do not necessarily reflect the views of UKDiss.com.
Most mammalian miRNAs are within introns of either protein-coding or noncoding TUs, whereas ~10% are encoded by exons of long non protein-coding transcripts, also known as mRNA-like noncoding RNAs (mlncRNAs) . Intergenic miRNAs have a dedicate promoter, whereas clustered miRNAs share one promoter and are co-regulated as a part of long pri-miRNAs .
In mammalian cells three major RNA polymerases are involved in the transcription of the nuclear genome: Polymerase I (Pol I) that is specific for the ribosomal RNA (rRNA), Polymerase II (Pol II) that transcribes the mRNA coding genes and Polymerase III (Pol III) that is involved in the transcription of non-coding RNA, included tRNAs.
Although miRNAs belong to ncRNA genes, several evidences suggest that they are transcribed by Pol II, although the possibility that a small number of miRNA genes might be transcribed by other RNA polymerases cannot be excluded. In particular it is reasonable to assume that Pol II transcribes the miRNAs involved in various regulatory processes and Pol III transcribes the miRNAs whose expression is required in all tissues . Similar to other genes transcript by Pol II, miRNAs have a cap structure and a poli(A) tail and their expression profiles indicate that miRNAs are under control during development and in various tissues. Moreover, a mature miRNA can be obtain from a plasmid that contains a pri-miRNA under the control of heterologous Pol II promoter and miRNA transcription activity is sensitive to the concentration of a-amanitin that inhibits Pol II activity but not the Pol I and Pol III . Finally the physical association of Pol II with promoter of several miRNAs has been demonstrated by immunoprecipitation analysis .
Transcription of miRNA genes by RNA Pol II results in the production pri-miRNAs which have a stem and loop structure with both 5′ 7-methylguanosine (m7) caps and 3′ poly(A) tails. Cleavage of the pri-miRNAs by the endoribonuclease III enzyme Drosha and its cofactor DGCR8 results in the production of pre-miRNAs, which are hairpins of approximately 70-100 nt in length and contain a monophosphate at the 5′ terminus and a 2-nucleotide overhang with a hydroxyl group at the 3′ terminus.
The complex formed by Drosha and DGCR8 is called Microprocessor and it has been demonstrated that both of these two factors are necessary for the maturation of pre-miRNA. Specifically neither recombinant DGCR8 nor Drosha alone is able to process pri-miRNA .
Drosha is a conserved protein of 160 kDa and belongs to the Rnase III family proteins, the enzymes responsible of endonucleolytic reactions. In miRNA biogenesis two members of this family are involved: Drosha in the nucleus and Dicer in the cytoplasm. The differences between them is in the substrate specificity and mechanism of action (46): Drosha binds to pri-miRNA and introduces a cut from the terminal loop, whereas Dicer cleaves from the 3′ terminus .
Figure 5. Mechanism of action for Drosha (a) and Dicer (b). Modified from . a) DGCR8 works as a molecular anchor for Drosha. DGCR8 recognises and binds the junction between the stem ans the loop of a pri-miRNA. Drosha binds DGCR8 and cuts approximately 11 pb away from the junction through its two RNase III domains, RIIIDa and RIIIDb. RIIIDa cleaves the 3’ strand, whereas RIIIDb cleaves the 5’strand. b) PAZ domain recognizes and binds the 3’ end of the pre-miRNA cleaved by DROSHA. Dicer cuts approximately 22 nt from the 3’end through the RIIIDa and RIIDb subunits.
Drosha is composed by two RNase III domains (RIIIDs) and a double-stranded RNA binding domain (dsRBD), both of them essential for catalysis .
DGCR8/PASHA is a 120 kDa protein that contains two dsRBDs and a putative domain (WW). The human DGCR8 gene is located on chromosome 22q11 and is expressed ubiquitously in both foetus and adults. Its name is linked to a group of diseases, which occur in case of mutation in this region and that are indicated as Di George syndrome.
The mechanisms that regulate how the Microprocessor can recognizes the target are largely unknown, however it is reasonable to assume that the complex is able to bind molecules with a stem and loop structures avoiding the widespread similar RNA stem loop structures that can form across the transcriptome [74, 75].
Basically the role of DGCR8 is to bind the pri-miRNA and the role of Drosha is to cleave it. Notably, Drosha seems not to have any pri-miRNA binding activity .
DGCR8 recognizes the target in correspondence of the ssRNA-dsRNA junction and interacts with all stem region (about 33 pb) and with the loop. A small region of DGCR8 (residues 484 through 750) is sufficient to facilitate pre-miRNA processing in vitro and includes both the dsRBDs and the C-terminal Drosha-binding domain; the N-terminal region is not involved in processing but it is important for nuclear localization of DGCR8. DGCR8 also stabilize Drosha protein.
Drosha cleavage is shown in the Figure 5: after DGCR8, the dsRBD of Drosha maytransiently interact with the substrate to place the processing centre approximately at11 bp from the point in which double strand is formed. In particular Drosha cleaves at 2 helical turns away from the terminal loop and 1 helical turn away from the basal segment. Pri-miRNA stem, defined by internal and flanking structural elements, guides the binding position of Drosha-DGCR8, which consequently determines the cleavage site .
