|Home | About | Journals | Submit | Contact Us | Français|
MicroRNAs regulate self-renewal, differentiation, and division of cells via post-transcriptional gene silencing. Aberrant microRNA levels, specifically an overall downregulation, are present in many cancers, as compared to their normal tissue counterparts. Therefore, a potential therapeutic use of microRNAs is to correct these aberrant transcript levels involved in the signaling pathways of cancer. This review focuses on the current knowledge of microRNAs and their involvement with cancer cells and cancer stem cells. The methods currently being used to develop miRNA-based cancer therapeutics are examined, and the limitations halting further progress are also discussed.
Cell differentiation is regulated, in part, by a recently discovered class of molecules – microRNAs (miRNAs). These regulatory molecules can directly affect differentiation in both normal stem cells and cancer stem cells [1–3]. MiRNAs are 21–23 nucleotides long and act as regulatory molecules in eukaryotic cells by binding to a non-coding region within target messenger RNAs (mRNAs), namely the 3′-untranslated region (3′-UTR). Through this mechanism miRNAs regulate self-renewal, differentiation, and division of cells via post-transcriptional gene silencing . MiRNAs play important roles in many cellular processes such as development , stem cell division [3,6,7], apoptosis [8,9], disease [10,11], and cancer [12,13]. MiRNAs regulate gene expression by either inhibiting translation or promoting degradation of specific mRNA transcripts. An estimated 3% of human genes code for miRNAs, yet these miRNAs may regulate around 30% of the protein-coding genes . This suggests not only their importance in various regulatory pathways, but also their potential for manipulation. MiRNAs themselves have been shown to act both as tumor suppressors and as oncogenes, which promote tumor growth. In additional, aberrant miRNA levels, specifically an overall downregulation, are present in many cancers, as compared to their normal tissue counterparts . Therefore, a potential therapeutic use of miRNAs is to correct these aberrant transcript levels involved in the signaling pathways of cancer cells, especially cancer stem cells (CSCs).
A canonical pathway for miRNA biogenesis begins with transcription of the encoded genes by RNA polymerase II. The majority of human miRNAs is expressed from introns, the non-coding regions of the genome [16,17]. An intron of about 400 nucleotides is excised from the primary transcript and becomes the primary miRNA (pri-miRNA). The pri-miRNA is then processed by the RNase Drosha into hairpin loops about 70 nucleotides in length, forming the pre-miRNA. The pre-miRNA is then exported to the cytoplasm by the nuclear membrane protein Exportin-5. Once in the cytoplasm, the RNase Dicer completes processing, forming mature miRNAs [4,15] (Fig. 1).
In addition to the canonical pathway described above, miRNA maturation involves much more complicated regulations. The transcribed and excised pri-miRNA has the following features: a hairpin stem (HPS) averaging 33 base pairs, a terminal loop (T loop), and two single-stranded regions (SSR) - one upstream and one downstream of the hairpin [18–20]. The pri-miRNA is then cleaved into a pre-miRNA by a complex consisting of the RNase Drosha and the DiGeorge critical region gene 8 protein (DGCR8) . The DGCR8 acts as a molecular ruler for the precise site of pri-miRNA cleavage, which occurs 11 base pairs away from the junction of the single- and double-stranded RNA. Drosha’s two RNase domains then cleave the 5′ and 3′ arms of the hairpin, forming the pre-miRNA  (Fig. 1). Pre-miRNA formation by Drosha occurs co-transcriptionally and before splicing, regardless of whether the primary transcript is protein-coding or non-coding . While miRNA encoding introns are spliced more slowly than adjacent introns, Drosha cleavage does not inhibit splicing because the latter process does not require a continuous intron [17,22].
MiRNA biogenesis is also very specific, depending on the miRNA being processed. Variations at the transcriptional level include the following: miRNA genes can be transcribed by either RNA polymerase II or RNA polymerase III, each of which recognizes specific promoters and terminators and undergoes specific regulation [23–25]; expression can additionally be controlled by transcription factors such as c-Myc or p53 . There are also many variations in pri-miRNA cleavage. In some cases, Drosha cleavage is replaced by splicing if the intron is the appropriate size to form a pre-miRNA hairpin [27,28]. In other cases, pri-miRNAs recruit heterogeneous nuclear ribonucleoproteins  or growth factors  to mediate and enhance Drosha processing. Other proteins such as RNA helicases, double-stranded RNA binding proteins, and Ewing’s sarcoma proteins can also be involved, depending on the specific miRNA .
