MicroRNAs predict and modulate responses to chemotherapy in colorectal cancer

Abstract

Colorectal cancer (CRC) is the fourth leading cause of cancer‐related death globally. Chemotherapy regimens consisting of 5‐fluorouracil (5‐FU) in combination with either oxaliplatin or irinotecan are the first‐line options for treatment of metastatic CRC. However, primary or acquired resistance to these chemotherapeutics is a major clinical challenge. MicroRNAs (miRNAs) are a group of small non‐coding RNAs that regulate gene expression post‐transcriptionally. miRNAs play important roles in many cancer‐related processes, including cell proliferation, apoptosis and invasion, and their dysregulation is implicated in colorectal tumourigenesis. Pertinent to chemotherapy, increasing evidence has revealed that miRNAs can be directly linked to chemosensitivity in CRC. In this review, we summarize current evidence concerning the role of miRNAs in prediction and modulation of cellular responses to 5‐FU, oxaliplatin and irinotecan in CRC. We also discuss the possible targets and intracellular pathways involved in these processes.

Introduction

Colorectal cancer (CRC) is the third most common cancer and the fourth leading cause of cancer‐related death in the world 1. Its incidence in many developing countries has increased 2‐ to 4‐fold over the last two decades and has now reached an alarming level 2. Prognosis of CRC is heterogeneous with 20–25% of patients presenting with metastases at diagnosis, and 50–60% of the remainder eventually developed disseminating disease 3. Localized CRC lesions can be removed surgically but half CRC patients survive <5 years subsequently, due to metastasis (principally) to the liver and lungs. Numbers of regimens, including FOLFOX (5‐fluorouracil [5‐FU], leucovorin and oxaliplatin), FOLFIRI (5‐FU, leucovorin and irinotecan) and XELOX (capecitabine and oxaliplatin) with or without cetuximab (an epidermal growth factor receptor‐directed monoclonal antibody) or bevacizumab (a vascular endothelial growth factor‐targeting monoclonal antibody), are the first‐line chemotherapeutic options for metastatic CRC 4. However, primary or acquired resistance is a major challenge to its successful treatment 5. Emerging evidence has shed new light on molecular mechanisms underlying resistance to commonly used chemotherapeutics. Understanding of these molecular events is key to improving prediction of treatment response and guiding treatment decisions in patients with metastatic CRC.

MicroRNAs (miRNAs) are a group of small (19–25 nucleotides), non‐coding RNAs that regulate gene expression post‐transcriptionally 6, 7. Through partial or complete binding to the 3′ untranslated region (3′‐UTR) of their target mRNAs, single‐stranded miRNAs can lead to blocking translation or mRNA degradation. miRNAs play significant roles in many biological and pathophysiological processes, including development, host defence, aging and tumourigenesis 7, 8, 9, 10. Most importantly, they exhibit differential expression in many, if not all, types of cancers, including CRC 6. Aberrant expression of miRNAs has been observed in many malignancies, where they function as either onco‐miRNAs or tumour‐suppressor miRNAs 11. Considering their extensive interactions with intracellular signalling networks, miRNAs regulate many cancer‐pertinent cellular processes, including cell proliferation, apoptosis, migration, invasion, stemness and chemoresistance 6, 7. Importantly, a subset of miRNAs may serve as diagnostic or prognostic biomarkers. With further advancement of RNA delivery technology, it is also anticipated that novel miRNA‐based therapeutics will emerge 12.

In the present review, we summarize results from the published literature on the role of miRNAs in prediction and response modification to three conventional chemotherapeutics, namely 5‐FU, oxaliplatin and irinotecan, in CRC. We also discuss possible involved molecular targets and intracellular pathways.

5‐fluorouracil and capecitabine

The fluoropyrimidine 5‐FU is an anti‐metabolite that exerts its anti‐cancer effects through inhibition of thymidylate synthase, and incorporation of its metabolites into RNA and DNA 13. Capecitabine was initially developed as a prodrug of 5‐FU with improved tolerability that could achieve higher intratumoural concentrations through tumour‐specific metabolic activation 14.

