An Exosomal miRNA Biomarker for the Detection of Pancreatic Ductal Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) remains a difficult tumor to diagnose and treat. To date, PDAC lacks routine screening with no markers available for early detection. Exosomes are 40–150 nm-sized extracellular vesicles that contain DNA, RNA, and proteins. These exosomes are released by all cell types into circulation and thus can be harvested from patient body fluids, thereby facilitating a non-invasive method for PDAC detection. A bioinformatics analysis was conducted utilizing publicly available miRNA pancreatic cancer expression and genome databases. Through this analysis, we identified 18 miRNA with strong potential for PDAC detection. From this analysis, 10 (MIR31, MIR93, MIR133A1, MIR210, MIR330, MIR339, MIR425, MIR429, MIR1208, and MIR3620) were chosen due to high copy number variation as well as their potential to differentiate patients with chronic pancreatitis, neoplasms, and PDAC. These 10 were examined for their mature miRNA expression patterns, giving rise to 18 mature miRs for further analysis. Exosomal RNA from cell culture media was analyzed via RTqPCR and seven mature miRs exhibited statistical significance (miR-31-5p, miR-31-3p, miR-210-3p, miR-339-5p, miR-425-5p, miR-425-3p, and miR-429). These identified biomarkers can potentially be used for early detection of PDAC.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) remains one of the most dismal types of cancers diagnosed with a 5-year survival rate of 10% [1]. PDAC is projected to overtake both breast and colorectal cancers as the second leading cause of cancer-related deaths before 2030 [2,3]. PDAC is typically asymptomatic and only 10-25% of patients are diagnosed in the early stages, while the majority are diagnosed during late stage disease [4]. Additionally, 60-80% of patients presenting resectable pancreatic tumors will exhibit recurrence, regardless of administration of adjuvant therapy [5,6]. To date, the only protein marker approved and designated for PDAC diagnosis and monitoring is Cancer Antigen 19-9 (CA19-9). CA19-9 is released at low levels by organs of the gastrointestinal tract as well as endometrial cells [7]. Elevated levels of the protein may be present in PDAC as well as various diseases including pancreatitis and gastrointestinal disorders and tumors [7,8]. Additionally, CA 19-9 is recommended as a prognostic indicator and not suggested for screening or as an early detection marker for PDAC [9,10]. Hence, there is a dire need for alternative screening methods in which high-risk patients may be monitored for the potential of not only developing PDAC but also monitoring for the presence of resurgent tumors.
Patients with high risk of developing PDAC are those with a history of smoking, obesity, chronic pancreatitis, hereditary PDAC, and onset of diabetes after 50 years of

Cell Lines
All cell lines were purchased from the American Type Culture Collection in 2019. Cells were used within 8 months of purchase for analysis; cell lines were not re-authenticated nor was mycoplasma testing conducted. PDAC cell lines PANC1 (ATCC ® CRL-1469), BXPC3 (ATCC ® CRL-1687), and CAPAN2 (ATCC ® HTB-80) were chosen for their differences in mutation profile. PANC1 has mutations in KRAS and p53, BXPC3 possesses mutant p53 and wildtype KRAS, and CAPAN2 exhibits a KRAS mutation and is wildtype for p53. We used pancreatic duct epithelial cell line hTERT-HPNE E6/E7/st (ATCC ® CRL-4037) as a non-cancerous control cell line. This is a stable cell line immortalized by telomerase catalytic subunit, with p53 and Rb deactivation by HPV oncogenes E6 and E7, respectively. SV40 large T protein is a simian-derived oncogene which was introduced to this cell line by Campbell et al. in 2007 to sensitize the cell line to KRAS mutation [48]. All cell lines were cultured as per ATCC guidelines specific for each cell line; however, standard FBS was replaced with exosome depleted FBS as media supplement for all cell lines. PANC1 media is comprised of 450 mL of DMEM, 50 mL of exosome-depleted FBS, and 5 mL of penicillin-streptomycin; BCPC3 media includes 450 mL RPMI-1640, 50 mL of exosomedepleted FBS, and 5 mL of penicillin-streptomycin; CAPAN2 media was formulated using 450 mL of McCoy's 5a Medium, 50 mL exosome-depleted FBS, and 5 mL of penicillinstreptomycin; and hTERT-HPNE E6/E7/st base medium is comprised of 375 mL of low glucose DMEM (Sigma Aldritch Cat# D-5030), 125 mL Medium M3 Base (Incell corporation, cat# M300F500), 27 mL exosome-depleted FBS, 5.4 mL of hr EGF stock (1 µg/mL, Gibco cat# PHG0314), 5.4 mL L-glutamine (ATCC cat# 30-2214), 1.4 mL D-glucose (Sigma cat# G8644), and 5 mL of penicillin-streptomycin. To prepare the 1 µg/mL of EGF stock solution, 1 vial of 10 µg EGF is combined with 10 mL of PBS and 10% BSA and filter sterilized before adding the EGF.

