MicroRNA-mediated posttranscriptional mechanisms of gene expression in proliferating and quiescent cancer cells

Abstract

MicroRNAs are small non-coding RNA regulators of gene expression that play important roles in critical biological processes, including cell division, self-renewal and cell state maintenance. Their deregulation leads to extensive clinical consequences in tumorigenesis. Cancers demonstrate heterogeneity in their cell states implicated in their resistance and resurgence. Apart from proliferating cells, cancers harbor a small proportion of assorted quiescent cells that resist conventional therapeutics and contribute to cancer recurrence. MicroRNA expression, targets, microRNPs (microRNA-protein complexes) and their functions have been demonstrated to be regulated in distinct tumor cell states and as an adaptive response to stress signals in tumor-unfavorable environments. In turn, altered microRNPs and their modified post-transcriptional mechanisms of gene expression may contribute to tumor resistance and influence tumor progression. An understanding of distinct microRNA mechanisms in cancer cells would provide extensive insights into the versatile roles of microRNAs in the perpetuation of tumors and indicate potential therapeutic avenues.

Introduction

MicroRNAs

MicroRNAs are a unique class of small, non-coding, 20- to 24-nt regulatory RNAs that modulate gene expression by recognizing target RNAs via base‐pairing, usually in the untranslated regions (UTR) of mRNAs, and guiding effector microRNPs (microRNA-protein complexes) in a controlled manner.¹ Often highly conserved and bioinformatically predicted to target a large range of genes involved in cell division, self-renewal and maintenance of cell states,^2-6 their deregulation or aberrant function leads to immense clinical consequences ranging from immune and developmental disorders to cancers.^7-11

MicroRNAs and microRNPs are intricately connected with tumorigenesis.^10-13 Typically found aberrantly overexpressed or deleted in cancers, microRNAs can function as tumor suppressors or oncogenes to control critical growth regulators and thereby, a network of downstream events to orchestrate multiple effects on cancer progression. Conventional oncogenes and tumor suppressors function as transcriptional regulators of pathways for tumor progression by controlling the expression of microRNAs as in the case of the miR-17–92 cluster, regulated by the oncogene, c-myc, which in turn results in suppression of tumor suppressors such as PTEN and p21CIP.^7,13-17 Several microRNAs play critical roles in controlling the cell cycle and thereby influence tumorigenesis. For example, miR-16 was discovered in a chromosomal region deleted in 68% of chronic lymphocytic leukemia; subsequent analyses revealed that the miR-16 family of microRNAs controlled expression of several cell cycle factors and of the anti-apoptotic factor, Bcl2, and thereby, influenced cell cycle arrest and apoptosis.^11,12 MicroRNAs have been observed to be stable clinical markers in tissue samples and may enable distinction between cancer subtypes, tissue-specific cancers, and different cancer cell states.¹⁸ In addition, circulating microRNAs have been identified in several malignancies and are suggested to influence the tumor microenvironment, surrounding non-cancerous and immune cells and contribute to tumor progression.^19-21

Quiescent Cell States in Cancer

Several studies indicate that cells that survive clinical therapy include dormant, quiescent (G0)-like cells^22-25 observed as a small (~1% or less) but clinically relevant population in leukemias and in certain solid tumors with poor survival rates. Quiescence or G0 is a unique, adaptive, non-proliferating state that provides an advantageous escape from harsh situations and chemotherapy that lead to permanent outcomes of tumor-negative environments: senescence, differentiation or apoptosis.^26-30 Instead the cell is suspended reversibly in an assortment of transition phases (called G0 state for simplicity) where it can develop and respond adequately to more favorable conditions by returning to the cell cycle.²⁶ Quiescent cells play critical roles in cancer including:

1. Quiescent side populations within tumors. Quiescent cancer cells, including some dormant cancer stem cells in the G0 state, can be isolated as a side population from heterogeneous tumors under conditions such as serum-starvation, hypoxia, specific signaling interference and upon chemotherapeutic ablation of dividing tumor cells^26-28,30-32 and contribute to tumor heterogeneity, resistance and recurrence.