Drug effects on miRNA expression

Effects of 5‐FU on miRNA expression were determined in C22.20 and HC.21 colon cancer cell clones. Seventeen up‐regulated (miR‐19a, miR‐20, miR‐21, miR‐23a, miR‐25, miR‐27a/b, miR‐29a, miR‐30e, miR‐124b, miR‐132, miR‐133a, miR‐141, miR‐147, miR‐151, miR‐182, miR‐185) and three down‐regulated (miR‐200b, miR‐210, miR‐224) miRNAs were identified in both lines after exposure to 5‐FU 15. 5‐FU also dose‐dependently reduced primary transcript levels of miR‐17‐92 cluster in human colon cancer KM12C cells 16. Akao et al. reported that 5‐FU exposure stimulated miR‐145 and miR‐34a expression in DLD‐1 cells 17 and miR‐19b and miR‐21 were found to be overexpressed in 5‐FU‐resistant CRC cells 18. The effect of capecitabine in combination with radiotherapy on miRNA expression was also investigated in tumour biopsies from patients with rectal cancer before and after therapy. Two miRNAs, miR‐125b and miR‐137, had consistent increase in expression levels 19.

Autophagy has been increasingly recognized as a pro‐survival cellular response activated in times of stress during chemotherapy 20. Hou et al. reported that 5‐FU down‐regulated four miRNAs (miR‐302a‐3p, miR‐548ah‐5p, miR‐133b, miR‐323a‐3p) and up‐regulated 27 miRNAs (miR‐203a, miR‐99b‐5p, miR‐195‐5p, let‐7c‐5p, miR‐320d, miR‐301a‐3p, miR‐30e‐5p, miR‐374c‐5p, miR‐181a‐5p, let‐7g‐5p, miR‐513b‐5p, miR‐30b‐5p, miR‐19b‐3p, miR‐19a‐3p, miR‐15a‐5p, miR‐106b‐5p, miR‐330‐3p, miR‐582‐5p, miR‐16‐5p, miR‐30a‐5p, miR‐23a‐3p, miR‐26b‐5p, miR‐98‐5p, miR‐186‐5p, miR‐30d‐5p, miR‐93‐5p, miR‐320c) that had the predicted target genes involved in regulation of autophagy in HT‐29 cells. These miRNAs also exhibited concordant alteration of expression upon starvation, in the same cell line 21.

Modulation of chemosensitivity

Modulation of cellular responsiveness to 5‐FU by chemosensitizing and chemoresistant miRNAs is illustrated in Fig. 1. miR‐494 was found to be down‐regulated in 5‐FU‐resistant SW480 cells compared to the parental cell line, whereas ectopic expression of miR‐494 increased sensitivities of CRC cell lines and chemoresistant SW40 xenografts to 5‐FU. Mechanistically, miR‐494 targeted DPYD, required for miR‐494‐mediated regulation of 5‐FU sensitivity 22 and likewise, miR‐22‐mediated inhibition of autophagy by targeting B‐cell translocation gene 1 (BTG1) is implicated in enhanced chemosensitivity to 5‐FU in CRC cells 23.

Figure 1.

Modulation of 5‐FU chemosensitivity by chemosensitizing and chemoresistant miRNAs in CRC.

Chemoresistance has been linked to epithelial–mesenchymal transition (EMT). To this end, up‐regulation of a group of EMT‐suppressive miRNAs (miR‐200b, miR‐200c, miR‐141, miR‐429 and miR‐101) has been implicated in the chemosensitizing effect of curcumin (a botanical substance with claimed anti‐tumourigenic properties) in 5‐FU‐resistant CRC cells 24. Restored expression of miR‐320, a tumour‐suppressive miRNA down‐regulated in CRC, also sensitized CRC cells to 5‐FU and oxaliplatin. In this regard, miR‐320 was shown to target Forkhead box M1 (FOXM1) that has an established role in chemoresistance in various types of tumour 25. miRNA‐497 levels correlated with 5‐FU sensitivity in CRC cells in which restoration of miRNA‐497 enhanced 5‐FU sensitivity. miRNA‐497 targeted Smurf1 whose expression levels were dramatically increased in 5‐FU‐resistant patients compared to treatment‐sensitive patients 26. Transfection of stemness‐related miRNAs (miR‐302a‐d, miR‐369‐3p and 5p, miR‐200c) has been shown to reprogramme CRC cells and enhance their sensitivity to 5‐FU probably through down‐regulating MRP8 that mediates cellular efflux of the toxic metabolite of 5‐FU 27. Ectopic expression of p53‐target miR‐34a sensitized CRC cells to 5‐FU by targeting lactate dehydrogenase A, c‐Kit, Sirt1 and E2F3 28, 29, 30, whereas inhibition of let‐7a conferred resistance to 5‐FU and radiation in CRC cells with single or double wild‐type TP53 alleles 31. miR‐142‐3p also elevated sensitivity of CRC cells to 5‐FU, which was paralleled by down‐regulation of CD133, Lgr5 and ABCG2 32. Restoration of miR‐143 and miR‐365 expression also promoted 5‐FU‐induced cell death, an effect probably mediated by repression of their common target Bcl‐2 (an anti‐apoptotic protein) 33, 34. Besides Bcl‐2, miR‐143 reduced extracellular‐regulated protein kinase 5 and nuclear factor‐κB 34. Aside from abovementioned miRNAs, miR‐96, miR‐122, miR‐129, miR‐145, miR‐203 and miR‐497 have been reported to enhance chemosensitivity to 5‐FU in CRC through down‐regulating XIAP, PKM2, Bcl‐2, RAD18, thymidylate synthase and IGF1‐R respectively 35, 36, 37, 38, 39, 40. miR‐145 may also target the Friend leukaemia virus integration 1 gene (FLI1) in CRC 41.

