Background
Colorectal cancer (CRC) constitutes a major public health burden [
1], being the third most commonly diagnosed cancer, and the fourth leading cause of cancer-related deaths worldwide. Interestingly, recent reports have shown that the incidence of colorectal cancer in Asian countries, which historically was relatively low, has increased dramatically during the last two decades [
2,
3]. Considering the high disease burden and mortality associated with this global disease, it is imperative to develop effective prevention and treatment strategies for the management of patients suffering from this malignancy.
CRC develops as a consequence of stepwise accumulation of multiple genetic and epigenetic alterations, which occur with tumor initiation and ensue during disease progression. In view of tumor heterogeneity, the prognosis and response to chemotherapy between individual patients can vary significantly. However, current guidelines for risk stratification of patients predominantly rely on the clinicopathological factors, which are inadequate and often result in under or over-treatment for CRC patients [
4,
5]. In view of this clinical challenge, identification of novel molecular targets that more robustly typify and represent disease biology would be of great value in improving prognosis and allowing precision therapeutic targeting in CRC patients.
Data gathered during the last decade has revealed that microRNAs play a crucial role in cancer pathogenesis, and may serve as promising disease biomarkers and potential therapeutic targets [
6,
7]. Meanwhile, emerging data from RNA-Sequencing efforts of cancer specimens have led to the discovery of additional classes of novel, small noncoding RNAs (ncRNAs), which may also significantly contribute to cancer pathogenesis [
8‐
12]; however, such data remain in their infancy at this time in point. Among these, PIWI-interacting RNAs (piRNAs), represent the most diversified group, but currently remain the least characterized class of small ncRNAs. The piRNA pathway consists of piRNAs that interact with PIWI proteins, in which the precursor piRNAs are transcribed from their clusters, cleaved by PIWI proteins, and subsequently amplified in the cytoplasm through sequence-complementary-dependent cycle. The piRNA-PIWI protein pathway was initially found associated with safeguarding of the germline genome against transposon-induced insertional mutations [
13‐
15]. However, emerging evidence indicates that piRNAs may also function on a somatic level, whereby they regulate gene expression; through histone modifications and DNA methylation [
16‐
19]. In other words, although such recent evidence suggests the role of PIWI-piRNA pathway in controlling epigenetic function, our understanding for the biological involvement of piRNAs in human cancers currently remains in its infancy, but presents an exciting new area of basic and translational research worthy of exploration.
Although piRNA-mediated gene expression regulation may have a broader implication for cancer research, recent studies have largely been limited to expression profiling of a handful or selected, small subset of piRNAs in different cancer types. [
9,
12] Furthermore, the role of piRNAs in CRC is poorly understood. Hence, we envisaged this first of its kind of study to systematically and comprehensively interrogate the molecular contributions of piRNAs in CRC, with a goal to identify novel, differentially expressed piRNAs that promote colorectal carcinogenesis, and decipher whether these piRNAs may have translational relevance as clinically relevant disease biomarkers. Accordingly, we performed a discovery step by performing small RNA-Sequencing-based expression profiling for piRNAs between cancer and normal tissues. Using a series of bioinformatic approaches, we identified candidate, CRC-specific piRNAs, followed by their validation in multiple CRC patient cohorts. We subsequently supported these findings by performing a series of functional assays and investigated downstream pathways and target genes of candidate piRNAs, which contribute to the neoplastic progression in colorectal cancer.