The result of processing is a pre-miRNA stem and loop which are hairpins that contain a monophosphate at the 5′ terminus and a 2-nucleotide overhang with a hydroxyl group at the 3′ terminus.
There is a post-transcriptional crossregulation between Drosha and DGCR8. In fact, DGCR8 stabilizes Drosha via protein-protein interaction but if the level both Drosha and DGCR8 are high in the cell, Microprocessor would cleave and destabilize the DGCR8-mRNA, decreasing the level of the correspondent protein. This system has been demonstrated in vitro and in vivo and suggests that it could contribute to control of miRNA biogenesis .
The transport of pre-miRNA from the nucleus to the cytoplasm is mediated by the nuclear pore complexes (NPCs), a proteinaceus channel embedded in the nuclear membrane .
In particular, exportin-5 (Exp5), a Ran-dependent importin-ß-related transport receptor, mediates nuclear export of pre-miRNA Exp5 and forms a complex cooperatively with the substrate (cargo) and the GTP-bound form of Ran (RanGTP) Upon export, hydrolysis of GTP to GDP on Ran results in release of the cargo from the export complex .
Nuclear transport receptors usually recognize specific sequences on the target but in pre-miRNAs no consensus sequences has been found. Probably Exp-5 recognizes the structural motif common in all pre-miRNAs which contain a 2 nt-3′ overhangs and a loop.
Following their export from the nucleus, pre-miRNAs are processed into 19-22 nucleotide miRNA duplexes by the cytoplasmic ribonuclease Dicer (Figure 4).
Dicer is a highly conserved protein found in almost all eukaryotes. Some organisms contain multiple Dicer homologues, such as D. melanogaster, which has two isotypes of the enzyme (Dicer-1 and Dicer-2).
Human Dicer is a protein of 160 kDa, belongs to the RNase III family of enzymes showing either an endonucleoliytic or a haelicasic activity. In fact, it first cleaves the pre-miRNA and then contributes to the loading of miRNA into RISC. Dicer contains the following domains: an N-terminal ATPase/RNA helicase domain, a DUF283 domain, a Piwi Argonaut and Zwille (PAZ) domain, two RNase III domains (RIIa and RIIb) and a dsRNA binding motif domain (dsRBD). Moreover Mg2+ ions are present in each catalytic centre [61, 80].
The dsRNA binding motif domain binds the pre-miRNA close the loop. The C-terminal RIIb, proximal to the dsRBD, cleaves the 5′ strand of the hairpin and RIIa cleaves the 3′ strand. On the opposite site, at the terminus, PAZ domain interacts with the 3′ overhang of the substrate. RNase III domains functions as a ruler measuring the distance from the 3′ end of pre-miRNAs to cleavage site to allow the correct production of the mature miRNA duplex (Figure 5) [61, 81, 82]. Notably, it has also been demonstrated that Dicer is able to cleave without the 3’overhangs, interacting directly with only 5′ strand .
The products of cleavage by Dicer are 19-22 nucleotide mature miRNA duplexes, which are usually approximately 21 nucleotides in length with both sets of termini having 5′ monophosphates, 3′ 2-nucleotide overhangs and 3′ hydroxyl groups. The miRNA guide strand, the 5′ terminus of which is energetically less stable, is then selected for incorporation into miRISC [84, 85].
Mature miRISC is composed by Ago proteins and binds both miRNA and mRNA targets. In mammals four different Ago proteins have been discovered (Ago1-Ago4), all characterized by the same two domains, PAZ, MID (middle) and PIWI.
PAZ-domain function is to anchor the 3′-end of the miRNA with its oligonucleotide/oligosaccharide binding (OB) fold and bind the 3’hydroxilated end .
On the other hand MID domain anchors the 5′ end of the miRNA, owing to its ability to discriminate between the four different bases. In particular it preferentially binds the uridine base .
The PIWI domain is structurally similar to RNase H, with an endonuclease activity Mg2+ dependent. Both of them are able to cleave the RNA but RNase H recognises the RNA-DNA interaction whereas RISC recognises the RNA-RNA interaction .
In mammals, all the proteins of Ago family can mediate both the translation repression of mRNA and mRNA decay in P-bodies, but only Ago2 has an endonucleolytic activity. In addition, other factors, such as the Trans-Activation Response RNA Binding Protein (TRBP) and Protein kinase RNA activator (PACT) are involved into strand selection and RISC assembly. In particular, TRBP in association with Dicer provides a platform for RISC and aids the recruitment of Ago2. PACT binds the N-terminal region of Dicer containing the helicase motif [68, 89]. Neither TRBP nor PACT are required for the pre-microRNA cleavage, however their role is crucial for the efficiency of miRNA processing. In fact it has been demonstrated that their depletion causes a decreasing in mature miRNA level [88, 89].