2. Quiescent disseminated and circulating tumor cells. Quiescence is observed among disseminated and circulating tumor cells that evade therapy and the immune system by virtue of their quiescent state to eventually extravasate into tissues and establish metastasis^33-35

3. Quiescent tumor associated macrophages (TAMs). Tumor associated macrophages are a heterogeneous assortment of quiescent and activated cells that can produce cytokines and other factors to promote the tumor environment, support establishment of metastasis and induce/maintain quiescent tumor cells, enabling their chemoresistance.^36-41

A cell exiting from active proliferation has four non-proliferating options: senescence, apoptosis, differentiation and quiescence, the latter having the advantageous characteristic of reversibility observed in quiescent cancer cells, immune cells, wound healing and germ cells.^26-28,42-44 Quiescent states are transient and distinct from G1 and G1 arrest, senescence, differentiation and other states of the cell cycle.^26-28,42-45 A block to cell division by several methods in cancer cell lines leads to cell death, G1 arrest or senescence rather than quiescence in cells lacking the capacity to enter the quiescent state, which, in the absence of specific tests and markers to distinguish G0 from G1 arrest and senescence, can misleadingly suggest G0.^{26,42-44,46-49} Additionally, quiescence is also influenced by cell to cell contact, where replating proliferating cells at high density inhibited entry into quiescence and related microRNA functions.^26,43-45 Therefore, true quiescence and associated microRNA functions in G0 cancer cells^46-49 is transiently and less frequently observed. MicroRNAs are required to maintain quiescence in tissues like muscles⁵⁰ and are involved in maintaining G0 side populations in cancers.^2,51-53

Differential characteristics involved in microRNA mechanisms between quiescent and proliferating cells

Such reversibly arrested cells are, in contrast, metabolically quite active although distinct from proliferating cells.^27,54 G0 cells demonstrate a switch to expression of specific genes that resist tumor-negative conditions, differentiation and cell death to maintain the quiescent state and secure the abilities of the cell to be restored to the proliferative state.^27,54 Maintenance of the G0 state can extend from a few hours to several months.^26-28,43,44 Quiescent cancer cells have different expression profiles from their proliferating neighbors due to transcriptional differences but also in part due to post-transcriptional microRNA-mediated mechanisms.^2,3,51-53,55 In G0, many mRNAs have extended 3′- UTRs, thereby, including microRNA and RNA binding protein regulatory sites.⁵⁶ Additionally, RNA binding proteins are altered, unmasking previously occupied 3′-UTR microRNA target and other RNA binding regulatory sites.⁵⁷ The G0-specific included or opened microRNA (and other RNA binding protein sequences like AU-rich elements^46,58) target sites alter gene expression,⁵⁶ which may play essential roles in tumor progression. Apart from altered mRNA UTRs and UTR-associated RNPs (RNA-protein complexes) that reveal microRNA and other regulatory target sites, G0 states express distinct microRNAs,^2,51-53 factors and microRNPs as well as mRNA regulatory mechanisms that are associated with specific gene expression changes,^26,27,54 indicating that post-transcriptional regulation plays an important role in maintaining the state^{26,42,46-50,54,59} (Fig. 1).

Figure 1. MicroRNA expression and mechanisms are modulated by cancer cell states and stress signals in tumors, leading to misregulation of translation, localization and stability/decay of mRNAs, which influences tumor progression and resistance. MicroRNAs are generally transcribed as pri-microRNA, which are polyadenylated and capped polymerase II transcripts. The pri-microRNAs are processed by the nuclear RNase III enzyme, Drosha, into 70nt pre-microRNAs that are exported to the cytoplasm by Exportin-5. The RNase III enzyme, Dicer, further processes the pre-microRNAs into double-stranded microRNAs that are transferred to Argonaute (AGO). The passenger strand is removed and mature single-stranded microRNAs are incorporated into AGO complexes, forming the microRNP or RISC.^66-70 Proliferating and quiescent cancer cells as well as other cancer cell states and distinct types of cancer can alter the composition of associated RNPs for specific microRNAs and mRNAs, which regulate microRNA levels and functions and lead to distinct gene expression outcomes: (a) Processing of microRNAs can be positively or negatively regulated by several RNA binding proteins (RNA-BP) such as KSRP (promotes processing), hnRNP A1 (promotes or represses processing) and Lin28A/Lin28B (inhibitory) by influencing Drosha and Dicer processing.^71-81 (b) GW182 interaction with AGO2 elicits repression and downregulation^{97,103,105,107-109} while GW182 interactions with AGO are altered and reduced in quiescent cells^112-115 leading to loss of repression¹²⁴ and activation of specific mRNAs.^{46-49,119,123,125} (c) The availability of the microRNA target site (microRNA site) on mRNAs can be regulated by alternative polyadenylation, which leads to shortened 3′-UTRs in proliferating cells and therefore, escape from microRNA-mediated control over gene expression; in quiescent cells, longer 3′-UTRs are favored and gene expression is subject to greater regulation by microRNAs and RNA binding proteins.^56,62-64,118 (d) RNAs can function as decoys in response to different cancer cell state conditions to activate expression: by microRNAs themselves functioning as RNA decoys to remove repressive RNA-binding proteins from binding and repressing mRNAs via RNA binding sites (RNA-BP site) in distinct cancer cell states¹³³ or by non-coding RNAs such as pseudogenes that bear similar target sites and sequester the microRNAs from their target mRNAs in different cancer subtypes.^135-137 (e) RNA-BPs can regulate microRNA functions both negatively and positively by competing with microRNAs for the same binding sites and by preventing or promoting microRNA functions by regulating target site access.^{57,89,118,138-141} (f) RNA and protein components of microRNPs can be modified in specific cancer subtypes: through addition of U/A nucleotides at the 3′-ends of microRNAs, which leads to altered microRNA levels and functions^81,143-149; by stress-induced addition of poly(ADP) ribose to AGO2/AGO by PARP, which leads to relief of repression and cleavage and thereby, expression of derepressed target mRNAs.¹³⁴

