MicroRNA implication in therapeutic resistance and metastatic dissemination of bone-associated tumors

Introduction

Breast cancer and prostate cancer are the most frequently diagnosed cancers of women and men, respectively, living in developed countries with more than 1 million new cases per year worldwide.

Despite the progresses made in the development of new treatments, mortality of cancer patients is still highly linked to the occurrence of distant metastasis. More than 70% of breast and prostate cancer patients present bone metastases in their advanced diseases. Lung, colon, bladder, or kidney cancers, if less frequently, are also found to metastasize to bones.

In addition, around 2000 new cases of primary bone cancers are being diagnosed within Europe every year (osteosarcoma being the most frequent bone cancer types), lung and bones been the preferential metastasis location of these primary bone cancers.

All these data show that the skeleton is one of the most common organ affected by metastasis. The high density of fenestrated capillaries present in the bone marrow (BM) as well as the high concentration of growth factors released as a consequence of the constant remodeling undergone make bones to be an easy-to-reach location that provides a fertile soil for tumor cells.

MicroRNAs implication in metastatic dissemination of bone-associated tumors

Two processes occur during the bone remodeling phenomenon: bone formation (by osteoblasts (OBs)) and bone resorption (by osteoclasts (OCs)). New bone formation occurs at the site of previous bone resorption [ ]. In normal conditions, those events are constantly and carefully balanced to maintain healthy bone structure. Any disturbance in this equilibrium leads to pathological bone structure development, associated with various collateral troubles [ , ].

A primary tumor is composed by heterogeneous tumor cell populations with different abilities and that proliferate at primary site. When the primary tumor reaches a significant size, tumor cells start to secrete angiogenic factors in order to establish a capillary network that ensures oxygen and nutrients intake that are both necessary for the primary tumor survival and development. Meanwhile, thanks to the heterogeneity of the primary tumor cell populations, the most unstable cells [ ] (also known as cancer stem cells (CSCs) [ ]) will acquire the capability to undergo epithelial–mesenchymal transition (EMT), process that allows them to lose cell–cell adhesion, to degrade basal membrane and to invade the extracellular matrix [ ]. This phenomenon is reversible. The detached tumor cells then enter lymphatic and blood circulation, through new-formed capillaries [ ]. A high number of circulating tumor cells (CTCs) will be eliminated during their transportation through the circulation [ ]. Among the surviving ones, only a few will become finally trapped in distant capillaries and will undergo mesenchymal–epithelial transition (MET) toward the metastatic site(s) [ ]. Those metastatic tumor cells can enter in a dormancy state before receiving the appropriate signals, allowing them to proliferate and to develop as actual metastasis [ , ].

Two types of metastases can be formed at bony level. They both act disturbing the normal osteoblast–osteoclast activity balance that prevails in healthy bones, but by different mechanisms, leading to two pathological situations: (1) osteoblastic metastases stimulate the osteoblastic population activity compared to the osteoclastic one, leading to pathologic bone condensation appearance and (2) osteolytic metastases where the OCs' activity is predominant and elevated bone destruction is the pathological outcome [ , ]. Usually, breast and prostate tumors are known to develop bone metastases that have osteoclastic and osteoblastic phenotypes, respectively. Even though this is the most accepted classification, in most cases patients present mixed phenotypes in their bone metastases. In fact, histological and biochemical studies revealed the presence of elevated bone resorption within osteoblastic metastases, meaning that osteolysis occurs in all bone metastases, regardless of their radiographic appearance [ ].

Osteolytic metastases are caused by the establishment of a ‘vicious cycle’ between metastatic breast tumor cells and bone cells. Metastatic breast tumor cells have the ability to produce direct osteoclastic stimulating factors, such as M-CSF (macrophage colony-stimulating factor) [ ], IL-8 (Interleukin-8) [ ], IL-11 (Interleukin-11) [ ], and indirect ones, like PTHrP (parathyroid hormone–related protein) [ ]. This protein stimulates OBs' activity, leading to an osteoblastic overproduction of RANKL (receptor activator of nuclear factor kappa-B ligand), an OC's development activator, compared to OPG (osteoprotegerin), and an RANKL inhibitor. Thus, PTHrP indirectly increases OC's activity. All these factors lead to osteolytic lesions' emergence. Resorption of bone matrix releases some factors such as TGF-β (transforming growth factor-β) that enhances metastatic tumor cell proliferation which in turn will release more direct or indirect osteoclastic stimulating factors, causing even more bone resorption [ , ].