In the canonical pathway of miRNA biogenesis, the imperfect base-pair between a miRNA and the sequence in the 3’UTR of target mRNA has as effect to inhibit protein synthesis by either repressing translation or promoting mRNA deadenylation and decay . Deadenylation is the first step in mRNA decay, and is generally followed by removal of the m7G cap (the 7-methylguanosine-triphosphate structure) at the 5′ end of mRNAs, which promotes their translation and protects them from degradation and exonucleolytic 5′ to 3′ degradation of mRNA. In the deadenylation of mRNA, Glycine-Tryptophan Protein of 182 kDa (GW182) and other proteins are involved and act in association with RISC. While the amino-terminal part of GW182 interacts with AGO through its GW repeats, the carboxy-terminal part interacts with the poly(A) binding protein (PABP) and recruits the deadenylases CCR4 and CAF1 proteins  ( Figure 4)
In addition to this canonical process, another non-canonical miRNA biogenesis pathway has been discovered in flies and mammals. This alternative process involves a group of miRNAs derived from short hairpin introns (mirtrons) that undergo to splicing and directly form the pre-miRNAs bypassing Drosha cleavage. Following the completion of splicing, the branch point of the lariat-shaped intron is resolved and the debranched intron forms a hairpin structure that resembles premiRNA. Some precursors (mirtrons) contain tails at either the 5’ or 3’ end, which therefore require exonucleolytic trimming for nuclear export .
1.4 MiRNAs and human pathologies
MiRNAs regulate all the biological functions, including cell growth, differentiation, proliferation, apoptosis and signal transduction pathways . Their de-regulation may be due to mutations in their encoding genes or in the miRNA biogenesis machinery and their levels may change during either the pre-natal development or the post-natal life. As epigenetic features, they are heritable but may change spontaneously or in response to environmental factors.
The first evidence of the correlation between miRNAs and human diseases has been identified by studying the DGCR8 protein, the RNase III enzyme that, together with Drosha, is responsible for the cleavage of the pri-miRNA in the nucleus. DGCR8 gene is located on 22q11.2 chromosome and deletions in this gene causes the so called Di George Syndrome, a rare disease. Children affected by the DGCR8 syndrome develop severe learning disabilities, such as autism, shows congenital malformations, such as the cleft palate, and have many other problems, such as congenital heart diseases and hormone-related disorders. Moreover, microdeletions of 22q11.2 are responsible for up to 1–2% of schizophrenia cases . It has been demonstrated that a Dgcr8 haploinsufficiency impairs the miRNA biogenesis and contributes to the behavioral and neuronal deficits associated with the disease in vivo .
Since then, the regulatory role of miRNAs in both physiological and pathological conditions has been widely demonstrated.
Some miRNAs have been also found in viruses, such as the gamma herpesvirus Kaposi’s sarcoma-associated herpesvirus (KSHV) and the simian virus 40 (SV40) [94, 95]. SV 40 gene encodes for a miRNA that downregulates the T-antigens by binding their mRNAs. This mechanism does not influence the infection directly but protects the infected cells from the effect of the T-cells, thus promoting the survival of the virus in the host cells.
MiRNAs regulate the immune and autoimmune response in many ways. MiR-21, for example, regulates the T-cell response, the differentiation of dendritic cells and the myeloid cell function . This is possible through the ability of a single miRNA to bind and represses the activity of multiple mRNAs at the same time.
MiRNAs are also involved in the normal development of the cardiovascular system, as well as in many pathological conditions of the heart, such as the miR-208a, that is encoded by the MYH6 gene. MiR-208a has shown to be upregulated in hypertrophic mice  and an anti-miR-208a lead to a reduced fibrosis, hypertrophy and obesity in vivo, thus suggesting a possible use of this miRNA as a therapeutics agent.
At the same manner of the genes, miRNAs are involved in the embryonic development and their de-regulation may cause important congenital defects during the different stages of specific pathways. For example, miR-140, miR-17-92 cluster (miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1 and miR-92a-1) and miR-200b play an important regulatory role during all the stages of palatogenesis . MiR-140-5p regulates the first step of the process of the craniofacial formation, by regulating the migration of the neural crest cells to a more ventral position in the embryo, in order to differentiate in multiple tissues, including the craniofacial cartilage. The levels of miR-140-5p may be affected by a SNP (rs7205289: C>A) located in the precursor, that impairs its maturation  or by a mutation on one of its targets, the Platelet Derived Growth Factor Receptor A (PDGFRA) . The palatal growth is the second stage of the palatogenesis and is characterized by the proliferation of the cells and accumulation of the extracellular components, such as collagen. MiR-17-92 cluster sustains the proliferation of the cells by inhibiting many anti-proliferative molecules. The activation of the cluster in the post-natal life, however, may contribute to the proliferation of some tumours . In the final stage of palatogenesis, the epithelium between the two palatal shelves is removed in a process named “epithelial-to-mesenchymal transition”. In this step, the expression of miR-200 must gradually decrease. MiR-200, which targets many factors involved in the Transforming Growth Factor-ß (TGF-ß) pathway, should be active to allow the closure of the palatal shelves.
1.5 MiRNAs and cancer
MiRNAs in cancer can act either as onco-suppressors (tumour-suppressor miRs) or oncogenes (oncomiRs) and may have a direct impact on cancer phenotypes, depending on their specific targets and the pathways that miRNAs interact with. MiRNAs act as onco-suppressors by targeting oncogenic mRNAs and as oncogenes when they repress the translation of mRNAs with tumour suppressor functions. This dual classification is complicated by the fact that miRNAs can act as onco-suppressor and oncogenes at the same time and this mainly depends on the context in which miRNAs are expressed. The levels of specific miRNAs in human tumours are like a signature that reflects and is associated with the diagnosis, staging, progression, prognosis and response to therapies .