Regulated Expression of microRNAs

MicroRNAs are generally transcribed as pri-cursor (pri-microRNA) polyadenylated and capped polymerase II transcripts. The pri-microRNAs are processed by the nuclear RNase III enzyme, Drosha, of the Microprocessor complex into approximately 70nt pre-cursors (pre-microRNA) that are exported by Exportin-5 to the cytoplasm. The RNase III enzyme, Dicer, further processes the pre-microRNAs into double-stranded 22-24mers and mature, single-stranded microRNAs are incorporated into Argonaute (AGO) complexes to form the microRNP or RNA-induced silencing complex (RISC, Fig. 1A).^66-70 The conserved terminal loop region of pre-/pri-microRNAs acts as a key regulatory switch for modulating microRNA levels by recruiting distinct RNA binding proteins, dependent on their relative levels in different cancers and cell states^66,70-81 including: heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) that can promote or abrogate maturation of specific microRNAs,⁷¹ the AU-rich element binding mRNA decay factor, KH-type splicing regulatory protein (KSRP) whose misregulation can promote cancer through loss of microRNA expression⁷² and Lin28 proteins that inhibit processing at multiple levels^73-81 to maintain the undifferentiated cell state and are overexpressed in several, distinct cancers.⁸¹

Distinct microRNA signatures are recognized as potential markers of hypoxia, tissue-specific cancers and cancer subtypes that can be correlated with clinical features, suggesting that microRNA expression is specifically regulated in cancers.^10,11 The cell cycle and cell density, which influences the cell cycle state, can directly affect the levels and functions of microRNA processing associated factors such as TRBP and additionally regulate specific microRNA levels by mediating their decay upon cell cycle and cell density changes.^82-84 MicroRNA processing related factors such as ARS2, a component of the Cap-binding complex, is essential for proliferation and for primary microRNA processing of several specific microRNAs implicated in transformation in proliferative cells where ARS2 is expressed.⁸⁵ Apart from direct and specific effects of microRNA levels on distinct tumors, mutations in microRNA processing and regulatory factors such as Dicer (loss of function or deletions), Drosha (a haplotype of 2% frequency associated with increased lung cancer survival), Exportin-5 (a dominant negative mutant present in some cell lines with microsatellite instability) and Lin28 (overexpression leads to decreased let-7, a tumor suppressor) also correlate with tumors of distinct types, either due to effects on the levels of families or groups of microRNAs in these tissues or due to a global loss of microRNA homeostasis, leading to deregulation of both oncogenic and tumor suppressive functions.^81,86-88

