Influence of microRNAs and Long Non-Coding RNAs in Cancer Chemoresistance

Innate and acquired chemoresistance exhibited by most tumours exposed to conventional chemotherapeutic agents account for the majority of relapse cases in cancer patients. Such chemoresistance phenotypes are of a multi-factorial nature from multiple key molecular players. The discovery of the RNA interference pathway in 1998 and the widespread gene regulatory influences exerted by microRNAs (miRNAs) and other non-coding RNAs have certainly expanded the level of intricacy present for the development of any single physiological phenotype, including cancer chemoresistance. This review article focuses on the latest research efforts in identifying and validating specific key molecular players from the two main families of non-coding RNAs, namely miRNAs and long non-coding RNAs (lncRNAs), having direct or indirect influences in the development of cancer drug resistance properties and how such knowledge can be utilised for novel theranostics in oncology.


Introduction
The discovery of the RNA interference pathway in 1998 and the widespread gene regulatory influences exerted by microRNAs (miRNAs) and other non-coding RNAs (ncRNAs) have certainly expanded the level of intricacy present for the development of any single physiological phenotype [1]. Such phenotypes can include clinical conditions of either an acute or a chronic nature. Undoubtedly, cancer best symbolizes clinical conditions relying on multifactorial influences for its development. Furthermore, the clinical presentation of any specific cancer condition in the patient can vary to great extents, depending on multiple tumour characteristics such as the degree of invasiveness, aggressiveness and angiogenesis. However, one of the most crucial cancer phenotypes that pose a major challenge to current conventional chemotherapeutic measures is the ability of the tumour to withstand the pharmacological effects of multiple cancer chemotherapy drugs, typically described as chemoresistance.
Since the influences of ncRNAs in the main facets of cancer development are described in great detail within the scientific literature, this review specifically places a spotlight on the emerging global research efforts that (in the authors' opinion) most effectively recognize the growing link pertaining to non-coding RNA activities with the regulation of cancer chemoresistance properties [2][3][4][5][6].

Cancer Chemoresistance Manifestations and Development Mechanisms
Tumours bearing a chemoresistance phenotype can irrevocably thwart the prognosis of the cancer patient, particularly when such characteristics evolve in relapse of the disease. This chemoresistance phenotype has two distinct development mechanisms, leading to the existence of innate and acquired chemoresistance phenotypes [7].
Innate chemoresistance refers to the scenario that an individual tumour can inherently possess unique genetic characteristics that render the tumour to withstand single (or multiple) chemotherapeutic agents, through various influences on drug cytotoxicity circumvention pathways, as described below [7].
In the case of acquired chemoresistance properties, the link between such cancer relapse phases and multi-drug resistant (MDR) tumours is predominantly due to the regular exposure of the tumour to conventional chemotherapeutic cyclical administration [7]. This essentially drives the tumour to evolve at the genetic level to a variant having increased withstanding potential against such conventional chemotherapeutic agents [7].
Since this article focuses specifically on the links identified so far between miRNAs/lncRNAs and cancer chemoresistance properties, it is important to highlight the main recognized mechanisms by which tumours can develop multi-drug resistance against conventional chemotherapeutic agents.
By far the most important and characterised mechanism for the emergence of cancer chemoresistance properties is the employment of drug efflux pumps that actively remove multiple drugs from the tumour cell cytoplasm, including conventional chemotherapeutic agents and eventually leading to MDR [8][9][10][11][12]. The key molecular players involved in drug efflux processes are primarily the ATP-dependent binding cassette (ABC) transporters such as ABCG2, ABCB1 (multidrug resistance 1 gene/P-glycoprotein) and ABCC1 (multidrug resistance-associated protein 1) [8][9][10][11][12]. Another important mechanistic branch leading to chemoresistance is the dysfunction or loss of p53-mediated apoptotic pathways typically triggered by DNA damage, with examples being dysfunctional activity of the mouse double minute 2 gene (Mdm2) and the p53 encoding gene (TP53) [13][14][15]. In a similar manner, other pro-apoptotic pathways that are typically triggered by cytotoxic drug activities can be hindered within chemoresistant tumour cells. Such pathway issues include cellular FADD-like interleukin 1 beta converting enzyme-inhibitory protein (c-FLIP) and the Bcl-2 protein family members [16][17][18][19]. Alternatively, triggering of proliferative/survival signalling pathways such as the ERK and PI3K pathways by means of protein tyrosine kinases, sirtulins, transcription factor kappa B (NFκB) or epidermal growth factor receptor (EGFR) family members can also lead to chemoresistance phenotypes within tumours [20][21][22]. Furthermore, increased efforts by key molecular components of the nucleotide excision repair pathway can take place as a means of limiting tumour cell DNA damage by cytotoxic drug activity [23][24][25]. Other mechanisms directing chemoresistance phenotypes in tumours include drug modulation through inactivation or attenuation of cytotoxic drug activity, modification of drug targets and inhibition of tumour suppressor genes that trigger DNA methylation pathways [26][27][28].