The first evidence of the involvement of miRNAs in cancer has been formulated by Croce et al . They found that two miRNAs, miR-15 and miR-16, were downregulated in patients affected by B cell chronic lymphocytic leukemia (CLL) and that the two miRNAs were located at chromosome 13q14, a region frequently deleted in CLL. The target of miR-15/16 cluster is the B-cell lymphoma 2 (BCL2), an anti-apoptotic protein. The downregulation of the two miRNAs contributes to a higher expression of BCL2, thus conferring the mlignant cells the resistance to the programmed cell death .
Since then, many groups investigated the role of miRNAs in cancer and demonstrated their usefulness either as biomarkers, as they were detected in a wide range of biological fluids, or indicators of the tumour subtypes and patient response to treatments .
MiRNAs are extremely plastics and their levels may change easily. As all the other epigenetic regulators, their levels may vary with the time and the stage of the diasease. Considering this, the variation of miRNA levels in cancer may help the progression of the disease. This is because MiRNAs themselves are under the control of genetic and epigenetic mechanisms that contribute to their mis-regulation. For instance, pre-transcriptional regulators are able to affect the biogenesis of miRNAs leading a gain or loss of function of miRNA gene copy number, mutations of the miRNA precursor, histone deacetylation and hypermethylation of miRNA promoters . Some miRNAs are clustered in regions of genomic instability or fragile sites and this may contribute to their level and function . MiRNA levels are also regulated by other transcription factors associated with RNA Polymerase II and all the other nuclear and cytoplasmic enzymes involved in their biogenesis. For instance, it has been shown that the deletion of Drosha, DGCR8 and Dicer1, the main components of the miRNA maturation machinery, enhances the tumour development in a K-Ras–induced mouse model of lung cancer . The level of miRNAs may also be influenced by Single Nucleotide Polymorphisms (SNPs) whitin their own sequnce or the sequence of their target genes, thus influencing the activity of the cancer cells . This is the case of a SNP in the 3’UTR binding region of let-7, which halts the binding of the miRNA and this leads to a higher level of KRAS levels in NSCLC patients .
The diysregulation of miRNAs may contribute to the malignant transformation in distant sites by directly modulating enzymes which take part in methylation-mediated silencing and chromatin remodelling, in a paracrine manner through exosomes, microvesicles and protein complexes able to influence the tumour microenvironment and by promoting the release of mediators which activate pro- or anti-cancer immune activity . In the tumour microenvironment, miRNAs have been found encapsulated in microvescicles, associated with protein complexes and, in some cases, free. They may derive from live, necrotic or apoptotic cells and released into the tumour microenvironment, where they may be incorporated by the recipent cells and contribute to the spread of the tumour in a paracrine manner . In a multiple myeloma model in vivo, miR-135 is encapsulated and released in the extracellular space by the malignant cells of the bone marrow. It then penetrates into the endothelial cells andenhances the angiogenesis by targeting the inhibitor of the Hypoxia-Inducible Factor (HIF) . The presence of miRNAs in all biological fluids, including saliva, breast milk and blood, put the attention on their role as biomarkers, as the de-regulation of specific circulating miRNAs can reflect the physiological and pathological status of the patients . Finally, the activity of miRNAs in cancer and other diseases may be influenced by other RNA molecules named “competing endogenous RNAs) (ceRNAs) that compete with the miRNAs as they share the miRNA binding site with other RNAs .
1.6 Strand selection and miRNA prediction tools
During miRNA biogenesis, one of the two strands of the double-stranded molecule is processed (guide strand or miR) and the other one (passenger strand or miR*) is degraded. This process is generally named “strand selection”. The guide and passenger strands are frequently named -3p and -5p, according to the direction of the mature miRNA strand. Altough the two sister strands are partially complemetary and come from the same precursor, they have different targets and are involved in distinct biological pathways or may target the same mRNA but in different sites. The miRNA strand selection may vary among tissues and between normal and malignant cells. Moreover mutations in the miRNA-precursor, or in proteins involved in miRNA biogenesis, may change the ratio between the sister strands, resulting in a deregulation of key biological pathways .
In Mammals, previous models have suggested that the choice of the mature miRNA depends on the thermodynamic and structural properties of the processed duplex. The more stable strand generally starts with a 5’ Uracil (U) base ans is rich in purines, whereas the less stable starts with a 5’ Cytosin (C) base and is rich in pirimidines . According to this theory, the less stable fragment is loaded into RISC because the purine residues may facilitate the strand loading through sequence-independent interactions. In fact, the PAZ domain of Ago2, which participates to the initial steps of DNA recognition and binding, contains numerous aromatic residues involved in RNA binding, and the interaction between these residues and the purine-rich duplex may contribute to strand selection . However, data from sequencing studies suggest that it is not always the case, as often both of the strands are active and the choice of the active strand may vary with time and tissues . In fact, the predominant form of a miRNA differs among tissues, times of development, and between species suggesting the existence of other mechanisms for controlling the selection of mature miRNAs, such as variations within the proteins that form the RISC complex [116, 117]. The most common modifications of the pre-miRNA are located at the 3’end, probably because they have a little effect on the target selection . As miRNAs recognise the 3’-UTR of their mRNA targets and for this reason, presumably, the modifications at the 3’ end may have a smaller impact on strand selection than the changes at the 5’ end. However, it is not possible to exclude that the modifications at the 3’end of the duplex may have a functional role. The strand selection process may vary according also to mismatches in the pre-miRNA sequence , as these variations are recognised by different proteins of the RISC. Finally, Single Nucleotide Polymorphisms in the miRNA gene sequence may affect both the miRNA processing and the binding with the mRNA targets .