Altered mRNA 3′-UTRs

In tumors, the 3′-UTR lengths are altered either by chromosomal breaks^60,61 and mutations or by alternative polyadenylation,⁵⁶ thereby disrupting microRNA-mediated control over gene expression, enabling oncogene expression and thus uncontrolled proliferation. A widespread effect of cell state differences between proliferation and quiescence on mRNAs is of alternative polyadenylation leading to preclusion of regulatory sites for RNA binding proteins and microRNAs, which along with a general decreased expression of specific microRNAs in cancers, are predicted to lead to deregulated expression of oncogenes in proliferating cells^56,62-64 (Fig. 1B). Increased levels of specific 3′-end processing factors were observed in proliferating cell states compared with their lower levels in non-proliferating and quiescent states, likely enabling greater use of the proximal, weaker, cleavage and polyadenylation sites on mRNAs in proliferating cells and thereby, shortening of the 3′-UTR.^56,62-64 Such alternative polyadenylation has been used to distinguish different cancer subtypes.^62,63 Mechanisms that alter the inclusion of target sites in mRNAs provide greater opportunity for the diversity in gene expression observed in heterogeneous cancers and may contribute to persistent populations in tumors and tumor subtypes. The ABCG2 transporter gene, critical for chemoresistance, was found to be alternatively polyadenylated forming shorter 3′-UTRs that were overexpressed in a resistant subtype of colon cancer cells; such cells were resistant in comparison to the parent colon cancer cell line, which expressed the longer 3′-UTR isoform and thereby, included target sites and associated microRNA-mediated regulation.⁶⁵ Variation in expression of such critical, resistance factors and thereby, differential chemoresistance within a tumor population and among tumor subtypes can thus be attributed at least in part, to shortened 3′-UTRs that exclude regulatory target sites and abrogate regulation by microRNAs and other RNA regulatory elements.

Regulated Mechanisms

Apart from altered targets and microRNAs, microRNP factors as well as microRNA-mediated mechanisms are altered in quiescent cells (Fig. 1). Unlike quiescent cells and cells cycling slowly when replated/restarted at lower densities, where several cases of indirect and direct upregulated gene expression have been observed,^89,90 high density populations of replated/restarted proliferating cells (cycling cells at high density that are not quiescent) demonstrate increased processing of specific microRNAs and enhanced repression/downregulation.⁴⁵ These processes suggest that the microRNA population, mechanism and target sites on mRNAs are coordinately manipulated to maintain the cancer cell state: proliferating cells promote oncogene expression and prevent expression of tumor suppressors by precluding microRNA target sites to enable oncogene mRNA expression, blocking the expression of tumor suppressive microRNAs (such as let-7)⁵ that repress oncogenes as well as by enhancing the levels of oncogenic microRNAs (such as miR-17–92)⁵ that repress tumor suppressor genes. Importantly, the quiescence and proliferation-induced microRNP characteristics also provide a unique opportunity to target these distinct populations of cancer cells. For example, downregulation of the ABCG2 transporter by overexpression of miR-520h diminishes the presence of tumor side populations along with resistance and invasion typical of micrometastatic cancer cells.⁵⁵ Therefore, an investigation of distinct, regulated mechanisms of microRNA expression and functions in proliferating and quiescent cancer cell states and in response to stress signaling induced in tumor environments, discussed in this review, would enable a greater understanding of the role of microRNAs in the persistence and progression of cancers.

Conventional and alternative mechanisms of microRNA-mediated regulation of gene expression

MicroRNAs act as targeting molecules via their sequence-specific patterns of base-pairing and guide associated effector RNP complexes to other target RNAs to dictate functional outcomes of target gene expression. The ratio of microRNA to mRNA target levels, which is manipulated by cellular conditions, influences the amplitude of the gene expression effect and range from fine-tuning of gene expression to significant changes. Additionally, the expression and levels of different microRNA families may be regulated by multiple signals, which may produce a combinatorial regulatory effect on many target mRNAs that bear multiple microRNA sites, thus relaying distinct signaling cues and adding greater versatility to gene expression control.^1,10,91 The AGO or eukaryotic initiation factor 2C (eIF2C) family of effector proteins associates with microRNAs to form the functional microRNP, which is directed to target RNAs by microRNA base-pairing to the target RNA/mRNA. Complete base-pairing of the microRNA to its target leads to mRNA cleavage, degradation and repression determined by the functional features of the AGO member present in the microRNP.^1,10,92,93 Depending on the characteristics of the microRNP-associated effector proteins, partial base-pairing to target sites by the microRNP leads to either the traditional outcome of downregulation/silencing of gene expression by mRNA deadenylation^94,95 and translational repression^96,97 or to alternative mechanisms that promote gene expression⁸⁹ (Fig. 1C). Mechanisms that promote gene expression include direct upregulation of specific messages by microRNAs, microRNA-mediated decoy of translation repressors and indirect gene expression by abrogation of conventional microRNA silencing or relief of repression^48,89,96,97 (Fig. 1C, d-f).