The mechanisms of osteoblastic metastasis establishment have not been fully elucidated. Some evidences indicate that metastatic prostate tumor cells could secrete factors, such as ET-1 (endothelin-1) [ ] and BMPs (bone morphogenetic proteins) [ ] which support OBs' activity stimulation, promoting pathological bone deposition. In turn, OBs produce factors that stimulate metastatic prostate tumor cell proliferation (TGF-β, IGF-1 (insulin-like growth factor-1), IL-6 (Interleukin-6), etc.). OBs also overproduce RANKL, increasing OCs' activity and allowing to release bone matrix–trapped factors which simulate metastatic prostate tumor cell proliferation in turn. This concurs to osteolytic lesions' emergence within so-called osteoblastic metastasis, explaining radiographic results from patients suffering from prostatic bone metastasis [ ].

Those skeletal lesions profoundly impair patient's quality of life by causing bone pain, fractures, spinal cord and nerve roots compressions, loss of mobility, and metabolic complications such as hyper/hypocalcemia [ ].

All current existing treatments are mainly based on the improvement of patients' quality of life through the prevention of skeletal complications. BPs (bisphosphonates) are widely used to prevent skeletal damages caused by bone metastasis. These molecules impair OC biology and activity. However, antitumoral effects of BPs and their efficacy are still controversial [ ]. In fact, some studies have shown that zoledronic acid, pamidronate, and other BPs treatments reduced skeletal complications and decreased bone pain in patients suffering from breast cancer with bone metastasis, whereas only transient pain relief was achieved in patients suffering from prostate cancer associated with bone metastasis [ ]. The combination of BPs with standard chemotherapeutic agents could induce a synergistic apoptotic effect on metastatic tumor cells, arising a new potential therapeutic frontline in order to find more effective treatments suitable for patients suffering from bone metastasis [ ]. Since 2010, a therapeutic antibody, called denosumab, is proposed to treat patients presenting bone metastasis. This antibody decelerates OCs' activity and thus bone degradation, by acting as an RANKL trap. However, side effects can be observed on denosumab-treated patients, like hypocalcemia [ ]. Bone defects/fractures can also be treated by surgical intervention. Concerning metastatic tumor cells themselves, surgical interventions are also an opportunity to extract them from bones. Chemotherapies and/or radiotherapies and/or hormonotherapies (if tumor cells are hormone dependent) can also be employed to eliminate bone metastatic tumor cells [ ].

All these therapeutic proposals do not act immediately on patients' pain, leading to painkiller prescription for the majority of patients suffering from bone metastasis.

For many years, new diagnostic tools needed to be developed for the detection of breast and prostate malignancies at their very early stages in order to both perform better predictions in the patients' outcomes and avoid further bone metastatic dissemination. Likewise, more efficient and adapted therapeutic strategies needed to be hunt down to improve survival of patients suffering from these cancers and affected by bone metastasis. To achieve these goals, we needed to reinforce our knowledge about the bone metastatic dissemination process. That is why, some recent research studies turned to miRNAs (microRNAs) study.

miRNAs are small (18–24 nucleotides) noncoding RNAs that regulate gene expression by targeting the 3′ untranslated region (UTR) of mRNAs, resulting in either mRNA cleavage or protein translation repression. miRNAs are transcribed in the nucleus as long hairpin chains (pri-miRNAs) that are cleaved by Drosha giving birth to precursors (pre-miRNAs) that will be exported toward the cytoplasm for further maturation. The enzyme Dicer will then excise these single hairpin chains into a duplex where one of the chains (the mature miRNA) will be incorporated into an effector complex called RISC (RNA-induced silencing complex), whereas the other will be eliminated [ ]. Since the discovery in the early 1990s of the first miRNA as a major player in Caenorhabditis elegans ' development [ ], hundreds of miRNAs have been identified in many other species, including Homo sapiens [ ], as important players in cell biology. They are not only implicated in natural cellular processes as proliferation, differentiation, development, or apoptosis, for example, they also have been found of critical importance in tumorigenesis [ , ]. During past decades, research studies revealed that miRNAs expression is often altered in cancers, leading to the “oncomiRNAs” term appearance. OncomiRNAs are classified into two groups: (1) oncogene miRNAs: usually overexpressed in cancer; they promote tumor cells development through downregulation of tumor suppressor genes expression and (2) oncosuppressor miRNAs: inversely; they function as oncogenes expression inhibitors. They are underexpressed in cancer, leading in fact to tumor cells development [ ]. OncomiRNAs expression profiles fluctuate during cancers evolution, knowing that variables like these cancers' origins, for example, also have an influence. Monitoring miRNAs expression profiles in all cancers' development steps allows to better understand metastatic dissemination process, especially bone metastasis development, because there was a lack in our knowledge in this phenomenon. This then allows to identify new diagnostic, prognostic, and therapeutic elements/targets to treat patients, including those suffering from bone metastasis. Furthermore, the identification of pathways that regulate those miRNAs expression as well as the downstream genes whose they modulate the expression, revealed supplementary attractive tracks.