To assess the biological role of a miRNA, the potential mRNA targets should be identified. As described previously, one single miRNA can target up to 2,000 transcripts and considering that the miRNA sequence is short and not perfectly complementary to the mRNA targets, this research may be difficult. The most common way to make this search easier is to use some bioinformatics tools. The main criteria used by these algorithms are the Watson-Crick base paring between the nt 2-8 at the 5’ end of the miRNA (seed region) and the 3’UTR of the mRNA, the free energy of the binding and the conservation of the miRNA binding sites between the species. Some of these tools will be further described in Section 2.16. However, not all miRNAs can bind their targets following the base paring, for example under the circumstance of a SNP in the pre-miRNA or for an imperfect binding. For this reason, in this study, the classical bioinformatics tools have been integrated with an alternative approach with the use of synthetic miRNA mimics designed with a biotin stretch at the 3’ end. These mimics can be transfected into cells and then purified by using strepdavidin beads in order to obtain just the mRNA bound to the miRNA of interest. The mRNA profile can then be finally analysed through sequencing and validated by qPCR. This method, as will be discussed later, will allow, ideally, to identify all the targets that are not predicted by the bioinformatics algorithms, especially the non-seed based interactions. Although it is a good approach to find the direct targets of a specific miRNA, this method also shows some limitations. The main limitation is due to inability to detect some transcripts because they might be regulated and expressed differently among the cells or because other transcripts compete for the miRNA binding . For these reasons we decided to use an integrate approach to better address the biological role of the two forms of miR-140 in lung cancer invasion.
1.7 Potential and current applications of miRNAs in human diseases, including cancer
Many studies have shown the possible applications of miRNAs as biomarkers, as in many cases a miRNA signature has been associated with the progression of specific diseases and/or the response to a treatment. According to the World Health organisation, a biomarker is “any substance, structure or process that can be measured in the body or its products and influence or predict the incidence of outcome or disease” . A good biomarker should be easily detectable by using non-invasive methods and should reflect the healthy status of the patient, thus allowing either an early diagnosis or prognosis. MiRNAs are ubiquitously expressed in all biological fluids, including saliva, urine, breast milk and blood and correlate with many pathological conditions . All these characteristics, together with the powerful techniques available that allow to detect small molecules, made them particularly attractive as a diagnostic and prognostic markers for many diseases, including cancer. MiRNAs are tissue and disease specific and their variation may be an indicator of a pathological condition. An early diagnosis, together with an accurate classification of the disease, is crucial for an incisive decision on the patient treatment.
MiRNAs have been found useful for the diagnosis of many diseases, such as cancer, cardiovascular and metabolic disorders [123-125]. In cancer, miRNAs can discriminate between two or more different cell subtypes because they vary according to the oncogenic pathway activated and the expression of miRNAs in blood reflects their levels in the tumour tissues [123, 126]. MiR-200 family, for example, is upregulated in an advanced state of breast cancer and this reflects the stage of the disease, as the members of miR-200 family target zeb1 and zeb2, two of the transcription factors deeply involved in the Epithelial-Mesenchymal Transition (EMT) . In other cases, miRNAs are predictive of cancer with specific mutations, as in the case of melanoma, where the low levels of miR-193a, miR-338 and miR-565 have been found in patients positive for BRAF mutations .
MiRNAs regulate many proteins involved in the physiological processes of the heart and recently many circulating miRNAs have been used for the stratification of the myocardial infarction . In the clinical practice, multiple factors such as age, gender, systolic blood pressure and smoking habits, are usually evaluated to estimate the cardiovascular risk and it has been demonstrated that the cardiovascular risk is associated with the circulating levels of miRNAs . The utility of miRNAs for the diagnosis of cardiovascular diseases has been further strengthen by the demonstration that specific miRNAs are released into the bloodstream after an infarction .
A specific miRNA signature has been recently discovered in patients affected by the type 2 diabetes mellitus (T2DM) and this may help to monitor the progression of this disease. In particular, five miRNAs, miR-661, miR-571, miR-770, miR-892b and miR-1303, have been found highly expressed in patient serum and a higher level have been shown in patients with complications . These five miRNAs may be useful as predictive biomarkers for T2DM-related complications and may be involved in the progression of the disease.