Posttranscriptional downregulation by microRNAs

MicroRNAs or siRNAs have predominantly been observed to cause downregulation or silencing by cleavage, repressed translation or deadenylation of the mRNA.^92,93,96,97 The key factor, AGO2, bears an RNA-dependent RNase H domain that enables cleavage of completely complementary mRNA sites usually to siRNAs⁹² but also to some naturally occurring microRNAs such as miR196 that is encoded within the A, B and C clusters of HOX8 genes and downregulates several HOX mRNAs with conserved, extensive complementarity.⁹⁸ Incomplete complementarity is usually associated with other post-transcriptional mechanisms, predominantly deadenylation and repression.^69,96,97

MicroRNAs can mediate repression prior to deadenylation of their messages^94,95,97 and non-polyadenylated transcripts are also repressed⁹⁵ while deadenylation was demonstrated to be a widespread effect of microRNAs on gene expression^99,100 and ribosome profiling of polyadenylated mRNAs demonstrated deadenylation in the case of several microRNAs.¹⁰¹ Target mRNAs may also be relocalized to storage/decapping bodies called P bodies.^96,97 MicroRNAs have been demonstrated to repress translation by several mechanisms including translation initiation, post-initiation/elongation and nascent peptide turnover.^96,97,102

GW182

The GW182 family of proteins is conserved from yeast to mammals and is a critical microRNP-associated factor for mediating silencing. GW182 interacts with the AGO family of proteins and with the effector complexes that mediate repression and deadenylation in mammalian cells. AGO2 recruitment of GW182 was demonstrated to be essential for downregulation by deadenylation and translation repression in mammalian cells.^103-107 Tethering GW182 is sufficient for repression and deadenylation suggesting that GW182 is a primary effector of repression recruited by AGO/microRNAs to specific targets.^103,108,109 GW182 can interact with poly(A) binding protein (PABP) and can recruit the CCR4-CAF1 and PAN2-PAN3 deadenylase complexes directly, as well as the PABP regulatory E3 ubiquitin ligase, EDD¹¹⁰ and mediates deadenylation^106,107 and repression.^109,111 GW182 was demonstrated to mediate both deadenylation/PABP -dependent and -independent repression.^109-111

Regulation by the cell cycle and density in proliferating cells

Core microRNP silencing factors as well as associated factors appear regulated by cell cycle states. GW bodies appear regulated across the cell cycle^112-114 and are reduced in quiescence, predicting an altered role in quiescent and quiescent-like cells.^46,49,115 GW bodies are not required for downregulation but are considered a consequence of the downregulation functions of microRNPs.¹¹⁶ GW bodies marked by GW182 but not P bodies marked by decapping enzymes were found to decline in G1, increase in S and G2 phases and appear reduced and diffused in G0 cells.^46,112-114 Repression is observed to be enhanced with synchronized replated cells in the S/G2 phases of the cell cycle⁴⁷ and with replated/restarted high density cells that are actively growing within the cell cycle (but not quiescent or low density cells that are cycling slowly in the absence of cell to cell contact)^45,47 indicating regulation by the cell cycle and cell density.

The RNA binding Pumilio 1 (Pum1)¹¹⁷ promotes microRNA repressive functions, in part by altering target mRNA structural conformation, thereby enabling access to repressive mechanisms like microRNAs. Pum1 binds specific target sequences on mRNAs, such as p27KIP1, in a cell cycle regulated manner. The Pum1-induced conformational alteration of the mRNA promotes microRNA access and consequent repression of p27 mRNA, exclusively in cycling cells but not in arrested cells.⁵⁷ Another target of Pumilio is the oncogene, E2F3, an important transcription regulator that functions as a powerful driver of proliferation and is overexpressed in many cancers. Pumilio represses E2F3 directly as well as enhances the activity of many microRNAs targeting E2F3. In cancers like bladder cancer and multiple tumor cell lines, such microRNAs are selectively downregulated or the 3′-UTR of E2F3 is shortened by alternative polyadenylation to remove the Pumilio sites and thus prevent Pumilio and Pumilio-enhanced microRNA -mediated repression.¹¹⁸ Pum1 is phosphorylated in cycling cells, enabling its ability to bind RNA while the RNA binding and overall stability of Pum1 is reduced in arrested cells. The inability of Pum1 to bind in arrested cells permits the RNA stem loop structure of p27 mRNA 3′-UTR to block microRNA access and thereby cause relief of repression.⁵⁷ Therefore, by activating the RNA binding abilities of repression-promoting regulators like Pum1 specifically in the proliferative state, specific tumor suppressors, like p27, can be suppressed by microRNA-mediated repression, perpetuating the proliferative state. Concurrently, by precluding the mRNA binding sites for Pum1, oncogene mRNAs, like E2F3 mRNA, are relieved of Pumilio-enhanced microRNA-mediated repression and are expressed, leading to a coordinated promotion of proliferation.

