Corroborative evidences have identified the potential use of miRNAs as genomic biomarker for disease diagnosis, specifically for CRC (18–20). However, there is still limited consensus on a single or a group of miRNAs that can solidly marked as CRC biomarker. The role of MiRNAs are heterogenous, they can act both as oncoMir by promoting cell growth or TS-miRs that control the cell growth (11). Thus, identifying the role of DEMs in CRC patients and associated signatures is beneficial in increasing our understanding of these molecular types. We screened the DEMs and investigated their functions and mechanisms. Based on web tool evaluations, we attempted to determine which miRNAs were strongly related to CRC.
To answer the association of miRNAs with potential related disease, a data from miRCancer database was retrieved. According to analyses, only hsa-miR-192-5p, has-miR-20a-5p, hsa-miR-21-5p, hsa-miR-215-5p, hsa-miR-24-3p, and hsa-miR-29a-3p were reported to be related with CRC (see Additional file 1). Although these miRNAs seem to be potential biomarker for CRC, we also discovered that each of the DEMs were also linked to other disease, such as hepatocellular carcinoma, lung adenocarcinoma, gastric cancer, osteosarcoma, rectal cancer, glioblastoma, ovary cancer, esophageal squamous cell carcinoma, and other. The dynamic results of the identified miRNAs are possibly due to the capability 2–8 of its 5’ complement sequence of the same miRNA that binds to several different of 3’ untranslated region (UTR) of mRNA (10, 21). For example, a study by Mokhlesi and Talkhabi concluded the hsa-miR-192-5p could potentially target genes associated with lung adenocarcinoma (22), but two years later, another study by Toolabi and team predict that hsa-miR-192-5p may targets genes associated with colon cancer (23).
According to our clinicopathological data in Table 1, half of the recruited samples have liver/lung metastases. We hypothesised that the CRC tumor in our patients had metastasized to other areas of the body, or that it was a secondary tumor that was spread from another part of the body, such as the lung or liver. Due to lack of relevant data, we were unsure to explain the relationship of the given DEMs with other disorders, particularly in lung and liver cancer. However, previous study claims that immunohistological examination can be perform to distinguish whether the metastasis originates from the colon or from other sources (24). Thus, from a clinical standpoint, patients diagnosed with CRC should be tested for other types of cancer, and the tumor excised should be subjected to immunohistological investigation (25). Histologically, the mucinous type of tissue might have different dysregulation of miRNAs from adenocarcinoma type of tissue. Mucinous histology is frequently correlated with the alteration of BRAF, PIK3CA and growth β-factor pathway than non-mucinous histology (26). In our study, we managed to have a minor number of mucinous tissue type, therefore we unable to analyse DEMs histologically.
The divergence of miRNA expression is also a point of contention. According to one study, hsa-miR-192 is downregulated in colon adenoma tissue (27), which contradicts our profile result of upregulation. There are various factors that can alter the dynamical expression of miRNAs such as miRNA availability and abundance in distinct cell types, miRNA compartmentalization, or cell state way (28). Interestingly, other researchers discovered that several of these miRNAs (hsa-miR-20a-5p, hsa-miR-21-5p and hsa-miR-24-3p) were also present in non-tumor tissue samples (29, 30). The presence of miRNAs in other sample sources, such as feces, urine and blood, may explained by miRNA transportation (31–33) and its ability to communicate with the nearby cells (34–36). Their conserved (37, 38) and stable features (12, 39, 40) make them as a convenient screening tools for disease diagnosis in future use.
Even so, there are few miRNAs that might not be reported yet in miRCancer database, as the results remains elusive or unavailable, such as hsa-miR-3653-3p, hsa-miR-3945, hsa-miR-5090, hsa-miR-5196-5p, and hsa-miR-6133. Notably, miRWalk database also unable to retrieve the target genes for these downregulated DEMs. However, the DIANA-micro T database can identify the target genes for the significantly expressed miRNAs, including the downregulated miRNAs identified in our work (see Additional file 1). Since there are numerous of available databases in identifying miRNAs mechanisms, the choice of databases is crucial as they pertained different algorithms. Despite of the different capabilities of each prediction tools, the prediction still run based on the common features of miRNA which are seed-match, conservation, free energy and site complementarity, and accessibility (41).
We enriched the miRNAs’ function by identifying their GO and pathway. We used two enrichment predictive tools; g:Profiler as shown by the result in Fig. 4 and DAVID in Table 2. The g:Profiler database is a free web tools that equipped with informative visual of results presentation. Its features that support the ranked gene lists reveal the potential GO based on the significant values from our listed gene. As presented in Fig. 4, the genes targeted by our DEMs have “aspartic-type endopeptidase” and “apeptidase activity” as the molecular function (MF). The biological process (BP) is enriched with “phosphatidylserine exposure on apoptotic cell surface” only but for the cellular component (CC), the genes are encoded for “late endosome lumen” and “endosome lumen”.
