MicroRNAs and miRceptors: a new mechanism of action for intercellular communication

Abstract

MicroRNAs (miRs) are small non-coding RNAs (ncRNAs) that control the expression of target genes by modulating (usually inhibiting) their translation into proteins. This ‘traditional’ mechanism of action of miRs has been recently challenged by new discoveries pointing towards a role of miRs as ‘hormones’, capable of binding to proteic receptors (miRceptors) and triggering their downstream signalling pathways. These findings harbour particular significance within the tumour microenvironment (TME), defined as the variety of non-cancerous cells surrounding cancer cells, but are relevant also for other diseases. In recent years it has become clearer that the TME does not passively assist the growth of cancer cells but contributes to its biology. Some of the mediators of the intercellular communication between cancer cells and TME are miRs shuttled within exosomes, a subtype of cellular released extracellular vesicles. This article will highlight the most recent findings on the biological implications of miR–miRceptor interactions for the biology of the TME and other diseases, and will provide some perspectives on the future development of this fascinating research.

This article is part of the discussion meeting issue ‘Extracellular vesicles and the tumour microenvironment’.

1. Introduction

RNA is conventionally considered an intermediate molecule, between DNA and proteins, in the cascade of events leading to gene expression. The relatively recent revolution of non-coding RNAs (ncRNAs) has highlighted a pivotal role for RNA also as a macromolecule finely regulating the expression of genes. The most studied members of the ncRNA family are microRNAs (miRs), short ncRNAs involved in the regulation of gene expression [1,2]. The best characterized mechanism of action of miRs consists in their sharing of the intracellular siRNA machinery (and especially of Dicer and Argonaute 2 proteins within the RNA-induced silencing complex) to prevent the translation of target messenger RNAs (mRNAs) into proteins. In humans, such inhibition mainly occurs by translational repression after partial matching of the sequence of the miR with the sequence of the target mRNA mostly in the 3′-untranslated region (3′-UTR) of the mRNA. It is also possible that the miR–mRNA interaction occurs within the coding region or at the 5′-UTR of the mRNA [3–6]. Intriguingly, it has also been shown that miR–mRNA interaction (mostly at the 5′-UTR) can increase the translation of the mRNA [7]. Regardless of its site and outcome, the described interaction defines miRs as molecules acting through an RNA–RNA base pairing mechanism. MiRs have profound effects in key mechanisms of cancer biology such as angiogenesis [8], metastasis [9], stress response [10] and drug resistance [11]. For example, Otsuka et al. showed that the let-7 family of miRs silences the expression of TIMP-1 in endothelial cells, promoting their proliferation and motility [12]. MiR-135a suppresses invasiveness and metastatic potential of prostate cancer cells by targeting ROCK1 and ROCK2 [13]. MiR-21-5p and miR-34a reduce their expression in tissues exposed to radiation-induced stress [10], whereas miR-21-3p has been shown to increase ovarian cancer resistance to cisplatin by silencing NAV3 [14]. While initially studied for their effects in gene regulation inside of cells, miRs have been recently identified as key players in cell-to-cell communication. In a seminal paper, Valadi et al. showed that miRs can be functionally transferred from one mast cell to another, shuttled within extracellular vesicles (EVs) [15]. This discovery provided the rationale for further investigating the possible role of miRs within the tumour microenvironment.