Exosome Isolation from Cell Culture Media
Cell cultures were grown for two passages before they were split into three T25 flasks and cultured for three days in 5 mL of their respective media. All cell lines exhibited at least 80% confluency and 90% viability as determined by Trypan Blue staining. Volumes of cell culture media from each sample were collected in accordance to a normalized cell count of 1 × 10 6 cells/mL across all cell lines and biological replicates. Exosomes were isolated from each normalized sample using the Thermofisher ® Total Exosome Isolation Kit (for cell culture media) as per manufacturer directions.

microRNA Nomenclature
MicroRNA have different naming schemes, depending on their level of processing. MIR refers to the gene encoding the miRNA; mir designates the pre-and primary transcript; miR denotes the mature form of the miRNA after it has been processed [50].

Database Generation and Lead Identification
Our workflow strategy incorporates extensive bioinformatics analysis to identify potential leads for in vitro analysis ( Figure 1). In order to generate this database, multiple knowledgebases were accessed to determine associated PDAC genes (N = 6136). Upon analysis and enrichment of the datasets provided by these various knowledgebases, a database of 383 PDAC-associated ncRNA was generated (Supplementary File S1). The majority (86%) of the ncRNAs in the generated database belong to the MIR subtype ( Figure 1). This provided the basis for a panel of miRNA for PDAC detection. To generate a dataset of MIRs which exhibited alteration in PDAC patients, the cBioPortal tool was utilized.
The cBioPortal tool provides valuable information regarding copy number alteration for a given gene of interest as well as tumor staging. It was predicted that if enough individuals exhibited alterations in the MIRs of interest it would translate as a change of expression of those MIRs in affected individuals. This would then potentially provide a fingerprint of potential miRNAs for PDAC detection. The University of Texas South Western (UTSW) PDAC dataset (N = 109) was used to batch analyze the 383 ncRNAs in the database. To be considered for further analysis, the cBioPortal tool had to report a copy number alteration of the specific ncRNA in a minimum of 10% of the patients in the sample. This criterion yielded 72 miRNAs.