In essence, a particular tumour can be clinically recognized as MDR through the identification of a unique spectrum of dysregulated expression patterns of multiple mRNA biomarkers [29-33]. However, the discovery of the miRNA and long non-coding RNA (lncRNA) families have created additional layers of genomic regulatory functions [2,[34][35][36][37][38][39]. Further insight into the roles conducted by miRNAs and lncRNAs in the development of innate and/or acquired chemoresistance properties by tumours is expected to lead to a more accurate depiction of the cancer patient's condition-both at diagnosis and during possible relapse condition.
The sections below represent a comprehensive summary of the global research efforts in delineating the influences and main mechanistic links of such non-coding RNA families on this specific tumour characteristic. Ultimately, the goal of developing novel theranostic measures for employment in the oncology clinic setting can be attained.

Influences of miRNAs in Cancer Chemoresistance
Each physiologically active, individual miRNA consists of a 19-22 nucleotide RNA duplex, bearing a guide strand that is fully or partially complementary to the target transcript to which it binds [40]. This binding, which is typically non-totally complementary, leads to either mRNA target cleavage or a hindering effect on the translational phase of protein synthesis by the ribosomal infrastructure. This has the ultimate effect of a reduction in protein level production for the affected target transcript, or effectively post-transcriptional gene regulation [40].
In essence, over 10% of all reported research findings of miRNA influences in cancer chemoresistance demonstrated that such a phenotype modulation was effected through simultaneous dysregulation of multiple miRNAs, rather than an individual putative chemoresistance miRNA-though no specific miRNA combination was identified as most prevalent by the authors [44,[46][47][48][49]62,67,71,76,81,85,88,132,133,158,168,201,203,[211][212][213]217,223,224,228,230]. This suggests that miRNA influences on cellular functions also occur at a more complex level, whereby it is a signature dysregulated expression pattern of two (or more) miRNAs that trigger simultaneous downregulation of a specific set of target transcripts leading to an ultimate change in cellular/tissue phenotype. Further evidence for this can be concluded by previous efforts carried out by the authors, whereby a set of seven putative chemoresistance miRNAs were identified for neuroblastoma-a paediatric cancer model [232].
The mechanistic links between miRNA activity and chemoresistance development that have been identified so far reveal that miRNA dysregulated expression can indeed influence the main cellular pathways (described in Section 2 above) that are directly affecting chemoresistance emergence in tumour models.
In the first instance, miRNAs have been recognized to regulate MDR-related molecular players such as multidrug resistance-associated protein 1 (MRP-1) [66]. The study carried out by Gao and colleagues identified the gene regulatory effects of miR-145-5p on MRP-1, ultimately enhancing the level of chemosensitivity in the doxorubicin resistant MCF-7 breast cancer cell line [66]. Furthermore, the study performed by Zhan and colleagues identified that MRP-1 could also be downregulated through the overexpression of miR-145-5p in cisplatin-resistant gallbladder tumour models, resulting in sensitizing of the tumour cells to cisplatin activity [99]. Similarly, the study conducted by Zhao and colleagues revealed that a combination of four miRNAs, namely miR-302a-3p/b-3p/c-3p/d-3p, concomitantly provide similar effects in doxorubicin-resistant breast cancer MCF-7 cell lines [71]. This study confirmed through RT-qPCR and Western blotting techniques that over-expression of these miRNAs induced sensitization to doxorubicin through a reduction in the expression level of MAP/ERK kinase 1 (MEKK1), leading to an overall downregulation of P-glycoprotein (P-gp) expression [71].