Some research groups argued that many factors make the applications of miRNAs in clinical practice quite hard . In fact, the levels of miRNAs may vary amongst people because of the diet or pharmacological treatments, such as aspirin, that may reduce the plasma levels of miR-126 in patients affected by T2DM, thus making this miRNA in the blood not predictive of cardiovascular diseases . However, the variability is the major concern of all the biomarkers currently used in the clinical practice and, in the case of the miRNAs, this is not always the case. Some miRNAs, in fact, are tightly associated with the genetic features of a disease and their levels are weakly influenced by other factors, such as miR-483-5p, that may be used to discriminate between the paediatric and adult liver tumours and may not be influenced by any treatment . Another issue related to the use of miRNA as biomarkers, is the normalisation method used in the quantification of circulating miRNAs as, to date, a stable housekeeping gene has not been identified. Moreover, it is still hard to establish the range within the levels of specific miRNAs may be considered physiologically. Finally, operation procedures should be produced 
All these observations, however, do not weaken the validity of the miRNAs, which may be used in concert with other predictive molecules/factors and should take into account for comorbidities and treatments.
MiRNAs in tumours can act as onco-suppressors when their expression is lower than in normal conditions, or as oncogenes when upregulated. Their low molecular weight and small size, together with their ability to target multiple molecules without toxicity, provides a new opportunity for cancer treatment compared to other methods such as the gene therapy. For instance, miR-181 regulates T-cell receptor sensitivity by targeting multiple phosphatases, an effect that was not observed by silencing a single component of the pathway by siRNA . Oncogenic microRNAs can be targeted by repression and therefore the inhibition of the interaction between miRNA and mRNA and onco-suppressor expression can be restored in cells using molecules that “mimic” their activity.
Oncogenic miRNAs can be inhibited using oligonucleotides complementary to the mature miRNA (antagomiRs) [130, 131]. AntagomiRs are cholesterol-conjugated single-stranded RNA analogues complementary to miRNAs. The chemical modifications at their 3’-end increase their stability in vivo and their efficient direct uptake via the cells membrane into viable cells. Another anti-miR-based strategy uses a stable transfection with virus-associated “miRNA sponges”. They are transcripts that expressed 3’ UTRs containing multiple miRNA binding sites from strong promoters, such as U6 or cytomegalovirus .
MiRNA replacement therapy aims at substitution of tumour suppressive miRNAs expressed at lower levels by using oligonucleotide mimics containing the same sequence as the mature endogenous miRNA. This can be achieved by vectors (plasmids or lentiviruses) expressing miRNA precursors or by the delivering of artificial miRNA duplexes (mimics). MiRNA precursors require transcription and processing, whereas miRNA mimics are directly incorporated into RISC. The challenges for miRNA re-establishment are the same of miRNA suppression: the specificity and efficiency of the delivery. Despite the miRNA replacement using virus vectors having produced good results , the application of non-viral vectors, such as nanoparticles, increased in the last few years. The efficiency of transfection in host cells is higher with viral vectors compared to non-viral vectors. However, the viral vectors show a high immunogenicity and cytotoxicity. The first person who died in a gene-therapy clinical trial suffered from a partial deficiency of a liver enzyme, the ornithine transcarbamylase (OTC). The patient died in 1999 for multiorgan failure, four days after the treatment with an adenovirus vector to deliver the gene for OTC to the liver. The autopsy showed that the vector circulated into the bloodstream and accumulated in the spleen, lymph nodes and bone marrow, although it had been infused directly into the liver through the hepatic artery . One of the phenomena associated with the use of viral vectors is the insertional mutagenesis, the ectopic chromosomal integration of viral DNA that disrupt the expression of tumour suppression gene or activates oncogene leading to the malignant transformation of cells. Due to their biosafety, the non-viral vectors, as the vehicles for gene therapy, have drawn significant attention.
Two non-viral methods to protect both miRNA mimics and inhibitors from the degradation by RNAases in serum or in the endocytic compartment of the cell . Firstly, the miRNA mimics chemistry can be modified by methylation of the passenger strand to increase the stability and anti-miRNAs can be modified through the addition of chemical groups that “lock” the nucleic acid (Lock Nucleic Acid, LNA) .Secondly, the vehicles to encapsulate and protect the small RNAs can be produced. And indeed the use of lipid-based nanoparticles is considered the best method to overcome the physical barriers in vivo, such as enzymatic degradation, and to access target cells with high tissue-specificity [136, 137].
The use of miRNA mimics therapeutics have been successful in both in vitro and in vivo models [136, 138, 139] and recently a pharmaceutical company, MiRNA Therapeutics (Texas), has started a multicentre Phase 1 clinical trial of a single miRNA agent in humans and they plan to advance into Phase 2 . More than one hundred patents affected by liver cancer will be treated with MRX34, a synthetic miRNA mimic of miR-34, which is known to be a tumour suppressor. MRX34 is a double-stranded oligonucleotide delivered using a liposome-formulation called NOV40, which is able to avoid interactions with normal cells and achieve maximal accumulation of the mimic into the liver, due to the high liposomal accumulation into the liver. Other miRNA therapeutics in clinical trials include the anti-miR-122 for the Hepatitis C, antimiR-103/107 for T2DM, antimiR-155 for cutaneous T cell lymphoma, miR-129 mimic for scleroderma and miR-16 mimic for NSCLC .
The personalised medicine aims to treat the patients with a particular condition by using specific approaches to better manage their health and address them to the best therapy. In the era of the molecular medicine, parameters other than the clinical and pathologic features are included in the risk stratification of patients, such as DNA methylation, gene expression and microRNA expression . The miRNA signature associated to each individual tumour is now easily possible through either RT-qPCR or Next Generation Sequencing and the data profiles generated are stable within the clinical samples and allow a better prognosis .