Tumour microenvironment (TME) is defined as the variety of non-cancerous cells surrounding cancer cells. This includes fibroblasts, immune cells (lymphocytes, granulocytes, monocyte/macrophages, dendritic cells), endothelial cells, and pericytes that surround cancer cells, regardless of the specific type of cancer. Historically, cancer has been considered a disease centred on the genetic and functional characteristics of the cancer cell itself. However, an increasing literature has clearly shown that the surrounding cells of the TME do not act as passive spectators but are actively involved in shaping and affecting the mechanisms leading to cancer growth, proliferation, metastatic spreading and development of resistance to therapy. At the same time, cancer cells are able to influence the biology of the cells of the TME or, using an expression that has been widely adopted, cancer cells ‘educate’ the surrounding TME. In order for such a reciprocal cancer–TME communication to occur, an exchange of functional molecules between the involved cell types is necessary. Soluble factors, such as cytokines and EVs are the main mediators of this exchange. Intriguingly, for a long time EVs have been considered the ‘garbage bins’ of the cells, containers of products of cellular catabolism or toxic by-products potentially dangerous that the cell needed to eliminate. Platelet-derived EVs were first isolated in the blood of patients with activated coagulation and fibrinolysis [16], whereas a few years later it was demonstrated that EVs derived from human neutrophils contain proteins selected with a specific sorting mechanism and can functionally bind to opsonized bacteria [17]. Once the functional role of EVs had been established, it has become clearer that there is not only one type of EV. Several types of classifications have been proposed, essentially based on their size, biogenesis or cell of origin, creating a jungle of confusing terms to indicate subtypes of EVs. Currently, it is generally accepted that EVs can be classified in the following four major categories: exosomes, microvesicles, apoptotic bodies and large oncosomes. Exosomes are the smallest kind of EV, ranging between 30 and 120 nm in size, and expressing transmembrane or lipid-bound extracellular proteins (such as tetraspanins CD9, CD63, CD81), cytosolic proteins (such as TSG101 or syntenin) and extracellular proteins (such as fibronectin 1 and acetylcholinesterase). Exosomes are not expected to express proteins from organelles that do not contribute to their formation such as cytochrome c, calnexin or histones. These criteria for the characterization of exosomes have been recently published by the International Society of Extracellular Vesicles (ISEV) as minimal experimental requirements for the definition of exosomes [18]. Microvesicles are intermediate-size EVs, ranging between 50 and 1000 nm and express integrins, selectins and CD40 ligand as specific markers [19]. Apoptotic bodies are between 500 and 2000 nm in size, originate from the external blebbing of the membrane of apoptotic cells and express abundant quantities of phosphatidylserine as marker [19]. Large oncosomes are the biggest category of EVs, whose size ranges from 1000 to 10 000 nm and have been described in prostate cancer cells stimulated by epidermal growth factor (EGF) through a non-apoptotic blebbing known as the ‘amoeboid’ phenotype, highly associated with malignancy [20,21]. One of the most characterizing properties of exosomes, compared to the other types of EVs, is their biogenesis. Exosomes are generated through an active process of the cell that begins from the invagination of the membrane of a large intracellular vesicle. This process mostly involves proteins of the endosomal sorting complex required for transport (ESCRT) and leads to the formation of a multivesicular body (MVB), which undergoes a maturation process ultimately leading to a mature MVB containing several vesicles. Several ESCRT-independent pathways of MVB biogenesis have been reported; however, exosome-genesis is completed by the fusion of the MVB with the plasma membrane of the cell (mediated by RAB and SNARE proteins) and the consequent release of exosomes [22]. Essentially, the functional properties of EVs and exosomes are determined by their cargo molecules. Exosomes contain proteins, RNA (both messenger and ncRNAs), and DNA [22,23]. Among the RNA cargo, miRs have emerged as essential players in the function of exosomes within the TME, and previously unsuspected functions for exosomal miRs have been recently identified and shown to be relevant in cancer and other human diseases.