Other studies have demonstrated miRNA influences on the p53 pathway status as a means of inducing chemoresistance properties. The study conducted by Shen and colleagues recognized the effect of up-regulated miR-29a-3p in exacerbating doxorubicin-resistance in breast cancer cell lines, through its regulatory function on PTEN and GSK3β, that are two major components of the PTEN/AKT/GSK3β signalling pathway providing feedback to TP53 [55]. Further evidence for such miRNA influences on p53 modulation include the study carried out by Qin and colleagues, who revealed that the over-expression of miR-182-5p can induce cisplatin chemoresistance in hepatocellular carcinoma HepG2-Rcells [130]. This exacerbation was found to occur through the direct regulatory effect of the miRNA on tumour protein 53-induced nuclear protein 1 (TP53INP1) [130].
Evidence for the mechanistic link between miRNA activity and apoptotic pathway dysfunction for the emergence of chemoresistance in cancer includes the study performed by Zhang and colleagues [89]. The outcome of this investigation identified the exacerbating effect of miR-425-5p on chemoresistance in colorectal cancer HCT116 cell lines as well as in xenograft models, through direct action on programmed cell death 10 (PCD10) [89]. In addition, Stojcheva and colleagues reported the acquisition of chemoresistance properties by glioblastoma to alkylators such as temozolomide due to miR-138-5p direct regulatory function on BIM, which is a Bcl-2 interacting mediator of apoptosis [111]. Furthermore, miR-182-5p was also identified as playing a key role in the development of chemoresistance in ovarian carcinomas due to its gene-regulating capacity on programmed cell death 4 (PDCD4) [214].
The influences of miRNA activity on cell proliferative function, as another mechanism for chemoresistance emergence, is described through the study performed by Ye and colleagues on chemoresistant breast cancer cell line models [59]. This seminal investigation recognized the downregulatory effect of miR-484 on cytidine deaminase (CDA), which is a major molecular player in controlling cell proliferation through its suppressive function on cyclin E-CDK2 signalling, thereby inhibiting any cell-cycle progress [59]. Interestingly, the investigation conducted by Phatak and colleagues on oesophageal squamous cancer cell line models revealed that overexpression of miR-214-3p resulted in the sensitization of such tissue cultures to cisplatin [171]. Such a reduction in chemoresistance by miR-214-3p was mediated by direct regulation of surviving expression, together with indirect regulation by means of downregulation of CUG-BP1, leading to decreased mRNA stability in the targeted tumour cells [171].
Cytotoxic drug-induced DNA damage response pathway was also found to be affected by miRNA influence. The study performed by Li and colleagues identified that such a pathway can be activated by upregulation of the inhibitor of growth 5 (ING5) gene, which is in turn regulated by miR-193a-3p within the bladder cancer cell line 5637 [49]. In addition, inhibition of ING5 in this study resulted in a drastic reduction of the DNA damage response pathway within the same cell line [49].
Evidence for the modulation of cytotoxic drug targets by miRNAs can be reflected in the study performed by Liang and colleagues on gemcitabine resistant MiaPiaCa-2 pancreatic cell cancer models [216]. Results of this study led to the regulatory role of miR-33a in β-catenin downregulation, with the latter being a key molecular player in directing the expression levels of multiple genes including cyclin D1, surviving and MDR-1 [216].
In addition to the effect of multiple miRNAs on cancer chemoresistance, several studies reported the possible influences exhibited by circulating miRNAs (in blood), typically through extracellular vesicle or exosome transport [67,168,172,205,228].
Exosomes can be characterized as endocytic vesicles that can be secreted by multiple cell types within the human body, carrying a spectrum of molecular players, including miRNAs, for the purpose of cell-cell communication (e.g., antigen presentation) [233]. The transfer of genomic components such as mRNA and miRNAs between two individual cells of varying morphology, such as from bone marrow to mast cells, was also identified to lead to novel protein synthesis within the recipient cells [233]. Consequently, exosomal transfer can be reliably considered as an additional method of influence on multiple molecular pathways, since the desired physiological effect/s is induced from cellular populations situation within remote locations.