Methods
Patients and study design
To identify CRC-associated piRNAs, we performed small RNA-sequencing on a subset of frozen cancer tissues and paired normal mucosa (NM) specimens (4 each), which were collected at the Mie University, Japan. To confirm the expression levels of candidate piRNAs between cancer and normal tissues, we measured their expression in matched cancer and normal tissues in three independent patient cohorts from the Mie University, Japan (
n = 20), Shanghai Tenth People’s Hospital, China (
n = 20) and Okayama University Medical Hospital, Japan (
n = 18). To investigate the prognostic potential of candidate piRNAs in CRC, we analyzed piR-1245 expression pattern in three different patient cohorts with a combined total of 771 CRC patients from the TCGA dataset, a clinical testing cohort and an independent validation cohort. The expression profiles of piRNAs from the TCGA (The Cancer Genome Atlas) dataset (
n = 387) was characterized by Martinez, et al. [
20]. We thereafter analyzed expression of candidate piRNAs in a clinical testing cohort (
n = 195, Shanghai Tenth People’s Hospital) and an independent validation cohort (
n = 189, Okayama University Medical Hospital). Both testing and validation cohort were formalin fixed paraffin embedded tissue specimens, which allowed us to perform micro-dissection for enriching the RNA content from the neoplastic cells. The baseline characteristic of these patient cohorts is described in Table
1. To further understand the mechanistic correlation of piRNAs with its downstream target genes, we evaluated their expression in a cohort of 159, fresh frozen tissues. Written informed consent was obtained from all patients, and the study was approved by the institutional review boards of all participating institutions. All CRC patients were followed up for survival for at least 5 years from their date of surgery. Patients treated with radiotherapy or chemotherapy before surgery were excluded from the study.
Table 1
Clinicopathological characteristic and piR-1245 expression in training and validation cohort
Gender |
Male | 91 | 45 | 46 | 0.9391 | 110 | 52 | 58 | 0.4255 |
Female | 104 | 52 | 52 | | 79 | 42 | 37 | |
Age |
≤ 69a/66b | 100 | 55 | 45 | 0.133 | 100 | 45 | 55 | 0.1687 |
> 69a/66b | 95 | 42 | 53 | | 89 | 49 | 40 | |
Tumor location |
Distal | 150 | 82 | 68 | *0.0123 | 121 | 55 | 66 | 0.1174 |
Proximal | 45 | 15 | 30 | | 68 | 39 | 29 | |
Histological type |
Well/moderate | 175 | 90 | 85 | 0.0566 | 180 | 90 | 90 | 0.7456 |
Poor | 18 | 5 | 13 | | 9 | 4 | 5 | |
Unknown | 2 | – | – | | – | – | – | |
Pathological T category |
pT1-3 | 48 | 34 | 14 | **0.0008 | 154 | 82 | 72 | *0.0434 |
pT4 | 147 | 63 | 84 | | 35 | 12 | 23 | |
Lymph node metastasis |
Negative | 132 | 73 | 59 | *0.025 | 85 | 53 | 32 | **0.0025 |
Positive | 63 | 24 | 39 | | 100 | 40 | 60 | |
Unknown | – | – | – | | 4 | – | – | |
Distant metastasis |
Negative | 187 | 96 | 91 | *0.0319 | 143 | 80 | 63 | **0.0027 |
Positive | 8 | 1 | 7 | | 46 | 14 | 32 | |
Stage |
I | 29 | 21 | 8 | **0.006 | 28 | 18 | 10 | **0.0052 |
II | 99 | 51 | 48 | | 53 | 33 | 20 | |
III | 59 | 24 | 35 | | 62 | 29 | 33 | |
IV | 8 | 1 | 7 | | 46 | 14 | 32 | |
Small RNA-sequencing, piRNA quantification and gene expression analysis
For RNA-sequencing, 1 μg of total RNA was used for library preparation with Illumina’s TruSeq small RNA sample preparation Kit using manufacturer’s recommended protocols and the previously published articles mirBase [
21,
22]. Expression of identified piRNAs was analyzed using Custom TaqMan small RNA assays as described previously [
23‐
25]. The average expression levels of tissue piRNAs was normalized against U6 using the 2
-ΔCt method. The relative expression of target genes was determined by 2
-Δct method using GAPDH as a normalizer as described in details in the Additional file
1: Supplementary Methods and primer sequences shown in Additional file
2: Table S1.