analysis and enrichment of the datasets provided by these various knowledgebases, a database of 383 PDAC-associated ncRNA was generated (Supplementary File S1). The majority (86%) of the ncRNAs in the generated database belong to the MIR subtype ( Figure  1). This provided the basis for a panel of miRNA for PDAC detection. To generate a dataset of MIRs which exhibited alteration in PDAC patients, the cBioPortal tool was utilized. The cBioPortal tool provides valuable information regarding copy number alteration for a given gene of interest as well as tumor staging. It was predicted that if enough individuals exhibited alterations in the MIRs of interest it would translate as a change of expression of those MIRs in affected individuals. This would then potentially provide a fingerprint of potential miRNAs for PDAC detection. The University of Texas South Western (UTSW) PDAC dataset (N = 109) was used to batch analyze the 383 ncRNAs in the database. To be considered for further analysis, the cBioPortal tool had to report a copy number alteration of the specific ncRNA in a minimum of 10% of the patients in the sample. This criterion yielded 72 miRNAs. Whole gene lists associated with PDAC were downloaded from the lnc and miRNA databases as well as GeneCards, NCBI Gene, and DisGenNET in order to identify as many PDAC-associated ncRNAs as possible (N = 6136). Protein-coding genes were then removed in order to isolate only ncRNAs (N = 383). This provided the basis for our ncRNA database. cBioPortal was used to identify genetic alterations of ncRNAs across patient data in order to determine the most attractive targets for diagnostic potential (N = 72). The expression databases (COSMIC, UniGene, SAGE, CGAP, and PED) and secretome tools (ExoCarta and GeneALaCart) were then accessed to determine if any of the remaining MIRs were secreted and exhibited changes in expression in PDAC to better determine the most reliable exosomal targets for PDAC detection (N = 50). Additional optimization using cBi-oPortal with putative secreted exosomal MIR markers (CNV 15% or greater) was used (N = 18). Comparison of MIR expression in chronic pancreatitis compared to PDAC was then used as the final metric for candidate MIRs for diagnostic potential (N = 10). The MIRs identified in this manner were considered diagnostic markers for further analysis in cell culture models. These 72 MIRs were enriched for their presence in exosomes using the exosome encyclopedia tool, ExoCarta, which provides general evidence for the presence of miRs in exosomes. This analysis yielded 50 exosomal MIRs (Supplementary File S2). We then used the cBioPortal batch analysis function on the UTSW dataset to test combinations of exosomal MIRs enriched from our dataset to determine the most optimal MIRs to collectively test as a potential diagnostic panel.  Whole gene lists associated with PDAC were downloaded from the lnc and miRNA databases as well as GeneCards, NCBI Gene, and DisGenNET in order to identify as many PDAC-associated ncRNAs as possible (N = 6136). Protein-coding genes were then removed in order to isolate only ncRNAs (N = 383). This provided the basis for our ncRNA database. cBioPortal was used to identify genetic alterations of ncRNAs across patient data in order to determine the most attractive targets for diagnostic potential (N = 72). The expression databases (COSMIC, UniGene, SAGE, CGAP, and PED) and secretome tools (ExoCarta and GeneALaCart) were then accessed to determine if any of the remaining MIRs were secreted and exhibited changes in expression in PDAC to better determine the most reliable exosomal targets for PDAC detection (N = 50). Additional optimization using cBioPortal with putative secreted exosomal MIR markers (CNV 15% or greater) was used (N = 18). Comparison of MIR expression in chronic pancreatitis compared to PDAC was then used as the final metric for candidate MIRs for diagnostic potential (N = 10). The MIRs identified in this manner were considered diagnostic markers for further analysis in cell culture models.

Selection of Biomarker Panel for Diagnostics and Screening
Tumor progression is marked by chronic inflammation and resultant aberrant gene expression. It was therefore of interest to establish whether the miRNAs in the proposed panel exhibited progressive deregulation associated with chronic pancreatitis and tumor stage progression. Expression of the miRNAs in chronic pancreatitis and PDAC were predicted using the pancreatic expression database (PED) [47] and partially verified using the pancreatic cancer database (PCD) [40] (Supplementary File S3). The presence of altered miRNAs in proceeding PDAC stages was predicted using the Staging clinical track provided by cBioPortal. These tools provided additional optimization of the proposed diagnostic and screening panel.  (Table 1). Additionally, four miRNAs (MIR27A, MIR203A, MIR210, MIR429) were shared between stages I and IV. Meanwhile stages III and IV appeared to share three miRNAs (MIR210, MIR429, and MIR1208).
The cBioPortal tool is a publicly curated database and enables the addition of several different tracks, including a tumor staging track. Because chronic pancreatitis increases the risk of developing PDAC, it was of interest to determine if differential expression could be observed within the miRNAs between the two diseases. The PED was used to determine the RNA expression of these MIRs in pancreatitis compared to PDAC. Eight miRNAs (MIR31, MIR96, MIR130B, MIR210, MIR339, MIR429, MIR590, and MIR1208) were downregulated in pancreatitis patients when compared to healthy pancreatic tissue. When the expression of these MIRs were compared between PDAC and pancreatitis, five miRNAs (MIR27A, MIR31, MIR93, MIR130B, and MIR330) were upregulated (Supplementary Table S1).