MiRNAs regulate gene expression by either inhibiting translation or promoting degradation of specific mRNA transcripts. Post-transcriptional gene silencing begins when an miRNA recruits RISC (RNA-induced silencing complex), which is a complex of proteins that localizes the miRNA to its complementary target mRNA . Although there are exceptions, the specificity of target localization involves both the miRNA and mRNA sequences. Nucleotides 2–8 of the miRNA, called the seed region (SR), must bind contiguously to a perfectly complementary sequence on the target mRNA. The binding sites for the miRNA seed region lie in the 3′-UTR of the mRNA, and the complementary sequence usually repeats multiple times within the 3′-UTR. Whether an miRNA promotes degradation or represses translation of its target mRNA likely depends on the degree of complementary binding beyond the seed region . More specifically, the silencing mechanism is likely dictated by the number, type, and position of mismatched base pairs between the miRNA and mRNA  (Fig. 1).
MiRNAs repress translation via different mechanisms either at the initiation or elongation step of translation, based on how they modulate the interaction between the 5′ cap and the 3′ polyadenylated tail of the mRNA . Although the exact mechanism is not well understood, translational repression involves cytoplasmic mRNA-processing bodies, or P bodies. These P bodies are believed to play a role in mRNA silencing because target mRNAs and RISC components are co-localized to this part of the cell [35,36]. P bodies house various proteins that control messenger ribonucleoprotein (mRNP) complexes [35–38]. These mRNP complexes consist of mRNA and repressor proteins, and they lack translation initiation factors . Thus, the localization of target mRNAs to P bodies, and subsequent formation of the mRNP complexes, inhibits translation. Although translational repression seems to be the default action of silencing , degradation of the mRNA appears to occur when the sequence beyond the seed region of the miRNA is perfectly complementary to the remaining region in the 3′-UTR of the mRNA. This degradation involves deadenylation, decapping, and exonucleolytic cleavage of the target mRNA transcript [40–42], but the exact mechanism is still unknown.
A tumor consists of a heterogeneous population of cells that differ by their relative states of differentiation . The outside of a tumor mass contains fully differentiated cells that are susceptible to radiation and chemotherapy because of their close vicinity to the non-tumorigenic microenvironment as well as the sufficient blood flow due to induced angiogenesis, or blood vessel growth . The region closer to the center of the tumor contains progenitor cells, which can undergo a limited number of mitotic cycles to form several daughter cells . These daughter cells can then differentiate into select types of cells based on their relative microenvironments . Within the heart of the tumor lies the cancer stem cells (CSCs), which are both structurally and functionally distinct from the other cells within a tumor mass .
CSCs have the ability to undergo self-renewing mitosis, where one or both of the daughter cells retain the identity of the stem cell. The other daughter cell could become a progenitor cell, which then undergoes several mitotic cycles to become a differentiated cell. It is through this mechanism that CSCs can cause proliferation and expansion of fully differentiated tumor cells [43,44]. CSCs are challenging to study because they rapidly differentiate when cultured; it is difficult to maintain a population in vitro that remains enriched in the CSCs . CSC-enriched populations exhibit three characteristics in vitro: they can be isolated with cell surface marker profiles, they form tumorspheres, which appear as spherical colonies in suspension cultures, and they have increased resistance to both chemotherapeutic agents and ionizing radiation . CSCs have been isolated for hematological malignancies , brain tumors [47,48], and colon tumors . A method for enriching breast CSCs has also been developed . As is often the case in research, further progress is limited by the available technologies and known techniques.
Cancer therapeutics in current use decrease tumor size, but they are unlikely to result in long-term remission unless the CSCs are also targeted. However, CSCs are difficult to target because of their innate properties. CSCs are characterized by their resistance to anti-cancer therapeutics and their repopulating ability; they can regenerate tumorigenic cells after tumor reduction from chemotherapy and radiation . The multidrug resistance trait of CSCs is associated with an overexpression of genes that code for transmembrane efflux pump proteins. The proteins themselves are regulated by reactive oxygen species within the cell, which could vary by tissue type and also among cells within a tissue . Thus, any therapeutics that target CSCs would require great specificity. A rapidly developing field in cancer therapeutics is targeting miRNAs, which exhibit aberrant levels in cancer cells. Functional studies of the specific miRNAs within different cancer cells, specifically CSCs, are necessary to find a therapeutic target.