Both “aspartic-type endopeptidase” and “apeptidase activity” were classified under peptidase activity. High expression on peptidase activity lead to tumor invasion, angiogenesis and metastasis. In addition, peptidases are crucial to modulate apoptosis, regulate immune response for anti-tumor, and promote growth and development of cancer stem cells and the transition of epithelial to mesenchymal cell type (42). Prior this, there is study reported that the “aspartic-type endopeptidase” is enrich in biological process of gastric cancer tissue (43). Similarly, the genes mostly contribute for “phosphatidase exposure on apoptotic cell surface” for BP that has been reported to be induced by apoptotic factor (44) and mostly associated with endothelial-tumor cells (45). For CC, our targeted genes are active in “late endosome lumen” and “endosome lumen”, which actively involved in transportation of intracellular membrane vesicles (46). In short, our dynamics GO analyses may be heavily reliant on the clinicopathological data that we have collected, where half of samples were diagnosed with advanced stage and the cancer cell have met to liver/lung mostly. Thus we postulated that our GO analysis revealed from g:Profiler, highly associated with tumorigenesis within the lumen lesion and cell apoptosis.
On the other side, the GO and targeted pathway predicted by DAVID are more robust. We ranked the data as can be seen in Table 2. DAVID also reveal the same enriched BP as g:Profiler which is phosphatidylserine exposure on apoptotic cell surface as one of the presented top 10 data. Again, the DAVID is synchronized with g:Profiler when reported “multivesicular body lumen” for the CC and “aspartic-type endopeptidase activity” for the MF. DAVID has more privilege in predicting the targeted pathway. We also managed to retrieved the additional pathways from two databases through DAVID; KEGG and Reactome. KEGG mostly reported on cell signaling pathway. E.g. “MAPK signaling pathway”, “Ras signaling pathway”, “Neurotrophin signaling pathway”, “Rap1 signaling pathway”, “PI3K-Akt signaling pathway”. Meanwhile, Reactome revealed the transcriptional activity. E.g. “Downregulation of SMAD2/3:SMAD4 transcriptional activity”, “Signal transduction”, “Transcriptional Regulation by TP53”, “Dimerization of procaspase-8”, and “Regulation by c-FLIP”.
Aside from presenting the data on targeted genes and ranking probable mechanisms from the DEMs given, we also examined the DEMs separately, which is expected to aid future decisions in researching a panel of single miRNAs or clusters of miRNAs as CRC biomarkers. According to individual analyses of DEMs, as shown in Additional file 2, we suggested that hsa-miR-20a-5p, hsa-miR-21-5p, hsa-miR-23a-3p, hsa-miR-24-3p, and hsa-miR-29a-3p either as a panel/panel cluster for CRC potential biomarkers based on the pathway recognition. They were recognized to commonly interact with “PI3K-Akt signaling pathway”, “WNT signaling pathway” and “FGFR signaling” that is highly associated with CRC pathogenesis.
Despite of promising evidence from web tools used, our enrichment finding analyses might be different from evidences provided through Malacards, The Human Disease Database, (https://www.malacards.org/#), which is an integrated database that compiled the information on genomic-associated disease. According to Malacards, the top 10 pathway that are directly related to CRC are signaling by “TCF7L2 Mutants”, “Defective MUTYH Substrate Processing”, “Defective MUTYH Substrate Binding” and “Defective Base Excision Repair Associated with MUYTH”. Other pathways that associated with CRC are “ERK Signaling”, “Signal Transduction”, “TGF-Beta Pathway”, “Prolacting Signaling”, “RAF/MAP Kinase Cascade”, “Phospholipase-C Pathway”, “Beta-Adrenergic Signaling”, “Signaling by WNT” and “NFAT and Cardiac Hypertrophy”.
Collectively, we believe that our data provide additional knowledge and insight regarding miRNAs underlying mechanisms in CRC of Asian Malay population. Apart from the genetic make-up, the study of miRNAs across the specific population might need to consider the gender, environmental causal and demographical barriers of population as well. For the time being, less report is critically discussed on this evidence. Although our findings showed that not all of the screened DEMs are linked to CRC, we believed that this could contributed to the miRNA’s profiling data, especially in Asian Malay community. Prior this, CRC was mostly diagnosed in western countries (47, 48) and recently was identified in Asian and Gulf country (49, 50). Seeing this emerging cases, a more practical approach in identifying epigenetic miRNAs as CRC biomarker is needed. Although our study may seem lack of data in comparisons of DEMs across the populations, we believe that this study provides an insight on miRNAs expression in Asian Malay population since less report is found regarding this. Other screening based on sex or stage is also preferable for possible correlation with clinicopathological data, because miRNA expression is epigenetic and may be regulated by other environmental factors.