Detection of the miRNA Panel in Pancreatic Cancer In Vitro Models
Cell culture media from each of the cell lines was collected after three days for exosomal RNA isolation, extraction, and analysis by RT-qPCR. It is well established that miRs may not be detectable due to low levels, therefore, any values over 35, or a lack of Cq values were replaced with the arbitrary low value of 36, as previously described [49]. Additionally, calculations were performed as long as 20% of the technical replicates exhibited values less than 35. This method was applied to the data for all four cell lines.
The data for miR-31-3p and miR-31-5p is consistent with current literature, which has shown a clear link between KRAS mutation and rampant overexpression of miR-31 [51][52][53][54]. Interestingly, overexpression of miR-429 has been observed in PDAC patient-derived xenograft models while also exhibiting associations with decreased metastasis in functional studies [55], increased sensitization to chemotherapy [56], and further suppression of metastasis in in vitro PDAC [57] and hepatocellular carcinoma models [58]. It is unclear why miR-429 may be overexpressed in PDAC models and yet exhibits associations with anti-tumor properties, though it may be due to different underlying biological mechanisms governing in vitro versus in vivo systems, or it could be due to alteration as tumors progress from one stage to the next. 0.05; CAPAN2 p = 0.381) (Figure 2A-C). Only PANC1 exhibited detectable expression changes for miR-93-3p (p = 0.7759) and miR-133a-3p (p = 0.3000), though neither were significant. miR-339-3p was detected in PANC1 (p = 0.1416) and CAPAN2 (p = 0.9769), but not in BXPC3. CAPAN2 and BXPC3 also exhibited significant log2 fold change expression for miR-429, p ≤ 0.01 ( Figure 2B,C). Figure 3 summarizes the miRs which were shared or specific to each cell line.  The data for miR-31-3p and miR-31-5p is consistent with current literature, which has shown a clear link between KRAS mutation and rampant overexpression of miR-31 [51][52][53][54]. Interestingly, overexpression of miR-429 has been observed in PDAC patient-derived xenograft models while also exhibiting associations with decreased metastasis in functional studies [55], increased sensitization to chemotherapy [56], and further suppression of metastasis in in vitro PDAC [57] and hepatocellular carcinoma models [58]. It is unclear why miR-429 may be overexpressed in PDAC models and yet exhibits associations with  The data for miR-31-3p and miR-31-5p is consistent with current literature, which has shown a clear link between KRAS mutation and rampant overexpression of miR-31 [51][52][53][54]. Interestingly, overexpression of miR-429 has been observed in PDAC patient-derived xenograft models while also exhibiting associations with decreased metastasis in functional studies [55], increased sensitization to chemotherapy [56], and further suppression of metastasis in in vitro PDAC [57] and hepatocellular carcinoma models [58]. It is unclear why miR-429 may be overexpressed in PDAC models and yet exhibits associations with

Cell Preferences for Mature miRNA
Each miRNA was analyzed for the presence of both its mature 5p and 3p arms. This was to determine if the cells exhibited preferential expression for an miR to develop a more targeted diagnostic. Previous studies have shown various tumor types exhibit non-equal expression levels for either the 5p or 3p arm of a mature miRNA, whereby one arm is significantly more expressed than the other [59][60][61][62]. The relative expression levels of each arm of the studied miRs were analyzed for significant differences to determine if there was a preference for the 5p or 3p arm in our cell culture models.