Since miRNAs are critical for both stem cell development and cancer pathogenesis, they are being examined for their regulatory roles in self-renewal, proliferation, and differentiation of cancer cells [2,51,52]. MiRNAs have been shown to act as both tumor suppressors, which help control growth, and oncogenes, which promote rapid growth . Global inhibition of miRNA processing increased tumorigenicity and transformation in a recent study , which suggests their important regulatory role. In addition, many cancer-associated regions of the genome contain miRNA genes . Aberrant miRNA levels, specifically an overall downregulation, are present in many cancers, as compared to their normal tissue counterparts . Specific miRNA expression patterns have been described in several cancers (Table I). Although it is not entirely clear if irregular miRNA expression is a cause or an effect of the tumorigenic state, the significance of these regulatory molecules in cancer is apparent.
As tumor suppressors, decreased expression of some miRNAs in cancer stem cells results in cell proliferation. These miRNAs normally regulate apoptosis, differentiation, and self-renewal . One example of these miRNAs is the miRNA-34 (miR-34) family, consisting of miR-34-a, b and c, which is down-regulated in many cancers  (Table I). Restoration of miR-34 in p53 deficient gastric and pancreatic cancer cells resulted in both cell cycle arrest at G1 and apoptosis suggesting that this miRNA is involved in p53 function [26,56,57]. Indeed, in cell lines and mouse tissue, p53 regulates the abundance of all three miRNAs of the miR-34 family, which inhibit a group of mRNAs that normally support tumor formation by inhibiting apoptosis; promoting cell cycle progression past the G1 checkpoint; preventing cellular aging; and promoting migration (Fig. 2). Interestingly, p53 normally induces growth arrest by activating p21, a cyclin-dependent kinase (cdk) inhibitor . However, miR-34 can induce G1 arrest independently of p21 in some cell types . Because of its regulation and function, the miR-34 family probably has a significant effect on p53 tumor-suppression function .
Because of mutant p53 in over 50% of human cancers, the miR-34 family is often down regulated in tumors [26,59]; however, its abundance can also be affected at the genomic level. In vertebrate genomes, one genomic locus encodes miR-34a and another locus encodes miR-34b and miR-34c. Even among closely related species, there is little conservation in these genes except in the miRNA-encoding sequences and in the short promoter proximal regions that contain a p53-binding site [26,57,59–61]. The miR-34a gene is commonly lost in many human tumor types, including neuroblastoma  and pancreatic tumors . Similarly, aberrant miR-34b and miR-34c expression has been observed in a subset of non-small cell lung cancers . The miR-34a targets the mRNA of an important transcription factor, E2F3, which binds specifically to retinoblastoma protein pRB in a cell-cycle dependent manner and initiates production of proteins that affect cell cycle checkpoints, DNA repair, and replication [63–65]. Reduction of E2F3 level by artificial elevation of miR-34 in primary neuroblastomas inhibited cell proliferation and activated cell death pathways , providing further evidence for miR-34’s role as a tumor suppressor.
Besides E2F3, the miR-34 family also targets Notch, HMGA2, CDK4, CDK6, Cyclin E2, and Bcl-2 gene products, which are involved in self-renewal and survival of cancer stem cells because of their effects on cell cycle control, apoptosis, and DNA repair [26,59,66] (Fig. 2). Bcl-2 is over-expressed in a majority of human cancers , and this increased expression has been correlated with both chemotherapeutic and radiation resistance of the CSCs [68–71]. Bcl-2 protects cancer cells from chemotherapeutic agents by preventing apoptosis [72,73]. Restoration of miR-34 levels in pancreatic and gastric cancer cells sensitized the cells to chemotherapy, inhibited tumor growth, and inhibited tumorsphere formation [74,75]. However, the tumor suppressing function of artificial elevation of miR-34 appears to be cell type dependent. While restoration of miR-34 expression has resulted in cell cycle arrest or senescence in some studies , it has resulted in apoptosis in others; apoptosis also decreased with inhibition or depletion of miR-34 [59,61]. Therefore, more studies are needed to fully understand the effects of therapeutic restoration of miR-34 on different cell types, because of the great complexity in the pathways targeted.