2. The discovery of miRceptors and implications in human diseases

In 2008, Mitchell et al. elegantly showed that miRs can be detected in human plasma and are remarkably stable from RNase degradation [24]. Moreover, they identified high levels of miR-141 in the serum as a valid cancer biomarker able to distinguish prostate cancer patients from healthy controls [24]. This work has been followed by a plethora of studies confirming that miRs can be detected in blood (as well as other biological fluids such as urines, pleural effusions, seminal liquid, sweat, tears) and specific miRs are differentially expressed in specific cohorts of cancer patients compared to the healthy counterpart [25]. Intriguingly, circulating miRs can also be exploited as biomarkers for human diseases other than cancer, such as childhood tuberculosis infection [26], allergic rhinitis and asthma [27], inflammatory diseases [28], cardiovascular diseases [29] and metabolic diseases [30], just to name a few. The identification of miRs in patients' blood together with the discovery that they can be shuttled intercellularly within EVs, sparkled the idea that miRs might mediate paracrine and endocrine interactions. Specifically, a molecule able to be secreted in the general circulation and targeting neighbouring or distant cells fulfils the definition of ‘hormone’. Traditionally, a ‘hormone’ is defined as a peptide or as a steroid. Therefore, assuming that an RNA might function as a hormone represents a challenge to the current knowledge. However, if we assume that miRs are hormones, a hormone-like mechanism of action should be hypothesized for miRs as well, meaning that a proteic receptor for miRs (also defined as a miRceptor) should exist. MiRceptors imply the direct interaction between a miR and a protein. In 2010, Eiring et al. provided the first evidence of a direct miR–protein binding by showing in chronic myelogenous leukaemia that miR-328 can sterically bind to hnRNP E2, an inhibitor of the transcription factor CEBP-α [31]. In leukaemic clones, the downregulation of miR-328 releases hnRNP E2 ability to inhibit CEBP-α, leading to a block in granulocytic differentiation, which triggers the blastic crisis. A summary of the studies in which the miR–miRceptor interactions has been assessed is summarized in figure 1. In 2012 we were the first to demonstrate that non-small cell lung cancer (NSCLC) cell lines secrete miR-21 and miR-29a within EVs, and that these vesicular miRs are taken up by tumour-associated macrophages (TAMs) mostly located at the level of the cancer invasiveness front [32]. We showed that NSCLC-derived vesicular miR-21 and miR-29a enter the endosomal compartment of TAMs and bind to human Toll-like receptor 8 (TLR8) or its murine orthologue Toll-like receptor 7. The binding of miR-21/29a to TLR8 triggers the downstream MyD88-mediated NF-κB activation, resulting in increased secretion of interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α) by TAMs [32]. As a result a pro-tumoral inflammatory response within the TME is initiated and promotes lung cancer growth and metastatic dissemination. In 2014, He et al. showed that the miR–TLR interaction is important in the genesis of cachexia, a debilitating syndrome characterized by predominant skeletal muscle mass loss and usually associated with the terminal stages of the disease [33]. In their paper, the authors show that lung and pancreatic cancer cell lines release vesicular miR-21 that binds to TLR7 on murine myoblasts, promoting muscle cell apoptosis requiring the activity of the c-Jun kinase [33]. Since high levels of IL-6 in the TME have been associated with increased drug resistance and the miR–TLR8 interaction is conducive of IL-6 release by TAMs, the role of the miR–miRceptor interaction in drug resistance has been investigated. In 2015, our group showed that miR-21 (but not miR-29a) is released by neuroblastoma cells within exosomes [34]. The vesicular miR-21 is then taken up by cancer surrounding TAMs, where it binds to TLR8 and triggers NF-κB-mediated upregulation of miR-155 in TAMs and in their exosomes. MiR-155 (normally not expressed by neuroblastoma cells) is then shuttled from TAMs into neuroblastoma cells via exosomes, and upon occurrence of the miR-21 binding to TLR8 in TAMs. When it reaches cancer cells, miR-155 directly targets TERF1, a member of the shelterin complex and inhibitor of telomerase [34]. The downregulation of TERF1 promotes telomerase activity in neuroblastoma cells and increases resistance to cisplatin both in vitro and in vivo [34]. Interestingly, this interaction has been observed also in other types of malignancies, highlighting the broader implications of the proposed mechanism [34]. For instance, Patel and Gooderham showed in a colorectal cancer model that the IL-6 secreted by THP1 human monocytes triggers the IL-6R/STAT3 pathway in the cancer cells, leading to increased production of miR-21 and miR-29b in the colorectal cancer cells. These cells then, release miR-21 and miR-29b in exosomes and shuttle them back to the immune cells, where they bind to the TLR8 miRceptor, further increasing the production of IL-6 by the immune cells [35]. This study identifies a very interesting feed-forward loop involving miR–miRceptor interactions which is responsible for the increased secretion of IL-6 within the colorectal cancer microenvironment.