Cell lines, RNA oligos, antisense and transfection
HCT116 and SW480 were obtained from the American Type Culture Collection (ATCC, Rockville, MD) and cultured in Iscove’s modified Dulbecco’s medium (Invitrogen, Carlsbad, CA). For the overexpression of piR-1245, both cells lines were transfected in triplicates with either single-stranded RNA oligos or scrambled RNA controls. For the inhibition of piR-1245 in CRC cell lines, we designed antisense oligos as described previously [
26] and as described in the Additional file
1: Supplementary Methods.
MTT, colony formation, invasion, migration and apoptosis assays were performed in CRC cell lines at different time points using standard approaches, and according to the manufacturer’s instructions, as described in the Additional file
1: Supplementary Methods.
Immunofluorescence (IF) staining
For IF, cells were fixed by 4% paraformaldehyde for 15 min, washed with PBS and blocking buffer (3% FBS, 1% heat-inactivated sheep serum, 0.1% Triton X-100), and thereafter incubated overnight at 4 °C with primary antibodies against Ki-67 (Santa Cruz, Dallas, TX), and fluorescent Alexa Fluor 488- conjugated secondary antibodies (Thermo Scientific, Rockford, IL) were subsequently used for fluorescence detection. The Ki-67 staining intensity was semi-quantified as follows: – for negative staining, ± for very weak staining, + for weak staining and ++ for strong staining.
Gene expression microarray analysis
To investigate the regulatory role of piR-1245 on genome-wide target mRNAs, we treated HCT116 cells with or without piR-1245 antisense, and subsequently performed Affymetrix GeneChip Human gene 2.0 ST arrays and subsequent bioinformatic analysis as described in the Additional file
1: Supplementary Methods.
piRNA target prediction
Based upon piRNA:mRNA sequence complementarity, we used Miranda v3.3a and RNA22 program to search for targets of piR-1245 against all human transcripts. The candidate piRNAs were selected based on prediction scores and binding energy. The whole transcript region of human transcripts were used for piRNA target prediction.
Statistical analysis
All statistical analyses were performed using the GraphPad Prism version 6.0 or MedCalc version 12.3 programs. Statistical differences between groups were determined by Wilcoxon’s signed rank test, the χ2 test or Mann-Whitney U test. Kaplan-Meier analysis and log-rank test was used to estimate and compare overall survival (OS) rates of CRC patients with high and low piR-1245 expression. The optimal cutoff values were determined by ROC curves to discriminate patients with or without death. The Cox’s proportional hazards models were used to estimate hazard ratios (HRs) for death. All P values were 2-sided, and those less than 0.05 were considered statistically significant.
Discussion
Colorectal cancer is one of the most common cancers worldwide. Consequently, elucidation of the molecular mechanisms underlying its progression is critical for the development of new diagnostic and prognostic biomarkers, as well as identification of better therapeutic targets for the management of patients with this deadly malignancy. Herein, we for the first time performed systematic piRNA expression profiling, and identified piR-1245, as a novel oncogenic piRNA mediating CRC pathogenesis. We have made several novel observations in this study. First, we discovered that piR-1245 is frequently overexpressed in CRC tissues from different cohorts, and its overexpression associated with several known risk clinicopathological factors including tumor differentiation and metastasis. Second, our data revealed that patients with high expression of piR-1245 had shorter overall survival, highlighting its applicability as a promising prognostic biomarker in CRC. Third, ours is the first study to demonstrate the biological relevance of this piR-1245 as a tumor-promoting noncoding RNA in CRC. Fourth, microarray analysis revealed thatpiR-1245 regulates several key cancer pathways, supporting its oncogenic role in CRC. Finally, we discovered several important tumor suppressors as direct downstream gene targets, and their expression was inversely correlated with the piR-1245, suggesting this small noncoding RNA promotes CRC development through inhibition of these target genes at the transcriptional level.