The miRNA molecular profiles correlate with patient survival in many cancers, such as lung and breast [141, 142]. In breast cancer, in particular, there is a specific miRNA signature that allow to discriminate amongst the subtypes . Because of the significant differences in prognosis and treatment, miRNAs may be particularly useful in concert with other molecular markers.
A miRNA signature helpful for a better diagnosis has been found also in other diseases, such as cardiovascular and metabolic disorders [143, 144]. In atherosclerosis, which is the primary cause of cardiovascular diseases, the changes in miR-134, miR-198, and miR-370 in the blood of patients may be useful to predict the clinical outcome of the disease . In metabolic disorders, such as type 2 diabetes, circulating miRNAs are associated with those patients with impaired wound healing .
1.8 MiRNAs in lung cancer
The first evidence of the involvement of miRNAs in human lung cancer was reported by Takamizawa et al. . In their work, they demonstrated that let-7 miRNA was downregulated both in vitro and in vivo and correlated with prognosis of lung cancer patients. Later on it has been shown that the let-7 reduction lead to an increase of RAS level . Also a SNP on let-7 is significantly associated with an increased risk for NSCLC among moderate smokers by modifying the binding to 3′ untranslated region of KRAS mRNA .
As previously stated, miRNAs can act as oncogenes or oncosuppressors, depending on the timing and the levels of expression. This is true for all the cancers, including NSCLC. For instance, miR-17–92 cluster, which includes miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1 and miR-92a-1, is overexpressed in lung cancer and enhances the cancer cell growth . Contrarily, miR-17~92 shows a tumour suppressive role in other cancers, such as breast cancer, squamous cell carcinoma of the larynx, retinoblastoma, hepatocellular carcinoma and nasopharyngeal carcinoma . This paradox may be due to the ability of a single miRNA to target multiple mRNAs involved in distinct pathways.
.Modifications in the key enzymes involved in miRNA biogenesis are also found to be correlated with the prognosis of lung cancer patients, such as in NSCLC patients, a higher Dicer expression leads a significantly better prognosis than low Dicer expression [150-152]. Moreover, miRNA-related genetic variants are also associated with the clinical outcomes in early-stage NSCLC cancer patients  thus confirming the importance of miRNAs as prognostic markers.
Emerging evidences suggest the useful role of miRNAs as biomarkers for the early diagnosis of lung cancer. In lung cancer tissues, a meta-analysis study indicates a specific miRNA signature associated with lung cancer tissues, with seven miRNAs up-regulated (miR-21, miR-210, miR-182, miR-31, miR-200b, miR-205, and miR-183) and eight down-regulated (miR-126-3p, miR-30a, miR-30d, miR-486-5p, miR-451a, miR-126-5p, miR-143, and miR-145) . As non-invasive biomarkers, miRNAs in sputum and blood, improved the sensitivity and the diagnosis of NSCLC patients. For example, miR-21 has been found upregulated in the sputum of patients and low levels of miR-21, miR-126, miR-210, and miR-486-5p in plasma have been associated with this disease .
Two different research groups have studied the miRNA molecular profile for lung cancer diagnosis and prognosis [141, 155] and found a five-miRNA signature (let-7a, miR-221, miR-137, miR-372, and miR-182), that is associated with survival and cancer relapse in NSCLC patients. MiRNA-let-7 is downregulated in NSCLC and further studies demonstrated its onco-suppressor activity in a xenograft model . MiR-221 levels are lower in tumour tissues than in controls and shows a tumour-suppressive activity in lung cancer together with miR-222 . The identification of a microRNA signature that can predict survival of patients with squamous cell carcinoma or adenocarcinoma is very important because it suggests that miRNAs may have an important role to better understand the progression of the disease. With the time other miRNAs important in NSCLC progression have been discovered, such as the miRNA cluster 17-92 mentioned above .
The current diagnostic methods for the diagnosis of lung cancer, such as the analysis of the sputum, the X-rays or CT scanning, are often not able to discriminate between SCLC and NSCLC or between adenocarcinoma and cell carcinoma . For this reason new molecular markers, such as the miRNAs, may help for the early diagnosis, prognosis and choice for the treatment for each individual patient, thus opening the new era of the “personalised medicine”.
Several studies have also demonstrated the association between specific miRNAs and the most frequent somatic mutations in lung cancer patients . Therapies are available for the treatment of people with specific mutations, such as EGFR, but very often the patients develop a resistance to the therapy and this is normally due to secondary mutations in the same or other genes . In vivo experiments have demonstrated that the treatment of the sensitive lung cancer cells with EGFR inhibitors leads to the downregulation of miR-30b-30c and miR221-222 and this causes the increase of the apoptosis. However, at the same time, MET overexpression upregulates those four miRNAs, thus making the anti-EGFR treatment useless. Researchers propose using MET-targeting miRNA inhibitors to improve the sensitivity to Gefinitib, an EGFR inhibitor, in a xenograft model .
NSCLC patients were also enrolled in 2013 for the clinical trial with MRX34, the miR-34 mimic developed by Texas and the results of the study will be show near in the future .