Figure 1.

Involvement of miR–miRceptor interactions in human diseases. Synopsis of the main studies in which the interaction of miRceptors (human TLR8 or murine TLR7) with the indicated miRs has been investigated. The numbers in brackets indicate the reference studies that showed that mechanism. TME, tumour microenvironment; aGVHD, acute graft versus host disease.

The interaction miR–miRceptor harbours implications that are relevant even beyond cancer. In 2014, Tudor et al. assessed the importance of this mechanism in sepsis, a condition that affects over 10 000 people/day, with a mortality rate of up to 50% in the United States [36]. This life-threatening condition is extremely time-sensitive, since it has been calculated that the survival rate of a patient diagnosed with septic shock drops by 7.6% every hour a proper antibiotic therapy is delayed, and the management of this condition is challenged by the absence of suitable sepsis biomarkers [37]. In this study, the authors observed that patients undergoing post-surgery sepsis expressed high levels of miR-K-10b and miR-K12-12*, two miRs produced by Kaposi sarcoma herpes virus (KSHV), compared to patients who underwent surgery but never developed the septic condition. Moreover, they showed that miR-K-10b and miR-K12-12* trigger the septic condition by binding to TLR8 in histiocytes and inducing the secretion of IL-6 and other cytokines that mediate the shock response [38]. Interestingly, activation of TLR7 and TLR8 has also been observed in response to the innate immune recognition of HIV-1 infection [39], leading to the hypothesis that HIV-derived miRs may also be able to bind to the miRceptors, therefore affecting the immune response to the HIV-1 infection [40].

Acute graft versus host disease (aGVHD) is a frequently lethal complication of allogeneic haematopoietic stem cell transplantation (HSCT) whose triggering mechanism is poorly understood. In 2013, Garzon's laboratory deep sequenced paired samples from 10 patients who developed aGVHD after HSCT and controls who did not develop aGVHD. They identified a signature of differentially expressed miRs, with miR-146 and miR-29a among the most upregulated in the blood of patients who developed aGVHD. The authors further showed that miR-29a can bind to TLR8 in antigen presenting cells, leading to upregulation of the immune cell marker of activation CD69 and CD86. Moreover, they showed that by binding to TLR8 in dendritic cells, miR-29a activates the NK-κB pathway in dendritic cells and increases their secretion of TNF-α [41]. Overall, this study highlights a role of the miR–miRceptor interaction in the pathogenesis of aGVHD and suggests that specific circulating miRs can be exploited as biomarkers of risk to develop this condition.

The role of the miR–miRceptor interaction in neurodegeneration was first demonstrated by Lehmann et al. in 2012 [42]. They discovered that miR let-7b activates TLR7 in murine microglia and macrophages leading to increased secretion of TNF-α and promotes neuron death in a TLR7-dependent manner. This study also identified higher levels of let-7b in the cerebrospinal fluid (CSF) of patients diagnosed with Alzheimer's disease, and showed that injection of let-7b in the CSF of mice induced neurodegeneration [42]. More recently, Yelamanchili et al. isolated EVs from the brains of macaques infected by simian immunodeficiency virus (SIV) or uninfected controls and showed increased levels of miR-21 in the EVs of infected brains (both in neurons and in macrophage/microglial cells) [43]. They proceeded by showing that vesicular miR-21 is able to induce neurotoxicity in a TLR7-dependent manner. They also showed that the miR-21-TLR7 pathway activation leads to a caspase-independent cell death that occurs through necroptosis, and can be reversed by the necroptosis-inhibitor necrostatin-1 [43]. Overall, these studies underline the importance of miR–miRceptor interactions in the pathogenesis of neurodegenerative disorders and highlight this mechanism as a suitable therapeutic target for these diseases.