There was an overwhelming preference for the mature 5p arm across all cell lines. There was an overwhelming preference for the mature 5p arm across all cell lines. Despite this, only 4 miRs exhibited an appreciable difference between 5p and 3p expression (Figure 4). miR-93-5p reported a statistical significance in expression over the 3p arm in both PANC1 and CAPAN2, p ≤ 0.01. The only miR which exhibited a 3p preference was miR-210, in PANC1, and did not exhibit any detectable levels of miR-210-5p (p ≤ 0.01). CAPAN2 also exhibited an increase in 5p expression over 3p for miR-339 (p ≤ 0.05). Lastly, for both PANC1 and CAPAN2, miR-425-5p significantly overexpressed compared to miR-425-3p (p ≤ 0.05 and p ≤ 0.01, respectively).
Each miRNA was analyzed for the presence of both its mature 5p and 3p arms. This was to determine if the cells exhibited preferential expression for an miR to develop a more targeted diagnostic. Previous studies have shown various tumor types exhibit nonequal expression levels for either the 5p or 3p arm of a mature miRNA, whereby one arm is significantly more expressed than the other [59][60][61][62]. The relative expression levels of each arm of the studied miRs were analyzed for significant differences to determine if there was a preference for the 5p or 3p arm in our cell culture models.
There was an overwhelming preference for the mature 5p arm across all cell lines. There was an overwhelming preference for the mature 5p arm across all cell lines. Despite this, only 4 miRs exhibited an appreciable difference between 5p and 3p expression (Figure 4). miR-93-5p reported a statistical significance in expression over the 3p arm in both PANC1 and CAPAN2, p ≤ 0.01. The only miR which exhibited a 3p preference was miR-210, in PANC1, and did not exhibit any detectable levels of miR-210-5p (p ≤ 0.01). CA-PAN2 also exhibited an increase in 5p expression over 3p for miR-339 (p ≤ 0.05). Lastly, for both PANC1 and CAPAN2, miR-425-5p significantly overexpressed compared to miR-425-3p (p ≤ 0.05 and p ≤ 0.01, respectively). . Preference for 5p and 3p arm across PDAC cell lines and HPNE control. Each cell line was examined for a preference in 5p or 3p arm of the mature miRNA Expression levels were calculated relative to the 5p arm of each miR and displayed as fold change values. (A) miR-93-5p was preferred in PANC1 and CAPAN2, while (B) miR-210-3p was preferred in PANC1. (C) miR-339-5p was preferred in CAPAN2 and (D) miR-425-5p were preferred in both PANC1 and CAPAN2. Significant (p ≤ 0.05, *) and very significant (p ≤ 0.01, **) are also noted and were determined using the student's t-test.

Discussion
Exosomes have numerous advantages as potential diagnostic vehicles. They provide a snapshot of the internal RNA and protein composition in cells at various stages of disease progression, and their stability in body fluids facilitates the relative ease of collection . Preference for 5p and 3p arm across PDAC cell lines and HPNE control. Each cell line was examined for a preference in 5p or 3p arm of the mature miRNA Expression levels were calculated relative to the 5p arm of each miR and displayed as fold change values. (A) miR-93-5p was preferred in PANC1 and CAPAN2, while (B) miR-210-3p was preferred in PANC1. (C) miR-339-5p was preferred in CAPAN2 and (D) miR-425-5p were preferred in both PANC1 and CAPAN2. Significant (p ≤ 0.05, *) and very significant (p ≤ 0.01, **) are also noted and were determined using the student's t-test.

Discussion
Exosomes have numerous advantages as potential diagnostic vehicles. They provide a snapshot of the internal RNA and protein composition in cells at various stages of disease progression, and their stability in body fluids facilitates the relative ease of collection compared to traditional invasive biopsy methods [63]. Despite numerous ncRNA expression studies in various cancer and tumor systems [37,51,[64][65][66][67][68][69][70], the potential clinical application of this breadth of information remains limited and it is only recently that their usage is being examined in several clinical trials [24,71].