miRNA‐23a and miR‐31 conferred resistance to 5‐FU in CRC cells in which the former targeted ABCF1 and apoptosis‐activating factor‐1 (APAF‐1) 42, 43, 44. miR‐21 also increased resistance of tumour cells to 5‐FU in CRC by directly targeting hMSH2, an important DNA mismatch repair gene 45, 46. Inhibition of miR‐21 also promoted cell differentiation accompanied by enhanced sensitivity to 5‐FU and oxaliplatin in CRC, in which direct targeting of the tumour suppressor gene sprouty2 has been demonstrated 47, 48. miR‐10b overexpression conferred chemoresistance to 5‐FU by targeting pro‐apoptotic BIM 49 and expression of miR‐520g correlated with 5‐FU resistance of CRC cells. Ectopic expression of miR‐520g attenuated 5‐FU‐induced apoptosis in vitro and reduced effectiveness of 5‐FU in inhibition of CRC tumour growth, in a mouse xenograft model. It has been proposed that miR‐520g mediates 5‐FU resistance by down‐regulation of p21 50, but nevertheless in the literature, contradictory evidence exists concerning the role of p21 in cellular responses to 5‐FU. For instance, antagonizing endogenous miR‐192 has been shown to attenuate 5‐FU‐induced accumulation of p21, thereby shifting 5‐FU's effect towards apoptosis 51.

In addition to direct inhibition of specific target genes, miRNAs have been shown to influence 5‐FU sensitivity by non‐specific mechanisms. In this respect, miR‐140 and miR‐192 increase resistance to 5‐FU probably by reducing S‐phase cells and thereby mitigating effects of S phase‐specific drugs, including 5‐FU 52, 53. miR‐224 is also involved in chemoresistance to 5‐FU in CRC where its knock‐down phenocopied KRAS mutation by increasing KRAS activity with ERK and AKT phosphorylation increasing 5‐FU chemosensitivity 54.

Response prediction

Polymorphisms in two microRNA genes (miR‐608 and miR‐219‐1) have been reported to be predictors of clinical outcomes in CRC patients receiving 5‐FU‐based chemotherapy. In this regard, carriers of the variant T allele of rs213210 in miR‐219‐1 had shorter overall survival, whereas variant G allele of rs4919510 in miR‐608 was associated with reduced risk of recurrence. No such associations were found in the group of CRC patients not treated with 5‐FU‐based chemotherapy 55. Variations in miRNA‐binding sites in BER gene 3′‐UTR also modulate CRC response to 5‐FU‐based chemotherapy 56.

Stage II/III CRC patients receiving 5‐FU‐based chemotherapy with high tissue levels of miR‐200a, miR‐200c, miR‐141 or miR‐429 had significantly longer overall and disease‐free survival 57. In contrast, high levels of miR‐125b and miR‐137 were associated with worse response to capecitabine‐based chemoradiotherapy 19. Svoboda et al. also reported that miR‐215, miR‐190b and miR‐29b‐2* were overexpressed in non‐responders, while let‐7e, miR‐196b, miR‐450a, miR‐450b‐5p and miR‐99a* had higher expression levels in responders to 5‐FU‐ or capecitabine‐based chemoradiotherapy 58. Low miR‐148a or high miR‐320e expression also correlated with a poorer response to 5‐FU‐based chemotherapy in advanced CRC patients 59, 60. High miR‐320e was associated with adverse clinical outcome in stage III CRC patients treated with 5‐FU‐based adjuvant chemotherapy 59. miR‐21 had predictive significance of pathological response to 5‐FU‐based neoadjuvant chemoradiotherapy in locally advanced rectal cancer patients, yielding an area under the curve value of 0.78 with 86.6% sensitivity and 60.0% specificity, in distinguishing rectal cancer from complete response to non‐complete response 61. let‐7g and miR‐181b levels were strongly associated with CRC patients' response to the 5‐fluorouracil‐based anti‐metabolite S‐1 62.