Direct posttranscriptional upregulation

Activation has been observed in a growing array of studies with specific mRNAs and microRNAs in distinct cellular conditions like G0, suggesting that activation by microRNAs is a regulated pathway of gene expression by microRNAs. Specific genes expressed in G0, are upregulated while others, particularly associated with the cell cycle, are downregulated.^{48-50,89,119,120} The alteration in microRNA profile and specific gene expression observed in G0 possibly creates a unique advantage for cancer cells to maintain their state and escape therapeutic targeting of dividing tumor cells.

A modified microRNP comprising AGO2 and a specific isoform of Fragile-X-Mental Retardation Related protein 1 (FXR1), called FXR1-iso-a, along with certain microRNAs in G0 like miR369–3p, associates with and mediates translation activation of specific mRNAs, such as TNFα, in a few quiescent cell lines.^46-49 FXR1 isoforms can bind the TNFα ARE and the FXR1 knockout mouse demonstrated cytokine deregulation, muscle wasting and neonatal death, suggestive of post-transcriptional deregulation of multiple cytokine targets including TNFα by FXR1 isoforms.¹²¹ Importantly, overexpression of FXR1-iso-a, leads to translation activation in low density proliferating cells, suggesting that modification of the microRNP via association of AGO2 with FXR1-iso-a leads to mRNA recruitment to polysomes in G0.⁴⁶ The upregulation of certain cytokines, signaling and growth regulatory factors, obtained by inducing G0 in cell lines is relevant to cytokine expression in quiescent monocytes⁵⁹ and dendritic cells⁹⁰ as well as in tumor associated macrophages. Monocytes extravasating into tissues, prior to differentiation into macrophages, undergo a G0-like arrest where TNFα and other cytokines are increased post-transcriptionally via their AREs prior to differentiation into macrophages; these effects can be reproduced by inducing G0 in monocytes in vitro in cell cultures.⁵⁹ Tumor associated macrophages produce cytokines and factors like TNFα that assist circulating some tumor cells, promote metastasis and induce quiescence in tumor cells, enabling their chemoresistance.^36-41

The mTOR pathway responds to nutrient and growth signals to regulate catabolic and anabolic processes, in particular, by regulating general and specific mRNA translation. Signaling control over mTOR is often deregulated in cancers, resulting in hyperactive mTOR signaling. The mTOR pathway regulates translation of 5′-terminal oligopyrimidine tract (TOP) mRNAs that usually encode ribosomal protein and other protein synthesis related factor mRNAs bearing the TOP sequence; translation is repressed upon cell cycle arrest at various points and upon nutritional deprivation.¹²² MiR-10a binds such target ribosomal protein mRNAs immediately downstream of the TOP sequence at non-canonical target sites, alleviates TOP-mediated repression and stimulates translation in response to treatments that activate TOP mRNA translation, such as anisomycin stress inducement or overexpression of a mutant activated RAS observed in cancers. The stimulated translation is inhibited by rapamycin (inhibitor of mTOR), suggesting involvement of the mTOR pathway.¹²³

In quiescent, confluent RK3E cells, miR206 levels increased and caused upregulated translation of KLF4 mRNA while in cycling cells, miR344 levels increased and caused repression of KLF4 mRNA and 3′UTR reporters¹¹⁹ at overlapping microRNA target sites. MiR-206 can repress KLF4 mRNA expression but only in proliferating breast cancer cells and not in normal immortalized MCF10A cells or confluent RK3E cells, suggesting that inducing miR-206 in quiescent cells leads to specific translation upregulation of KLF4 in response to the non-proliferative, quiescent state. Since KLF4 induces its positive regulator miR-206 as does G0, the positive feedback regulation may enable distinct effects of KLF4 in quiescent compared to proliferating cells: prolonging the doubling time in mesenchymal cells, promoting epithelial differentiation in normal cells and a malignant, mesenchymal-like loss of epithelial phenotype in tumor cells. These studies demonstrate that under quiescent conditions, specific G0-modified mRNAs with G0-induced microRNAs/microRNPs are capable of functioning as translation upregulatory complexes.