In this study, we utilized extensive bioinformatics analysis and enrichment processes to propose a panel of exosomal miRNAs to be used as a potential diagnostic for PDAC. Of the 10 miRs studied, we found seven (miR-31-5p, miR-31-3p, miR-210-3p, miR-339-5p, miR-425-5p, miR-425-3p, and miR-429) that were differentially expressed in PDAC cell lines compared to the control, and also found four (miR-93-5p, miR-210-3p, miR-339-5p, and miR-425-5p) that exhibited a preference for one arm over the other in PDAC cell lines but not arm preference in the control. These biomarkers are strong candidates for the development of a novel non-invasive diagnostic panel for PDAC, with the potential to improve early detection of pancreatic cancer.
Interestingly, miR-133a-5p, miR-210-5p, miR-330-5p, miR-330-3p, miR-1208, miR-3620-5p, and miR-3620-5p were not expressed in any of the tested cell lines. Alternatively, miR-133a-3p was detectable only in the PANC1 cell line. Previous studies suggest that miR-133a, miR-1208, and miR-3620-3p exert tumor suppressive effects [72][73][74][75][76]. The absence of these miRs in the PDAC cell lines may therefore be consistent with the tumor suppression phenotype observed in these previous studies. However, the presence of detectable levels of miR-133a-3p in the PANC1 cell line may be due to the specific mutations associated with this specific cell line. Previous studies have also shown that miR-210-5p is over expressed in bladder cancer, breast cancer, kidney tumors, lung squamous cell carcinoma (LSCC), and stomach cancer but not in PDAC [77]. Meanwhile, miR-210-3p exhibited high levels of expression in bladder, breast, kidney, LSCC, and pancreatic tumors, which is consistent with our observations and further supports tissue-specific expression. Previous studies have found that miR-330-5p may also function as a tumor suppressor by inhibiting PDAC progression, and low levels of the miR have been reported in PDAC tissues [78,79]. Contrary results have been observed with miR-330-3p, with one study suggesting a pro-tumorigenic property in PDAC [80], while others suggesting anti-tumor effects in liver and ovarian cancers [81,82]. Finally, miR-3620-5p is known to form G-quadruplex structures with itself [83]. In the present study, miR-3620-5p was the only miR that had detectable levels of expression (Ct < 35) in the no template controls of the RT-qPCR analysis (data not shown). This was likely due to the miR-3620-5p specific-probe forming a dimer-like structure thus yielding a detectable signal. Thus, it remains difficult to accurately assess miR-3620-5p expression.
In the cBioPortal dataset, MIR31 exhibited deep deletion in 27 patients, which would suggest decreased expression of MIR31 in about 25% of patients. This decrease in expression was not observed in our in vitro models, nor the other 75% of patients sampled in the UTSW dataset. This underscores the variability in expression of MIR between patients and the importance of a multi-marker diagnostic. To that end, our in vitro data for miR-31-5p and miR-31-3p showed significantly upregulated levels in CAPAN2 (miR-31-5p and miR-31-3p) and PANC1 (miR-31-3p) compared to HPNE. This observation is consistent with the literature which has shown a strong positive correlation between mutant KRAS and rampant miR-31 overexpression [51,52,54]. Regardless, it is well established that in vivo and in vitro analyses are often conflicting. For example, one in vitro study utilizing aggressive PDAC cell lines suggests that overexpression of miR-429 correlates with poor survival in later stages [84]. On the other hand, additional in vitro studies state the opposite and have labeled miR-429 as a potential tumor suppressor [56][57][58] as has an in vivo study using patient pancreatic cancer xenografts [55]. Lack of consistency in the expression of target genes amongst the literature is largely due to differences in patient polymorphisms [70,85] as well as potential differences between specific models being used. These variables highlight and support the requirement for diagnostic panels, rather than a single biomarker.