1.9 The two forms of miR-140 in lung cancer
MiR-140-5p was first studied in craniofacial abnormalities and the platelet-derived growth factor-a (PDFG-A) was identified as direct target . The sequence of human pre-miR-140 is predicted based on homology to a miRNA from mouse and its expression was later verified in human [162, 163]. Pre-miR-140 was found to locate on chromosome 16 (16q22.1), in one intron of the WW domain containing E3 ubiquitin protein ligase 2 (WWP2) gene and processed to produce two mature miRNAs, miR-140-3p and 140-5p, with different seed sequences . As described in figure 6, miR-140-5p originates from the 5’arm of the hairpin of the precursor, while miR-140-3p from the 3’ arm. Both the forms of miR-140 are downregulated in cancer tissues, including lung, suggesting their role as onco-suppressors [164-166].
Figure 6. Pri-miR140 stem and loop structure. This miRNA sequence is predicted based on homology to a verified miRNA from mouse and its expression was later verified in humans .
Croce et al. demonstrated that most of miRNAs downregulated in cancers are located in cancer-associated regions or fragile sites and their lower levels are due to deletions in the loci from where they are transcribed . Both of the two forms of miR-140 have been found downregulated in lung cancer and, presumably, their downregulation may be due to mutations in 16q22.1 locus.
MiR140-3p and 140-5p are downregulated in lung cancer cells and several studies showed that these miRNAs not only have a negative role in tumour growth and metastasis but also suppress the migration and invasion of NSCLC cells by downregulating the expression of key target genes, such as Insulin-Like Growth Factor 1 receptor (IGF1R) regulated by miR-140-5p and ATP8A1 targeted by miR-140-3p [168, 169].
These two miRNAs may also influence the expression of other non-coding RNAs. MiR-140-5p promotes the expression of the nuclear long non-coding RNA Nuclear Enriched Abundant Transcript 1 (NEAT1) that is upregulated in different malignancies, including lung cancer [170, 171]. The long non-coding RNA Metastasis Associated Lung Adenocarcinoma Transcript 1 (MALAT1), identified for the first time in lung cancer, influences miR-140-5p levels in glioma tumours .
Starting from these premises, in this study both of the two forms of miR-140 were investigated.
1.10 Importance of the study of both the two forms of miR-140 in lung cancer
In the classical model of miRNA biogenesis, the miRNA duplex is produces by the cut of the hairpin precursor by the endoribonuclease Dicer. Then, through some mechanisms that are largely unknown, one of the strand acts as a guide for the mature miR-mediated silencing complex (miRISC). This strand selection process is finely regulated and may vary among tissues, during the development and diseases. Recent studies have demonstrated that both the sister strand may be accumulated, thus leading to the hypothesis that the strand selection might not be tissue-specific . Moreover, mutations in the precursors may affect the strand selection, especially if they cause a change in the seed sequence of the mature strand, because the mutated miRNA may affect different mRNAs. In the case of the pre-miR-146a, for example, a SNP leads to the maturation of three active strands, two of which generate from the passenger strand and contribute to the development of thyroid cancer .
MiR-140-3p and miR-140-5p were identified in congenital malformations. Together with miR-17-92 cluster and miR-200b, they both play an important regulatory role during all the stages of palatogenesis and mutations in their precursor impairs their regulatory functions, thus leading to some serious developmental defects .
It is known that the de-regulation of some miRNAs contributes significantly to the loss of regulation of key pathways involved in cell homeostasis, thus allowing the growth of cancer. New studies suggest that a differential strand selection may be one of the mechanisms used by many tumours, including hepatocellular carcinoma, breast and gastric cancer, to repress the physiological regulation and promote its survival . It has been demonstrated in vitro and in vivo, that the levels of the mRNA targets may influence the abundance of their regulatory miRNAs [174, 175] and this makes the arm expression preference investigation more complicated and confirms that the thermodynamic hydrogen bonding theory is not the only mechanism that influences miRNA biogenesis.
In this study, both of the two forms of miR-140 will be investigated to see whether there is a preferential strand selection and if this may affect the progression of the disease.
1.11 Aims and hypothesis
The hypothesis of this study is that the two forms of miR-140 reduce the progression of NSCLC by targeting specific mRNAs.
• To verify if miR-140-3p and miR-140-5p are differentially expressed in NSCLC and normal cells by using tissues from patients and cell lines.
• To investigate if there is any strand selection during the biogenesis of miR-140-3p and miR-140-5p and whether this plays any role in the progression of NSCLC.
• To evaluate the effect of both miR-140-3p and miR-140-5p on cancer cell behaviour. This include the study of proliferation, migration, invasion, adhesion to the key the extracellular matrix (ECM) proteins, cell cycle, apoptosis and metabolism of cancer cells following the treatment with specific miRNA mimics.
• To investigate the effect of the treatment with miR-140-3p and miR-140-5p mimics on the tubule formation ability of the endothelial cells in vitro.
• To identify novel targets of miR-140-3p and miR-140-5p that are crucial for the changes on cancer cell behaviour following the treatment with miR-140-3p and miR-140-5p mimics.
• To dissect the molecular mechanisms in which the two forms of miR-140 are involved.