Quiescence-induced alterations in microRNPs and translation mechanism

In quiescent cells, the essential AGO2 associated repressor, GW182,^{97,103,105,107} was demonstrated to be altered in its interaction with AGO^112-115 and is marked by a decrease in GW bodies^112-114 although the bodies are not essential for repression.¹¹⁶ Abrogating GW182 or its interaction with AGO2 leads to a loss or relief of repression¹²⁴ or to activation as in the case of dAGO2 (a Drosophila AGO ortholog that does not interact with GW182) specifically with unadenylated reporters,¹²⁵ suggesting that the interaction of AGO2 with GW182 is altered on select transcripts in G0 to mediate translation activation.^46,49

A second characteristic, at least on some target mRNAs, is the recruitment of an altered translation activating microRNP complex, AGO2-FXR1-iso-a, which leads to relocalization of the mRNA to polysomes^46,48 (Fig. 1C activation). FXR1 has been demonstrated to be associated with ribosomes¹²⁶ and FXR1-iso-a in particular, on overexpression enhances translation.⁴⁶

Third, in several examples of translation activation, the activation complex fails to interact with GW182^46,115,125 and the target mRNAs lack a poly(A) tail¹²⁵ or have short poly(A) tails due to general poly(A) shortening,¹²⁷ thereby, potentially altering PABP-mediated interactions in such situations. Deadenylation activity by Poly(A) Ribonuclease (PARN) is enhanced in quiescent mammalian cells,¹²⁷ suggesting a distinctive mechanism of translation activation by microRNAs of poly(A) tail shortened mRNAs, as indicated by some of the studies on microRNA-mediated translation upregulation.^89,125

Fourthly, upregulation has been observed with transcripts and cellular conditions that involve specialized alternative translation mechanism features: IRES and a lack of a cap and poly(A) tail with dAGO2 in Drosophila extracts and with HCV, a 5′-TOP sequence and the mTOR pathway, short poly (A) mRNAs and conditions, such as oocytes and G0 cells, where general translation is reconfigured to enable specific translation instead.⁸⁹ AGO and AGO-related PIWI proteins like MILI have been demonstrated to associate with general translation initiation factors,^128,129 directly or via other RNA binding proteins, enabling recruitment of the ribosome^130,131 and consequent translation activation. These features suggest that activation involves alternative translation mechanisms for specific gene expression in distinct conditions.

Finally, G0 conditions also display downregulation of expression^50,89,120 especially of cell cycle factors and other such growth arrest-refractory genes^26,54 that must be depleted in addition to expression of G0 maintenance factors to perpetuate the quiescence state. Activation in G0 occurs with specific G0-expressed microRNAs and G0-expressed or -mobilized mRNAs that would be biologically required for G0-related functions. Activation was also observed by expressing tethered AGO2 or FXR1-iso-a or select, synthetic microRNAs with their corresponding reporters in G0 cells.^46,48 Since free, unbound AGO2 is limiting in the mammalian cell,¹³² these data suggest that specific G0-induced or mobilized mRNAs/ microRNAs would have the opportunity to be bound by new AGO complexes associated with different cofactors due to altered GW182 interaction in G0, alleviating repressive effects in G0 and mediating upregulation instead.^{46,48,112,114} These examples indicate that specific mRNAs and microRNAs induced or mobilized in specific G0 cellular conditions, such as miR-369–3p^46,48 and miR-206¹¹⁹ expressed in serum-starved or confluent G0 cells, may be recruited into distinct microRNPs that promote activation of specific G0-expressed mRNAs, required for the G0 state, while repression ensues with G0-induced microRNAs on other pre-existing mRNAs such as those encoding proliferation related genes that need to be suppressed to maintain the G0 state.^50,89,120

MicroRNA-mediated decoy of repressive proteins

A new functional role for microRNAs was recently demonstrated as decoy RNAs that remove repressive proteins, preventing them from accessing their target mRNAs, which are then expressed (Fig. 1D, microRNA binding RNA-BP). MiR-328 expression rescues differentiation and impairs survival of leukemic blasts in blast crisis chronic myelogenous leukemia (CML-BC), where the microRNA levels are usually decreased. MiR-328 exerts dual mechanisms to lead to a common biological consequence of promoting differentiation and inhibiting tumor progression: the microRNA binds and removes a repressive protein, hnRNP E2, from its target mRNA, C/EBPα, whose expression is then upregulated in a seed sequence- and microRNA-target binding independent manner,¹³³ promoting differentiation; however, miR-328 also functions via the normal seed sequence based target mRNA binding, repressing the mRNA encoding the survival factor for leukemic blasts, PIM1, to promote differentiation. Thus, miR-328 employs a novel mechanism in addition to its normal microRNA function and contributes to differentiation, via both positive and negative regulation of gene expression.¹³³ Therefore, microRNAs control pathways and biological outcomes in cancer not only by targeting multiple target genes but also through distinct mechanisms operating on specific transcripts and factors to lead to a common biological purpose. Whether other microRNAs also exert such concurrent mechanisms to elicit a shared outcome in addition to their defined sequence-specific, base-pairing mediated roles remains to be investigated.