Here we proposed to discuss about recently identified miRNAs implicated in bone metastatic dissemination.

In his “Seed and Soil” hypothesis, published in 1889, Sir Stephen Paget explained that some cells, in a primary tumor, can be compared to seeds which have an affinity for specific microenvironment(s)/soil(s). Metastasis from a primary tumor will develop only if some compatible tumor cells/seeds are “plant” in the right microenvironment(s)/soil(s) [ ].

Tumor cells' integration in host metastatic microenvironment(s) is much more favored if they gain mimicry: the ability to imitate host cells. This is accomplished by the expression and the production of markers and factors normally belonging to cells composing the target microenvironment(s) [ ]. miRNAs recently show their implication in this phenomenon, especially in breast and prostate tumor cells osteomimicry, favoring their metastatic development in bone tissues [ ].

miR-135 and miR-203 regulate runt-related transcription factor 2 (Runx2) expression, a transcription factor implicated in osteoblastic lineage precursors differentiation. These two miRNAs are highly expressed in normal breast epithelial cells, making Runx2 gene expression undetectable. On the contrary, Runx2 gene is expressed by metastatic breast tumor cells because of miR-153 and miR-203 expression absence. miR-135 and miR-203 expression impair tumor cells' viability, decreasing tumor growth, and impair their migration capacity, reducing bone metastasis development. These miRNAs expression level also decreases bone resorption induced at metastatic sites. Mechanistically, Runx2 expression by breast tumor cells stimulates their multiplication and allows PTK2 (protein tyrosine kinase 2) and ROCK1 (Rho-associated coiled-coil containing protein kinase 1) genes expression which are implicated in cell adhesion and cell motility, respectively. In addition, Runx2 promotes IL-11 and PTHrP production, stimulating directly or indirectly OC development and activity and thus bone degradation [ ].

Similarly, a recently published study shows that miR-466 is underexpressed by prostate tumor cells compared to normal prostatic cells. In vitro tests show that artificial miR-466 overexpression by prostate tumor cells impairs their proliferation and increases their apoptosis. It also leads to EMT process reversion and decreases their migration and invasion capacities.

In vivo tests highlight that miR-466 overexpression by prostate tumor cells reduces tumor growth and bone metastasis development. miR-466's target genes research leads to the identification of Runx2 gene as such. miR-466 underexpression in prostate tumor cells leads to Runx2 gene overexpression and its downstream targets, including ANGPT1 (angiopoietin), ANGPT4 , Osteopontin , Osteocalcin , and MMP11 (matrix metallopeptidase 11) genes expression rise. These genes are implicated in tumor growth and bone metastatic processes [ ].

miRNAs are now listed as implicated in osteomimicry phenomenon by their capacity to modulate bone-related factors expression by breast and prostate tumor cells and thus to orient their metastatic tropism through bone tissue.

Secondly, recent studies have shown that metastatic target microenvironment(s) may be prepared by primary tumor cells for subsequent metastatic deposits through the building of so-called premetastatic niche(s). This improves primary tumor cells homing and growth at the metastatic target site(s). Cells from the primary tumor play a key role in the development of these niches by producing factors (soluble or exosome/vesicle-transported) which modify host cells behaviors, rendering target microenvironment(s) amenable for metastatic development [ ]. miRNAs have been also recently identified as responsible for premetastatic niche establishment [ ].

Metastatic breast tumor cells exhibit high miR-218 blood secretion, through extracellular vesicles production. These vesicles target OBs, leading to subsequent miR-218 release in their cytoplasm. Then, this miR downregulates Col1a1 (collagen type I α1) gene expression, decreasing osteoblastic type I procollagen deposition at the bone matrix. Furthermore, OBs exhibit increased TIMP3 (tissue inhibitor of metalloproteinase 3) secretion, an inhibitor of the N-procollagenase ADAMTS2,3 (ADAM metallopeptidase with thrombospondin type 1 motif 2,3), impairing type I collagen processing. Complementary, miR-218 directly inhibits YY1 (Yin yang 1) gene expression, increasing inhibin βA gene expression and secretion by breast tumor cells. miR-218 also inhibits inhibin βB gene expression in breast tumor cells. Paracrinally, inhibin βA, after its association with inhibin α, inhibits phosphorylation of SMAD2/3 in OBs, increasing TIMP3 expression and secretion. This occurs to also impairs type I procollagen processing. Thus, miR-218 indirectly further promotes osteolysis, to make bone tissue adapted to metastatic breast tumor cells development [ ].