3. Future implications for the field

The discovery that miRs can function as ‘hormones’, and can bind to and activate a receptor has changed our understanding of how miRs operate. Consequently, an increasing interest on the implications of miR–miRceptor interactions has pervaded the scientific community in an attempt to clarify the consequences of such a molecular mechanism. Clearly, we are at the infancy of a new field that leaves several unanswered questions. The first obvious question is whether, in addition to human TLR8 and murine TLR7, there are other miRceptors out there. This is a quite interesting question, since it could involve considerations on the many currently existing ‘orphan’ receptors (meaning receptors for which a ligand has not been identified yet). Perhaps the classification of them as ‘orphan’ is a bias due to the fact that we do not consider the possibility that their ligand might be an RNA, instead of a protein. Another point of discussion is about the ‘geographic’ distribution of the miRceptors within the TME: which cells express the miRceptor and which one(s) do not? In our experimental approaches there is a need to simplify the complexity of the TME in order to reach rigorous scientific deductions. However, focusing on a two-cell interaction model has the disadvantage of losing track of the ‘big picture’, with the risk of ignoring perhaps a different cell population within the TME that mediates a much stronger miRceptor-mediated response to the miR-binding. The next set of questions regards the binding of the miR to the miRceptor: what determines the binding of specific miRs but not of others? In our publication in 2012, we observed that lung cancer cells also release vesicular miR-16 and that miR-16 enters the endosomal compartment in the TAMs but does not bind to TLR8 [32]. Which type of information differentiates a TLR8-binding miR (such as miR-21 and miR-29a) from a non TLR8-binding miR (such as miR-16)? Preliminary data suggest that part of this difference resides within the sequence of the mature miR itself [32], but further work is warranted to clarify this critical point. On the same topic, it is still unclear whether the miR binds directly to TLR8 or whether an intermediate protein is necessary for this interaction to occur. Finally, the stoichiometry of the miR–miRceptor interaction is completely unknown and has not been addressed yet. Another important consideration is whether only vesicular miRs are able to interact with miRceptors or even extra-vesicular miRs (bona fide carried by Ago2) can activate a miRceptor-mediated signalling. And, of course, there is the possibility that other ncRNAs (in addition to miRs) can bind to a proteic receptor and activate its downstream signalling pathways. It is noteworthy that the ability of a miR to interact with a miRceptor does not preclude its ability to work according to the ‘traditional’ mechanism of action of miRs. For instance, vesicular miR-21 not only is directed in the endosome of the recipient cell where it binds to TLR8 and triggers NF-κB activation, but likely is also released in the cytoplasm of the cell where it mediates the transcriptional repression of target mRNAs. Therefore, it would be important to be able to determine how much of the vesicular miR follows the path of the ‘traditional’ mechanism and how much is working through the TLR8 binding. While all of these questions are still unanswered, a solid point is that circulating miRs can be successfully used as biomarkers of human diseases and likely the miR–miRceptor interaction can generate by-products that could also be detected as biomarkers of the interaction and, indirectly, of an existing condition. But the most promising future regards the therapeutic implications of the miR–miRceptor interaction. We can devise a series of strategies to interfere with the miR–miRceptor interaction in order to impair a pathogenetic intercellular communication. These strategies include the targeting of the miRceptor downstream signalling, as well as interfering with the secretion of the miR and/or the expression of the miRceptor. A better understanding of the stoichiometry and of the complex of molecules involved in the interaction will in the near future allow us to antagonize the binding and prevent the pathogenetic response from occurring. In conclusion, the discovery of a receptor-mediated mechanism of action for miRs has identified new possible molecular targets for the treatment of cancer and other human diseases, has identified a new set of possible biomarkers and has fundamentally challenged our definition of ‘hormones’ by including nucleic acids among the type of macromolecules capable of mediating autocrine, paracrine and endocrine interactions.

Data accessibility

This article has no additional data.

Competing interests

I declare I have no competing interests.

Funding

M.F. is supported by the award no. R01 CA215753 from the National Cancer Institute. M.F. is a St Baldrick Foundation's Scholar and is supported by a Hyundai Hope on Wheels grant, a William Lawrence & Blanche Hughes Foundation grant, a Jean Perkins Foundation grant, the Nautica Malibu Triathlon Funds, a STOP Cancer grant, the Hugh and Audy Lou Colvin Foundation grant, and the award number P30CA014089 from the National Cancer Institute.