The effect of exosome activity can also be linked within the context of cancer chemoresistance, such as the findings described by Chen and colleagues [67]. This study highlighted the effect of exosomes on chemoresistant breast cancer cell lines, focusing on the transfer of miR-222-3p from doxorubicin resistant MCF-7 cell line to a chemosensitive MCF-7 cell line model [67]. This exosomal transfer ultimately rendered the recipient breast carcinoma cell line more resistant to doxorubicin activity following post-transfer functional assays [67].
Furthermore, the study carried out by Au Yeung and colleagues identified and validated the effect of exosomal transfer of miRNAs in drug resistant ovarian cancer models [205]. This study identified the exosomal transfer of miR-21-5p from cancer-associated adipocytes (within the omental stroma) into ovarian carcinoma cell populations [205]. This transfer ultimately conferred paclitaxel chemoresistance properties to the recipient ovarian carcinoma cells due to the direct regulatory effect of miR-21-5p on the transcript for apoptotic protease activating factor 1 (APAF1) [205].
Exosome transfer was also identified as being implicated in chemoresistance conferring to prostate cancer, as highlighted by the investigation by Li and colleagues [228]. The study identified 29 dysregulated miRNAs within exosomes derived from two paclitaxel resistant prostate cancer cell line models [228].

Influences of lncRNAs in Cancer Chemoresistance
Notwithstanding the myriad of networking interactions leading to miRNA directed gene regulatory effects within cellular populations, the more recent discovery of a separate family of non-coding RNAs, namely lncRNAs, leads to the identification of an additional level of gene regulation within the human body. According to the latest version of LNCipedia, there are over 60,000 members of the lncRNA family that have been catalogued [234,235]. LncRNAs are non-coding RNA genes of at least 200 nucleotides long. LncRNAs can either act as positive or negative regulators of target gene expression, with this activity being directed either on transcripts originating from the same locus as the lncRNA itself (cis-acting) or directed on target transcripts originating on other loci (trans-acting) [6]. Table 2 and Supplementary Materials Table S2 highlight in detail the currently reported scientific evidence for the influence of multiple lncRNAs on varying cancer models. Interestingly, previous studies have highlighted the detrimental effects of one individual lncRNA, known as homeobox transcript antisense RNA (HOTAIR), having an elevated prevalence within multiple tumour chemoresistance phenotypes [236][237][238].
The investigation carried out by Fang and colleagues focused on the possible effects of HOTAIR on chemoresistance in small cell lung cancer, mainly through knock-down of the lncRNA in chemoresistant and parental cell line models, followed by viability assays [236]. Apart from confirming HOTAIR knock-down with enhanced chemosensitivity of the affected cell lines to doxorubicin, cisplatin and etoposide, the study also recognized the chemoresistance phenotype was additionally linked to increased methylation of homeobox A1 (HOXA1), suggesting HOTAIR influences in affecting such a methylation status [236]. The study also confirmed that HOTAIR inhibition, through short hairpin RNA antagonist employment in murine tumour xenograft models for small cell lung cancer, led to a reduction in tumour growth [236].
In a similar study conducted by Liu and colleagues, HOTAIR expression was discovered to be up-regulated in the cisplatin-resistant A549 lung adenocarcinoma cell line model, with consequent re-sensitisation of the cell line to cisplatin exposure following HOTAIR knock-down [237]. This short interfering RNA (siRNA)-induced HOTAIR knock-down effect was also linked to enhanced cell cycle arrest and apoptosis, together with a reduction in cell proliferation, through control of p21 WAF1/CIP1 expression [237].
Furthermore, the study performed by Ozes and colleagues investigated the possible influences of HOTAIR on ovarian cancer chemoresistance properties, specifically for platinum-based chemotherapeutic agent chemoresistance [238]. The results of the study highlighted exacerbated HOTAIR expression within platinum drug resistant ovarian tumour samples when compared to primary ovarian tumour counterpart samples [238]. The study also revealed that HOTAIR up-regulation allows for prolonged NF-KB expression, leading to extended DNA damage response mechanisms to take place, following platinum-based drug exposure and therefore contributing to the chemoresistance phenotype development [238].