KRAS proto-oncogene and TP53 mutations are ubiquitous for pancreatic cancer. In many cases, a KRAS mutation is the initiating driver of pancreatic tumorigenesis [86]. The KRAS proto-oncogene is a GTPase central to the RAS/MAPK pathway. RAS proteins are crucial to cell growth, proliferation, migration, differentiation, and apoptosis in a tightly regulated cellular system [87]. A single point mutation in residue 12 of glycine to aspartic acid or valine (KRASG12D or KRASG12V) renders KRAS proteins constitutively active [88,89]. This results in rampant growth and proliferation. TP53 is a well-established tumor suppressor, that encodes a cell cycle checkpoint monitor thereby initiating DNA damage/repair pathways and apoptosis in the event that damaged DNA cannot be repaired [90]. A mutation in TP53 results in the inactivation of the protein, allowing for unregulated proliferation when coupled with mutant KRAS. These two hallmark genes of pancreatic cancer serve as a basis for our choice of, PANC1 (p53/KRAS double mutant), CAPAN2 (p53-WT/KRASmut), and BXPC3 (p53-mut/KRAS-WT) as robust in vitro models of PDAC. However, there is a possibility that the use of additional in vitro models which exhibit different causative mutations may give rise to differing results than are presented here.
KRAS mutant PANC1 exhibited four statistically significant miRs and CAPAN2 expressed six statistically significant miRs compared to BXPC3, which significantly expressed two miRs. KRAS is well established as the most common mutation in PDAC. Thus, it is possible that the reason why KRAS mutant cell lines yielded a greater number of differentially expressed miRs is due to the ability of constitutively active KRAS to cause rapid growth, via signaling cascades resulting in rapid transcription of genes, thereby enabling higher expression levels of these miRNAs [91,92]. Interestingly, miR-429 exhibited increased expression levels in the KRAS mutant CAPAN2 (Log2 fold change 7.28), and p53 mutant BXPC3 (Log2 fold change 7.67) but was undetectable in the KRAS/p53 double mutant PANC1 model. Previous studies suggest that miR-429 may be a tumor suppressor, despite being commonly overexpressed in PDAC [55]. Given the observed overexpression in CAPAN2 and BXCP3, but not in PANC1, it is possible that p53/KRAS double mutant asserts an antagonistic effect on miR-429, suppressing its expression. Additionally, the overexpression of miR-429 in the KRAS WT BXPC3 as well as KRAS mutant CAPAN2 may predict its beneficial use as a monitor for patients with a mutation in KRAS or P53, but not patients exhibiting mutations in both. Alternatively, this discrepancy may be due to the comparison of PDAC cell lines to an immortalized pancreatic duct model instead of a primary tissue model. The model of HPNE used for this work was originally designed to test the instability of KRAS, thereby providing a model for a precancerous control. Regardless HPNE E6/E7/st is a stable cell line and does not exhibit tumorigenic properties until transfected with mutant KRAS [48]. The HPNE culture we used possesses knockouts of two tumor suppressors (p53 and retinoblastoma, Rb) but maintains a differentiated phenotype and exhibits a precancerous genotype. The use of a precancerous control is a viable option for testing a diagnostic biomarker for early PDAC detection and monitoring.
A preference for the 5p arm of most of the miRs tested was observed in this study. Previous studies have observed this phenomenon with reports of arm switching in tumors [62]. Although arm switching was not observed in this study, the data shows that there was a significant overexpression of the 5p arms compared to the 3p arms in the cancer cell lines compared to the control cell line for MIR93, MIR210, MIR339, andMIR425. This statistically significant overexpression of these mature 5p arms in our miR panel may be beneficial for use as an additional diagnostic feature.
Many of the exosomal miRs identified in this study have also been observed in plasma exosomes and thus may prove beneficial as a diagnostic panel [44,93]. Additionally, six of the 18 MIR transcripts (MIR31, MIR93, MIR210, MIR330, MIR425, and MIR429) identified in the bioinformatics analysis exhibited differential expression between chronic pancreatitis patients and PDAC patients, of which all but MIR93 appear to exhibit overexpression in PDAC models compared to immortalized pancreatic duct model, HPNE. Further work testing the proposed mature miR markers and their expression in healthy, chronic pancreatitis, and PDAC patients is required to determine their efficacy as early markers for monitoring and detection. Additionally, it may be of interest to test this panel in other tumor types in order to establish specificity of the proposed marker. Thus, our proposed diagnostic marker may prove beneficial in identifying PDAC in patients as well as monitoring high risk patients and patients undergoing treatment.