of PDAC at the time of specimen procurement are summarized in Table 1. A SYBR Green I assay was used to compare the expression of four mRNAs (CCDC88A, ARF6, Vav3, and WASF2) and five snoRNAs (SNORA14B, SNORA18, SNORA25, SNORA74A, and SNORD22) in serum samples from patients with PDAC and disease-free controls. The expression of all of the snoRNAs and mRNAs examined was significantly increased in pairwise comparisons between PDAC and control patients (P < 0.05; Fig. 3).

ROC curve analyses
Receiver operating characteristic curve analyses showed that two mRNAs (WASF2 and ARF6) and two snoRNAs (SNORA74A and SNORA25) in serum provided excellent accuracy (AUC > 0.9) for distinguishing PDAC patients from controls (Fig. 4). The accuracy of each exosomal RNA in distinguishing PDAC from control patients is summarized by the AUC of ROC curves (Table 3). The AUC of CA19-9 was 0.897 (95% CI, 0.797-0.997; Table 3).
The relationships between the serum concentrations of these mRNAs (WASF2 and ARF6) and snoRNAs (SNORA74A and SNORA25) and various clinicopathologic features were analyzed using the Wilcoxon rank-sum test (Tables 4-7). There were no significant associations between serum exosomal RNA concentrations and the clinical characteristics of age, gender, tumor size, or clinical stage in PDAC patients.

Association between RNA levels and risk of PDAC
The significance of differences in the serum levels of two mRNAs (WASF2 and ARF6) and two snoRNAs (SNORA74A and SNORA25) in terms of PDAC diagnosis was evaluated using logistic regression to obtain crude odds ratios (ORs; Table 8). Crude ORs adjusted for CA19-9 were then adjusted to exclude possible effects of age and gender. The results suggested that the serum level of WASF2 mRNA was the most highly correlated with the risk of PDAC.
3.6. Detection of early-stage PDAC using exosomal RNAs and CA19-9 The AUC values of ROC curves for five snoRNAs, four mRNAs, and CA19-9 are summarized in Table 9.  = 13), as determined using a SYBR Green I assay. ROC curves of levels of the exosomal RNAs and CA19-9 in serum of PDAC patients and controls; X-axis, 1 À specificity; Y-axis, sensitivity.
Two mRNAs (WASF2 and ARF6) and two snoRNAs (SNORA74A and SNORA25) for distinguishing patients with stage 0/I/IIA and stage IIB/III/IV PDAC from disease-free controls had values > 0.9 (Table 9). The AUC values of ROC curves for CA19-9 for distinguishing patients with stage 0/I/IIA and stage IIB/III/IV PDAC from controls were 0.933 and 0.897, respectively (Table 9, Fig. 5A). The levels of WASF2, ARF6, SNORA74A, SNORA25, and CA19-9 in serum samples from patients in the early stages of PDAC (stages 0, I, and IIA) were significantly higher than the level in controls (Fig. 5B). Interestingly, the AUC values of ROC curves for SNORA22 and SNORD22 in patients with stage 0/I/IIA PDAC were higher than those in patients with stage IIB/III/IV PDAC.