In addition to HOTAIR, another putative chemoresistance lncRNA of particular interest is MRUL (NR_024549), since the chromosomal locus for MRUL is in close proximity to the locus for the Multi Drug Resistance 1 (MDR1) gene-the latter being recognised as the most important gene to induce cancer chemoresistance phenotypes [239]. Such a study highlights the unique properties of lncRNAs in their capacity to perform cis-regulatory functions on neighbouring transcripts of clinical relevance.
Evidence for the regulatory role of lncRNAs in directing both cell proliferative signalling and drug modulation mechanisms affecting chemoresistance can be found in the study performed by Dong and colleagues [240]. This particular study recognised the effect of GAS5 in enhancing apoptosis due to gefitinib activity within innate EGFR tyrosine-kinase inhibitor (TKI) resistant lung adenocarcinomas (A549 cell line), through the gene regulating role of GAS5 on insulin-like growth factor 1 receptor (IGF-1R) [240]. Interestingly, the investigation conducted by Cheng and colleagues identified UCA1 as being upregulated in acquired (non T790M) EGFR-TKI resistant non-small cell lung cancer [241]. UCA1 knockdown assays confirmed that this lncRNA, when downregulated, allowed for increased gefitinib sensitivity and furthermore inhibited AKT/mTOR functions [241].

Conclusions and Perspectives
In essence, it can be stated that ncRNAs do have a place in regulating cancer chemoresistance properties, merely based on the body of evidence described above. Furthermore, the recent scientific literature on this niche research reveals that both miRNAs and lncRNAs have important roles in affecting the main mechanisms currently known to lead to the development of cancer chemoresistance phenotypes (see Figure 1).

Conclusions and Perspectives
In essence, it can be stated that ncRNAs do have a place in regulating cancer chemoresistance properties, merely based on the body of evidence described above. Furthermore, the recent scientific literature on this niche research reveals that both miRNAs and lncRNAs have important roles in affecting the main mechanisms currently known to lead to the development of cancer chemoresistance phenotypes (see Figure 1). Undoubtedly, the recent progress in molecular analytical and sequencing technologies has advanced to the levels that the entire miRnome/lncRNome can be quantified in a rapid and reliable manner, facilitating investigators' efforts to identify unique expression profiles that are linked with defined tumour chemoresistance properties.
Such breakthroughs in technology are proving to be essential for biomarker researchers since evermore studies are leading to the paradigm that tumour clinical characteristics such as chemoresistance are the result of influence by multiple miRNAs and/or lncRNAs acting in a simultaneous manner, and not merely the outcome of one individual non-coding RNA's dysregulated expression. The issue with this paradigm is the degree of complexity and resource consumption in carrying out detailed functional analyses to validate each permutation of non-coding RNA influences from an identified expression profile comprising just a handful of miRNAs/lncRNAs. Hopefully, further advances in bioinformatics and analytical technologies can permit more accurate trawling efforts to pinpoint such biomarkers and/or possibly allow for high throughput functional analyses for the entire miRnome/lncRNome in a rapid and effective manner.
The clinical importance for all global research efforts to identify and validate novel non-coding RNA biomarkers for cancer chemoresistance must certainly not be underestimated. The validation of reliable miRNA and/or lncRNA biomarkers for individual cancer chemoresistance (be it innate or acquired) can lead to the exploitation of such biomarkers as novel drug targets. Ultimately, antagonists and/or mimics (depending whether the miRNA/lncRNA is up-or down-regulated) for each non-coding RNA drug target can be developed and safely delivered as adjunct therapy together with conventional chemotherapeutic drugs. The adjuvant therapy leads to enhanced tumour sensitivity for the conventional chemotherapeutic drugs, therefore markedly enhancing chemotherapy effectiveness.
Alternatively, in patients who are particularly prone to the dose limiting adverse effects of conventional chemotherapy, the doses for the latter can be reduced due to the addition of the novel Undoubtedly, the recent progress in molecular analytical and sequencing technologies has advanced to the levels that the entire miRnome/lncRNome can be quantified in a rapid and reliable manner, facilitating investigators' efforts to identify unique expression profiles that are linked with defined tumour chemoresistance properties.