Indirect gene expression by relief of repression

Apart from direct upregulation of select transcripts by specific microRNAs in G0 and other distinct conditions, indirect relief of repression is an alternate mechanism that also mediates gene expression where downregulation by microRNAs is abrogated. Relief of repression is regulated by cell cycle state and stress conditions associated with cancers.^96,97,134 Many RNAs and RNA-binding proteins are expressed specifically in cancer cell states, subtypes and conditions which, along with microRNP modifications induced by cell cycle state and stress conditions, alter microRNA/microRNP expression and functions and alleviate microRNA-mediated repression thereby, manipulating cancer progression.

Pseudogene transcripts and non-coding RNAs that bear target site sequences of specific oncogene and tumor suppressor mRNAs, function as microRNA regulators by competing with these mRNAs for microRNA binding and sequestration, resulting in alleviation of repression (Fig. 1D, microRNAs sequestered by non-coding RNA) and increased expression of critical oncogenes or tumor suppressors.^135-137 RNA-binding proteins like HuR either alter mRNA conformation or overall mRNP formation to control access and thereby, recognition of the target sites by microRNAs (Fig. 1E).^57,138,139 These functions of RNA binding proteins are regulated, microRNA/mRNA specific and can also promote microRNA functions¹⁴⁰ in a regulated manner in response to cancer cell state and stress conditions.¹³⁸ Alternatively, microRNAs compete directly with RNA binding proteins for the target sites that overlap with RNA binding protein target sequences or structures leading to two outcomes: stabilization and increased expression by increased mRNA levels due to microRNAs preventing access to decay factors or downregulation of expression by microRNAs along with their preclusion of stabilizing factors.^89,141

Stress response in tumors, in particular in response to harsh chemotherapy, can trigger either apoptosis or survival depending on the modifications on the microRNP. In response to stress, AGO and target mRNAs are relocalized from P bodies¹³⁹ to a lesser extent to stress granules¹⁴² and primarily to the cytoplasm where the mRNAs are expressed due to suppression of microRNA-mediated repression and cleavage of target mRNAs by poly-ADP ribose polymerase mediated modifications of AGO,¹³⁴ an important, therapeutic target in breast, ovarian, uterine and hematopoietic cancers (Fig. 1F). 3′-end modification (adenylation or uridylation) is extensively prevalent among microRNAs of several species.^143-149 Interestingly, while in cancer cell lines, uridylation has been associated with degradation intermediates, in non-proliferating tissues, uridylation is observed yet the microRNA levels are abundant,¹⁴⁶ suggesting that uridylation may have other undiscovered functions (Fig. 1F), that are regulated by non-proliferative conditions. A recent study demonstrated that depletion of the uridylase, TUT4/Zcchc11, had tumor-inhibitory effects in human HER2 positive breast tumors but not in triple negative breast tumors, correlating with the expression of specific microRNA processing regulator proteins.⁸¹ These results highlight the importance of understanding microRNA-regulatory mechanisms observed in distinct cancer states and subtypes for developing precise clinical targets and improved therapeutic options.

Conclusions

MicroRNAs demonstrate distinct mechanisms of post-transcriptional regulation of their levels and functions in control over target gene expression in different cancer cell states within tumor populations and in response to the signals in the tumor environment. These regulated, microRNA mechanisms are produced as a result of specific manipulations by RNAs and RNA-binding proteins or by modifications to the mRNA or microRNP; the regulators and modulatory events are themselves controlled by specific cancer state conditions. These mechanistic dynamics provide opportunities for greater specificity and versatile gene expression that enable tumor persistence and progression. An in-depth understanding of microRNA-mediated mechanisms and their regulation in distinct cancer cell states and in response to the tumor environment will provide highly specific markers, clinical targets and advance therapeutic options.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors and this review were supported by a Cancer Research Investigator Award, the D. and M-E Ryder Award and a Smith Family Foundation Award to S.V.