Discussion
To decrease the mortality caused by PDAC, efficient screening methods capable of detecting the disease in the early stages are needed (Marengo and Robotti, 2014). This study describes a novel panel of serum exosomal RNAs, including four mRNAs and five snoRNAs, for use in diagnosing PDAC. We demonstrated that the measurement of these serum exosomal RNAs enables detection of the early stages of PDAC. The AUC value of the ROC curve for levels of the current standard marker, CA19-9, in serum of patients in the early stages of PDAC (stages 0, I, and IIA) was 0.93 (Table 3), but the use of this marker is limited to monitoring the response to therapy rather than diagnosing the early stages of PDAC (DiMagno et al., 1999).
We recently reported that an RNA-binding protein, insulin-like growth factor-2 mRNA-binding protein 3 (IGF2BP3), and IGF2BP3-bound RNAs localize in cytoplasmic RNA granules that accumulate in the membrane protrusions of PDAC cells (Taniuchi et al., 2014a,b). Local translation of IGF2BP3-bound mRNAs induces formation of the protrusions, thereby promoting the motility, invasion, and metastasis of cancer cells (Taniuchi et al., 2014a,b). The mRNAs for  CCDC88A, ARF6, Vav3, and WASF2 bind to IGF2BP3 (Taniuchi et al., 2014a), and CCDC88A, ARF6, and Vav3 proteins accumulated in membrane protrusions promote the motility and invasion of PDAC cells (Taniuchi et al., 2014a;Tanouchi et al., 2016;Tsuboi et al., 2016). snoRNAs for SNORA14B, SNORA18, SNORA25, SNORA74A, and SNORD22 also bind IGF2BP3 in PDAC cells (Taniuchi et al., 2014a); however, the mechanisms by which IGF2BP3bound mRNAs and snoRNAs are transferred from cytoplasmic RNA granules to intracellular exosomes are unknown. The mechanism underlying the packaging process and its importance is thus an important subject for future studies. Small nucleolar RNAs are precursors for functional small RNAs, and snoRNA-derived small RNAs can function like microRNAs (Ender et al., 2008). The snoRNA ACA45 is processed into 20-25 small nucleotide RNAs, and ACA45-processing RNAs bound to Ago proteins contribute to posttranscriptional gene silencing of target mRNAs (Ender et al., 2008). The U/A-rich SNORD50A inhibits 3 0 -processing of its target mRNAs by blocking interactions with the Fip1poly(A) site, resulting in changes in alternative polyadenylation profiles and/or posttranslational regulation of subsets of target mRNAs (Huang et al., 2017). As the functional roles of snoRNAs in PDAC remain unclear, future studies will explore whether the snoRNAs analyzed in this study are functionally linked to posttranscriptional regulation and the motility and invasiveness of PDAC cells.
A SYBR Green I assay showed that mRNAs for CCDC88A, ARF6, Vav3, and WASF2 and snoRNAs for SNORA14B, SNORA22, SNORA25, SNORA74A, and SNORD22 were present in intracellular exosomes of S2-013 cells. All these mRNAs and snoRNAs except for CCDC88A mRNA were secreted and present in extracellular exosomes of cell culture medium harvested from S2-013 cells. The SNARE protein YKT6, which is targeted and regulated by miR-134 and miR-135b, regulates exosome release in lung cancer cells (Ruiz-Martinez et al., 2016). Suppression of YKT6 reduces exosome release from lung cancer cells (Ruiz-Martinez et al., 2016). Since exosomes are highly heterogeneous (Kowal et al., 2016) and the expression of CCDC88A mRNA was significantly increased in serological samples from PDAC patients, it is possible that a key molecule may Table 8. Univariate and multivariate logistic regression analyses. CA19-9, cancer antigen 19-9; Model 1: odds ratio adjusted for CA19-9; Model 2: odds ratio adjusted for CA19-9, age, and gender.
regulate CCDC88A mRNA-containing exosome release from S2-013 cells. The serum concentration and composition of exosomes are altered by different pathophysiological conditions (Patel et al., 2016). Hypoxic conditions in the tumor microenvironment promote tumor metastasis by altering exosome release to regulate cell-cell communication (Ackerman and Simon, 2014;Lu and Kang, 2010). Although the process of how particularly selected exosomes are secreted from PDAC cells is largely unknown, it is likely that pathophysiological conditions affect exosome release from PDAC cells.
It is therefore predicted that the serum level of CCDC88A mRNA would not be upregulated in PDAC patients whose PDAC tumors contain a high population of PDAC cells, such as S2-013, that have inhibited release of CCDC88A mRNA-containing exosomes. In contrast, the serum level of CCDC88A mRNA would be upregulated in PDAC patients with a low population of PDAC cells that have inhibited release of CCDC88A mRNA-containing exosomes. The PDAC-specific mechanisms by which key molecules regulate exosome release and the particular exosomal RNAs that are secreted are important subjects for the identification of diagnostic markers associated with PDAC exosomes. The number of exosomes is elevated in the systemic circulation of patients with PDAC (Melo et al., 2015). Exosomes from cancer cells secrete mRNAs, microRNAs, snoRNAs, and lncRNAs into body fluids such as the blood, urine, milk, and saliva (Laurent et al., 2015). The isolation of exosomes from cancer cells could lead to the identification of specific diagnostic markers capable of distinguishing exosomes from cancerous and noncancerous cells (Melo et al., 2015). Levels of the microRNA 17-5p are elevated in serum exosomes of PDAC patients, and increased 17-5p microRNA expression is positively correlated with metastasis and staging (Que et al., 2013). hTERT mRNA, the transcript of the enzyme telomerase, is shuttled from PDAC cells via exosomes into telomerase-negative fibroblasts; however, whether hTERT mRNA is present in serum PDAC-derived exosomes remains unclear (Gutkin et al., 2016). Serum concentrations of the snoRNA SNORD91B are decreased in PDAC patients (Liu et al., 2014), and to date, there is no evidence indicating upregulation of snoRNA expression in PDAC development. To the best of our knowledge, this study is the first report of elevated snoRNA levels in the circulation of PDAC patients.
It is important to identify PDAC-specific circulating mRNAs and snoRNAs that could be highly sensitive and specific serum markers of this disease. Although the sample size of this study was limited, the present study demonstrated that ORs adjusted for  showed that the serum level of WASF2 mRNA was the most highly correlated with the risk of PDAC (Table 8), and the lower limits of the 95% CI of the AUCs were relatively high, especially those of WASF2 and ARF6 (0.875 and 0.867, respectively; Table 3), and distribution bias was not detected between the early stages of PDAC (stages 0, I, and IIA) and the late stages of PDAC (stages IIB and III; Fig. 5B). Thus, the present study provides the statistically meaningful finding that analyzing serum levels of SNORA74A, SNORA25, WASF2, and ARF6 would be a useful novel approach for distinguishing patients with PDAC from patients without the disease. More extensive studies aimed at clarifying the potential superiority of SNORA74A, SNORA25, WASF2, and ARF6 to CA19-9 in distinguishing patients with stage 0/I/IIA PDAC from controls are needed. Additionally, the combination of measuring the levels of one of these RNAs along with that of CA19-9 may be superior to the use of CA19-9 alone in establishing a diagnosis of PDAC. Toward this end, we have started two prospective clinical validation studies (UMIN #000021938 and UMIN #000031970) at the Kochi Medical School Hospital, Kochi Health Sciences Center, Hata Prefectural Hospital, and Kanagawa Cancer Center to assess the utility of SNORA74A, SNORA25, WASF2, and ARF6 as potential diagnostic markers for the early detection of PDAC in comparison with CA19-9.

Conclusions
The present study revealed that WASF2, ARF6, SNORA74A, and SNORA25 may be novel, noninvasive diagnostic biomarkers for the early detection of PDAC. In particular, monitoring serum levels of WASF2 mRNA may be useful, as it was the most highly correlated with PDAC risk. These findings should be validated in large-scale prospective clinical studies to use them as diagnostic markers for the detection of PDAC.