Such breakthroughs in technology are proving to be essential for biomarker researchers since evermore studies are leading to the paradigm that tumour clinical characteristics such as chemoresistance are the result of influence by multiple miRNAs and/or lncRNAs acting in a simultaneous manner, and not merely the outcome of one individual non-coding RNA's dysregulated expression. The issue with this paradigm is the degree of complexity and resource consumption in carrying out detailed functional analyses to validate each permutation of non-coding RNA influences from an identified expression profile comprising just a handful of miRNAs/lncRNAs. Hopefully, further advances in bioinformatics and analytical technologies can permit more accurate trawling efforts to pinpoint such biomarkers and/or possibly allow for high throughput functional analyses for the entire miRnome/lncRNome in a rapid and effective manner.
The clinical importance for all global research efforts to identify and validate novel non-coding RNA biomarkers for cancer chemoresistance must certainly not be underestimated. The validation of reliable miRNA and/or lncRNA biomarkers for individual cancer chemoresistance (be it innate or acquired) can lead to the exploitation of such biomarkers as novel drug targets. Ultimately, antagonists and/or mimics (depending whether the miRNA/lncRNA is up-or down-regulated) for each non-coding RNA drug target can be developed and safely delivered as adjunct therapy together with conventional chemotherapeutic drugs. The adjuvant therapy leads to enhanced tumour sensitivity for the conventional chemotherapeutic drugs, therefore markedly enhancing chemotherapy effectiveness.
Alternatively, in patients who are particularly prone to the dose limiting adverse effects of conventional chemotherapy, the doses for the latter can be reduced due to the addition of the novel non-coding RNA-directed therapy. This leads to a great reduction in dose-limiting adverse effects and consequent discomfort in the cancer patient.
Finally, such chemoresistance biomarker expression profiles can be easily quantified from tumour biopsy through real time quantitative polymerase chain reaction (RT-qPCR) assays. The additional clinical information regarding the chemoresistance properties can provide the oncologist with valuable pre-emptive knowledge. Such additional information aids in developing a bespoke chemotherapy drug combination for the cancer patient that maximizes therapeutic efficacy and therefore minimises "trial and error" chemotherapy regimes, since the tumour would be exposed only to the chemotherapeutic agents to which it is fully sensitive.
However, efforts to render such a powerful theranostic technology is confronted with two main issues for it to become commonplace within global reach.
Firstly, the most effective technologies for accurate quantitative analysis of ncRNAs remain to be real-time, reverse transcription quantitative PCR (RT-qPCR) and next-generation sequencing. Both such technologies require sophisticated equipment and highly skilled staff dedicated to the processing of clinical samples for miRNA and lncRNA expression profiling. Eventually, these technologies can be miniaturized and simplified to the level of the development of a cost-effective point-of-care diagnostic apparatus that can be utilised by healthcare professionals with limited experience in the analytical technologies being employed.
Secondly, the issues regarding safe and effective drug delivery of novel miRNA and lncRNA therapeutics still pose a hurdle to rapid development of such translational medicine and effective availability for use by the individual cancer patient. Notwithstanding this issue however, the pharmaceutical industry are currently focusing hard on multiple circumvention methods for effectively providing efficient drug delivery options. These efforts are concentrated in pharmaceutical companies that are entirely dedicated to the research and development of miRNA and lncRNA-based therapeutics.
However, the authors sincerely believe that, despite such challenges, the advent of such novel clinical oncology drug treatment/management protocols will become a reality within the hospital setting in the not too distant future.
Supplementary Materials: The following are available online at www.mdpi.com/2073-4425/8/3/95/s1, Table S1: Frequency chart depicting prevalence of individual miRNAs in specific cancer models reported in the literature (as depicted in Table 1 above) and from the data reported in the review article by Garofalo and Croce [42], Table S2: Frequency chart depicting prevalence of individual lncRNAs in specific cancer models reported in the literature (as depicted in Table 2 above).

Conflicts of Interest:
The authors declare no conflict of interest.