Both normal and cancer stem cells have the ability to divide symmetrically, creating two identical daughter cells. They also have the ability to divide asymmetrically, forming one stem cell and one progenitor cell, the latter of which will undergo differentiation . This asymmetric division facilitates healthy growth in normal cells due to the polarity involved in cell division. When this polarity is lost, the stem cells multiply and form a tumor . Stem cell self-renewal is controlled by both intercellular mechanisms, via signaling from neighboring cells, and intracellular mechanisms, involving differential gene expression that is under epigenetic, transcriptional, translational, and post-translational control . For example, the polarity of cell division in mammary stem cells is regulated by many genes including tumor suppressor p53. Loss of p53 results in more symmetric divisions and leads to tumor growth . In non-tumorigenic cells, regulatory miRNA levels are generally lower in cells that are less differentiated (i.e., stem cells) . In adult stem cells, miRNAs regulate hematopoiesis, myogenesis, cardiogenesis, astrocyte/neuronal differentiation, osteogenic differentiation, and skin differentiation . Since miRNAs play an important regulatory role in normal differentiation, aberrant expression could cause significant changes within the cell, possibly leading to cancer.
One extensively studied model for miRNA-regulated stem cell differentiation is the maturation of blood cells. Normal hematopoietic stem cells differentiate into several cell types with myeloid and lymphoid lineages, and certain miRNAs have been shown to regulate various differential steps (Fig. 3). In normal hematopoiesis, long-term reconstituting hematopoietic stem cells divide into their short-term counterparts, which then give rise to multipotent progenitors. These multipotent progenitors differentiate into either lymphoid stem cells or myeloid stem cells; this differentiation step is inhibited by miR-128a and miR-181a, but it is activated by miR-223 [80,81]. Furthermore, the differentiation of lymphoid stem cells into common lymphoid progenitors is inhibited by miR-146 and activated by miR-181 . Common lymphoid progenitors then differentiate into T cells, B cells, and natural killer cells, the first of which is activated by miR-150 . On the other hand, the differentiation of myeloid stem cells into common myeloid progenitors is inhibited by miR-155, miR-24a, and miR-17 . Common myeloid progenitors differentiate into either granulocyte-macrophage progenitors, the formation of which is inhibited by miR-16, miR-103, and miR-107, or megakaryotic-erythroid progenitors . Granulocyte-macrophage differentiation into granulocytes is inhibited by miR-223, while differentiation into monocytes is inhibited by miR-17-5p, miR-20a, and miR-106a [80,83]. Megakaryotic-erythroid progenitors differentiate into either megakaryocyte progenitors, which develop into platelets, or erythroid progenitors, the formation of which is inhibited by miR-24 . Erythroid progenitors then develop into red blood cells via activation from miR-451 and miR-16, or this process is inhibited by miR-150, miR-155, miR-221, and miR-222 [80,82,85–87] (Fig. 3).
Like non-tumorigenic stem cells, CSCs contain low miRNA levels , but it is still unknown if this is the cause or the effect of the differentiation state. Some miRNAs suppress proliferation in differentiated cells through tumor suppressor pathways, suggesting miRNAs are the cause of differentiation state . On the other hand, specific miRNAs that are characteristic for normal tissue stem cells are present at the same level in CSCs . However, aberrant miRNA expression levels, both an upregulation of some miRNAs and a downregulation or absence of others, have been found in CSCs . Certain cancers have specific miRNA profiles (Table I) as well as specific membrane proteins served as surface markers. In comparison to the profiles of other cells within the tumor, cell surface marker profiles for CSCs have been characterized for various tumor types including acute myeloid leukemia , adult glioblastoma multiforme , brain tumors , breast cancer , bone sarcoma , chronic myeloid leukemia , colorectal cancer [49,94], head and neck squamous cell carcinoma , lung cancers , metastatic melanoma , pancreatic cancer , and prostate cancer [99,100]. With this knowledge, therapies can be targeted for delivery to cancer cells, especially cancer stem cells. In order to develop therapeutic applications that target CSCs, the functional role of the miRNAs in the specific tumor would need to be established.