Circulating miRNAs have been utilized to predict CRC patients' responses to chemotherapy. Changes in circulating miRNA‐126 levels during treatment were predictive of tumour response in metastatic CRC patients treated with capecitabine and oxaliplatin combined with bevacizumab. Non‐responding patients had median increases in circulating miRNA‐126 of 0.244 compared to median reduction of −0.374 in responding patients 63. Aberrant elevation of a further circulating miRNA, miR‐19a, also predicted non‐responders to FOLFOX chemotherapy in advanced CRC 64. Re‐elevation or sustained elevation of serum miR‐155 levels after surgery and chemotherapy also foreshadowed chemoresistance in CRC patients treated with 5‐FU and leucovorin plus cetuximab 65. In contrast, reduction in blood miR‐296 predicted chemotherapy resistance and poor clinical outcome in patients receiving capecitabine and sunitinib 66.

Oxaliplatin

Oxaliplatin is a diaminocyclohexane‐containing platinum‐based anti‐neoplastic agent that exerts its cytotoxic action through diverse mechanisms, including induction of DNA lesions through adduct formation, arrest of DNA synthesis, inhibition of mRNA synthesis and triggering of immunological reactions 67.

Drug effects on miRNA expression

Zhou et al. profiled miRNA expression in two CRC cell lines (HCT‐8 and HCT‐116) exposed to oxaliplatin. Although oxaliplatin altered expression of multiple miRNAs in respective cell lines, only miR‐151‐5p and miR‐1826 were found to be commonly down‐regulated 68. One further study identified miR‐203 as the sole up‐regulated miRNA in all three tested oxaliplatin‐resistant CRC cell lines compared to their parental lines 69.

Modulation of chemosensitivity

Kalimutho et al. compared mechanisms of tumour sensitivity to satraplatin versus oxaliplatin and found that loss of p53‐mediated miR‐34a expression increased resistance to oxaliplatin but not to satraplatin 70. miR‐520g, a miRNA negatively regulated by p53, also conferred resistance to oxaliplatin by targeting p21 in CRC cells 51. By microRNA expression array and quantitative reverse transcription‐PCR, Chai et al. found that chemoresistant SW620 cells had elevated miR‐20a expression compared to chemosensitive SW480 cells. Knock‐down of miR‐20a sensitized SW620 cells to oxaliplatin, whereas its overexpression conferred resistance to the drug in SW480 cells. BNIP2, a Bcl‐2‐interacting partner, has been confirmed to be the direct target of miR‐20a 71. Ectopic expression of miR‐203 induced oxaliplatin resistance in CRC cells, whereas miR‐203 inhibitor produced the opposite effect. Such actions were mediated through its negative regulation of ataxia telangiectasia mutated (ATM), a primary mediator of DNA damage response 69. LIN28B, a negative regulator of the let‐7 microRNA family, also conferred resistance to oxaliplatin in SW480 and HCT116 colon cancer cells 72.

Response prediction

A set of 13 miRNAs (up‐regulated: miR‐1183, miR‐483‐5p, miR‐622, miR‐125a‐3p, miR‐1224‐5p, miR‐188‐5p, miR‐1471, miR‐671‐5p, miR‐1909∗, miR‐630, miR‐765; down‐regulated: miR‐1274b, miR‐720) is strongly associated with complete pathological response in rectal cancer patients receiving oxaliplatin–capecitabine and radiotherapy before surgery. Of these, miR‐622 and miR‐630 had a 100% sensitivity and specificity in selecting responsive cases 73. miRNA‐126, the only known endothelial cell‐specific miRNA that may be used as a surrogate marker of angiogenesis, has also been shown to predict response to oxaliplatin and capecitabine in patients with metastatic CRC 74, 75. In KRAS‐mutated tumours, increased miR‐200b and reduced miR‐143 expression were associated with better progression‐free survival in patients treated with oxaliplatin in combination with capecitabine, cetuximab and bevacizumab 76. High miR‐320e expression was associated with worse therapeutic response in advanced CRC patients treated with oxaliplatin‐ and 5‐FU‐based chemotherapy 59, whereas high expression of miR‐196b‐5p and miR‐592 predicted improved outcome of XELOX regimen (oxaliplatin and capecitabine) with or without bevacizumab 77. Five serum miRNAs (miR‐20a, miR‐130, miR‐145, miR‐216 and miR‐372) that were significantly up‐regulated in oxaliplatin‐chemoresistant CRC patients also predicted chemotherapeutic response with areas under the receiver operating characteristic curve of 0.841 and 0.918 in two independent cohorts 78.