In contrast to the growing body of studies that underpin the miRNA-cancer connection, knowledge of piRNAs in tumorigenesis, particularly in CRC, remains currently in its infancy. PiRNAs are roughly 26-30 nucleotides in length and associate specifically with Argonaute proteins that belong to the PIWI subfamily [
40,
41]. Although previously considered to be germline specific and guardians for protecting the integrity of the genome against transposon-induced insertional mutations [
42], mounting evidence now point towards novel active role of piRNAs in somatic gene regulation, through other mechanisms such as transcriptional gene silencing and sequence-specific DNA methylation [
16,
40,
41]. Interestingly, recent studies have demonstrated that piRNAs are widely expressed and play important roles in somatic cells [
17]. Furthermore, few studies have begun to investigate the differentially expressed piRNAs in human cancers and their benign counterparts. It is noteworthy that piR-651 and piR-823 were found to be dysregulated in gastric cancer [
43,
44], and moreover, recent study showed a panel of piRNAs are associated with prognosis in breast cancer [
28], suggesting that piRNAs which previously considered as “junk” RNAs, are indeed involved in tumor progression and could be used as clinically-relevant biomarkers.
Until now, there are limited studies reporting the functional or clinical significance of piRNAs in CRC. We noted Cheng, et al. revealed piR-651 was overexpressed in several types of cancers including CRC [
44]. However, the clinical and biologic significance of this piRNA in CRC remains unknown. In this study, through small RNA-seq analysis, we identified another specific piRNA, the piR-1245, which was consistently overexpressed in colorectal cancer tissues across different cohorts, highlighting its important role in CRC development. Notwithstanding its overexpression in cancer, we discovered that piR-1245 is a promising cancer biomarker, since its overexpression correlated with known risk clinicopathological features such as tumor depth, tumor differentiation and metastasis. Furthermore, another major finding of our study was that piR-1245 was a robust prognostic biomarker for survival prediction in CRC patients. These findings may help provide a better understanding of the mechanisms of piRNA in cancer progression and metastasis in CRC, and suggest that this novel small RNA may be an important disease biomarker and a potential therapeutic target in this disease.
To fully appreciate the clinical significance of piR-1245 in CRC, its biological significance as a contributor to colorectal pathogenesis should also be considered. Our functional experiments provide convincing evidence to support for the associations of piR-1245 with an aggressive clinical phenotype, where piR-1245 promotes CRC cells survival, migration and invasion as well as suppression of apoptosis. Consistent with this paradigm, our gene expression profiling results revealed that piR-1245 affects cancer-related pathways and functions as an oncogenic regulator. Accordingly, our results successfully proved our hypothesis, whereby overexpression of piR-1245 affected gene regulatory network for CRC and resulted in an aggressive phenotype, both biologically and clinically.
To further decipher the mechanic role of piR-1245 in CRC, we interrogated its potential downstream gene targets. By using bioinformatics approach, we identified nine ‘functionally relevant’ cancer-related genes. Interestingly, these nine candidates are involved in key tumor suppressive pathways and their expression inversely correlated with piR-1245 expression, supporting the oncogenic role of piR-1245 in CRC. Surprisingly, piR-1245 was found to not only bind to the exonic regions but also within the intronic regions. A recent study reported that piRNAs are able to bind to pre-mRNA introns and subsequently lead to the decay of targeted pre-mRNA through nuclear exosomes [
38], suggesting that piR-1245 may use a similar mechanism to downregulate the expression of target genes. Furthermore, Watanabe, et al. suggested that piRNAs may suppress expression level of mRNAs harboring transposon sequence in 3’UTR or 5’UTR region [
40]. Besides, piRNAs may also serve as natural antisense molecules that target genes by binding to their CDS regions or function as siRNAs to target 3’UTR [
39‐
42]. In our study, we observed that piR-1245 could target 3’UTR, CDS or 5’UTR region via perfect or imperfect base-pairing between the two types of RNAs, by a mechanism that closely resembles that of antisense or siRNA. Although a number of possible scenarios could account for the interaction between piR-1245 and its target mRNAs, our data clearly demonstrated that the expression of these targets was significantly altered following gain or loss of piR-1245 expression in CRC cell lines.