Therapeutic microRNAs can be introduced into tumor cells to correct the aberrant miRNA levels, which could potentially result in reversal of the some of the cells’ tumorigenic properties. These therapeutics are aimed to silence tumorigenic genes post-transcriptionally via the RNA interference (RNAi) pathway, in which a small RNA causes degradation of its complementary mRNA . Cells use endogenous miRNAs for RNAi, as described in previous sections. Therapeutic RNAi can be utilized with two molecules: synthetic small interfering RNA (siRNA) and vector-based short hairpin RNA (shRNA). SiRNAs, like miRNAs, are double-stranded, are 21–23 nucleotides long, contain two-nucleotide overhangs at the 3□ ends, and can be incorporated directly into RISC. ShRNAs are delivered in vectors, so they are actually synthesized in the nucleus like miRNAs . ShRNAs are stem loop structures, similar to pre-miRNA hairpins, and undergo the same processing mechanisms as miRNAs, including Dicer processing before RISC incorporation [21,103–108]. There are practical advantages and disadvantages to the small RNA therapies. SiRNA is easy to chemically modify, but bulk production of complex structures is more costly. On the other hand, shRNA is difficult to modify and easy to produce in bulk. However, there are methods for shRNA alterations, which involve changing the inserted sequence, altering promoter regulation, using different plasmids or viral vectors for delivery, and regulating or inducing shRNA expression [109–111]. Both siRNAs [112–114] and shRNAs [112,114,115] have demonstrated effectiveness in vivo.
RNAi technology has increased the potential of targeted therapies. Several animal studies have shown the therapeutic efficacy of RNAi, and some RNAi-based therapeutics are currently undergoing clinical trial testing for treating cancers . The therapeutic efficacy of siRNAs and shRNAs varies, depending on several factors: immune system avoidance, silencing efficiency, and off-target effects . SiRNAs longer than 29–30 nucleotides induce an interferon immune response similar to double-stranded RNA, but shortening the strand or making structural modifications virtually eliminates this effect . Both siRNA and shRNA also induce Toll-like receptors (TLRs) and downstream immune responses; however, this effect can be effectively diminished by using minimal plasmid vectors or unmethylated CpG-free plasmid vectors . SiRNA, even when protected with chemical modifications, is susceptible to degradation and metabolic processes. While shRNA shows desired target effects when present in fewer than five copies, siRNA requires doses in the nanomolar range to achieve the same effects. This is a result of shRNA undergoing continual synthesis, yielding less transient effects .
On the other hand, RNAi-based therapeutics have many off-target effects that impact gene expression, resulting in off-target phenotypes. This is due to the fact that such small RNAs can bind some mRNA transcripts with imperfect complementarity . SiRNAs, like miRNAs, have a seed region sequence of nucleotide positions 2–8 that complementarily bind to the 3′-UTR of mRNA . Chemical modifications, especially at position 2, can greatly improve target specificity . However, the large dose of siRNA required increases the probability of off-target effects . When comparing siRNA and shRNA treatments of similar doses, shRNA had fewer off-target effects, likely due to the different mechanism of action . ShRNAs face the same nuclear and cytoplasmic regulation as miRNAs, while siRNAs do not . Currently, shRNA sequences are being made into artificial pri-miRNA transcripts .
Since miRNAs are critical for both stem cell development and cancer pathogenesis, they are being examined for their regulatory roles in self-renewal, proliferation, and differentiation of cancer cells [2,51,52]. Global inhibition of miRNA processing increased tumorigenicity and transformation , which suggests their important regulatory role. MiRNAs have been shown to act as both tumor suppressors, which help control growth, and oncogenes, which promote rapid growth . In addition, many cancer-associated regions of the genome contain miRNA genes . Certain cancers have specific miRNA profiles (i.e., upregulation or downregulation of certain miRNAs) as well as specific membrane proteins called surface markers. This knowledge enables development of therapeutics targeted to cancer cells, including cancer stem cells. Functional studies of the specific miRNAs within different cancer cells, specifically CSCs, are needed to find a drug target and develop effective and safe therapeutics.
This work is partially supported by RO1 CA086928 (to S. W.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest Statement