Irinotecan

Irinotecan, also known as CPT‐11, is a derivative of camptothecin. It prevents religation of the DNA strand by inhibiting action of topoisomerase I, thereby causing double‐strand DNA breakage that leads to apoptosis or premature senescence 79.

Drug effects on miRNA expression

Cellular changes in miRNA expression upon irinotecan treatment in CRC remain largely uncharacterized. To this end, only the effect of irinotecan on expression of senescence‐related miRNAs in normal colonic fibroblasts or normal colonocytes has been investigated. In this regard, strong up‐regulation of miR‐34a, miR‐128a and miR‐449a has been associated with premature senescence in irinotecan‐treated colonic fibroblasts. In contrast, irinotecan has been found to only moderately increase expression of miR‐34a, but had no significant effect on the other two miRNAs, in normal colonocytes, which underwent apoptosis upon treatment 80.

Modulation of chemosensitivity

Evidence concerning effects of miRNA on responsiveness to irinotecan in CRC remains scarce. Chemoresistance is a common property of cancer stem cells. miR‐451 has been found to be down‐regulated in colonspheres with CRC stem cell properties compared to parental cells. Concordantly, restored expression of miR‐451 caused a reduction in self‐renewal, tumourigenicity and chemoresistance to irinotecan 81.

Response prediction

miRNA expression levels, or polymorphisms of miRNA genes or miRNA‐binding sites on targets, have been shown to predict response to irinotecan‐based chemotherapy. In metastatic CRC patients treated with salvage cetuximab–irinotecan, Graziano et al. reported that G‐allele carriers of the let‐7 complementary site in the KRAS 3′‐UTR, displayed worse overall survival (P = 0.001) and progression‐free survival (P = 0.004) than T/T genotype carriers. Multivariate analysis revealed that prognostic significance of this variant was independent of other clinicopathological parameters, including KRAS mutation status 82. A subsequent study by the same group further revealed that higher let‐7a levels were significantly associated with better survival outcomes in this cohort of patients 83. For miRNA genes, Boni et al. found that single‐nucleotide polymorphisms (SNPs) rs7372209 and rs1834306 in pri‐miR26a‐1 and pri‐miR‐100 genes, respectively, were associated with tumour response and/or time to progression in metastatic CRC patients treated with irinotecan and 5‐FU 84. In metastatic CRC patients receiving cetuximab or panitumab (another epidermal growth factor receptor‐directed monoclonal antibody) with or without irinotecan, low levels of miR‐31* and high levels of miR‐592 in tumour tissues, favoured disease control over progressive disease 85. A recent study further reported that high circulating levels of miR‐345 were associated with lack of response to treatment with cetuximab and irinotecan in metastatic CRC 86.

Concluding remarks and future perspectives

miRNAs are a class of small, non‐coding RNA molecules regulating gene expression at post‐transcriptional levels. Accumulating evidence has implicated that miRNAs could be used to predict clinical outcomes of treatment, as well as modulate efficacy of anti‐cancer chemotherapy. In our review, alterations in miRNA expression profiles have been found to predict the success of 5‐FU‐, oxaliplatin‐ and irinotecan‐based chemotherapy in CRC. Furthermore, miRNAs seem to regulate cancer cell sensitivity to these drugs. Thus, combining miRNAs with existing chemotherapeutic agents might be used to maximize therapeutic effect and improve clinical outcomes in patients with metastatic CRC. However, detailed mechanisms and intracellular pathways in miRNA regulation of chemosensitivity in CRC remain largely unspecified. Further investigations are needed for systematic identification of miRNAs involved in modulation of chemosensitivity as well as their associated molecular targets and signalling pathways. In addition, many functional studies have been conducted with CRC cell lines in vitro with limited numbers of studies reporting results from mouse xenograft models. There is also no study as yet on safety and efficacy of miRNA‐based treatment in humans; in time, this will be investigated.