In vitro tests show that exosomes containing miR-141-3p, released by prostatic tumor cells, are able to be endocytose by OBs. miR-141-3p are then released into OB's cytoplasm and inhibit DLC1 (deleted in liver cancer 1) gene expression, leading to p38MAPKs (mitogen-activated protein kinases) signaling pathway activation. This promotes OBs' activity and subsequent OPG overproduction compares to RANKL secretion. OPG inhibits RANKL activity, compromising OC activation. In vivo experiences show that exosome-carried miR-141-3p injections in prostate tumor mice models lead to bone metastasis development that are associated with an osteoblastic phenotype. miR-141-3p promotes a satisfying bone microenvironment formation for metastatic prostatic tumor cells development [ ].

In addition to the miRNAs implication in the capacity of some tumor cells to mimic host cells behavior and their ability to modify their target microenvironment(s) proprieties, miRNAs are also implicated in tumor cells spread initiation toward host microenvironment(s) through newly formed capillary network [ ].

A study published this year explains that miR-181a-5p and miR-181b-5p are implicated in EMT and invasion of prostate tumor cells. MIIP (migration and invasion inhibitory protein), an oncosuppressor factor, underexpression in prostate tumor cells leads to miR-181a-5p and miR-181b-5p overexpression, directly impairing KLF17 (Kruppel-like factor 17) production which is normally a negative regulator of EMT and invasion thanks to its repression action on SNAIL1/2 and TWIST genes expression. Factors, such as E-cadherin, are thus underexpressed. Different analyses highlight that MIIP, KLF17, and E-cadherin proteins are underproduced whereas miR-181a-5p and miR-181b-5p are overexpressed by prostate tumor cells compared to normal prostatic cells, extract from patients' samples. MIIP overexpression during in vitro experiences decreases prostate tumor cells invasion. Moreover, in vivo MIIP underexpression promotes tumor growth when prostate tumor cells are grafted subcutaneously, and induces bone osteolysis when they are injected in tibia [ ].

Very recently, miR-199-5p shows its implication in breast cancer angiogenesis.

miR-199-1p expression by breast tumor cells, which impairs ALK1 (activin receptor–like kinase 1) gene expression in human umbilical vein endothelial cells (HUVECs), is decreased during the angiogenesis process. In vitro miR-199-1p overexpression inhibited capillary-like tubular structures formation and migration of HUVECs. Mechanistically, miR-199-5p underexpression by breast tumor cells increases ALK1 gene expression in HUVECs, leading to the accentuation of SMAD/Id1 (inhibitor of DNA binding 1) signaling pathway activation, and initiation of the angiogenesis process (concomitant with BMP9 (bone morphogenetic protein 9) stimulation). miR-199-5p overexpression during in vivo experiences allows to decrease breast tumor angiogenesis, causing tumor size and weight reduction [ ].

Only a restricted number of CTCs survive during their spread from their primary development site toward their metastatic target microenvironment(s). The surviving one corresponds to tumor cells that employ various mechanisms to escape from immunity system or to divert it.

It has been very recently showed that during breast tumor cells apoptosis, miR-375 can be released in their close environment. In vitro cocultures between breast tumor cells and macrophages increase intracellular macrophagic miR-375 presence. This miR intracellular uptake occurs through CD36 (cluster of differentiation 36) intervention. Once in the macrophagic cytoplasm, miR-375 directly inhibits TNS3 (Tensin 3) and PXN (Paxillin) genes expression, enhancing macrophages migration and tumor infiltration. Concerning breast tumor cells, miR-375 directly regulates CCL2 (C–C motif chemokine ligand 2) gene expression, increasing macrophages recruitment. Thus, this publication highlights the fact that, among others, miRNAs can be implicated in tumor-associated macrophages (TAMs) development, macrophages that support tumor evolution [ ].

Once tumor cells reach their target microenvironment(s), they can enter in dormancy state and stay inoffensive until they receive appropriate signals, allowing them to install as actual metastases. Because of their slow proliferation during their dormancy, these metastatic tumor cells are often unresponsive to chemotherapy. Metastatic microenvironment(s) play an important role in the establishment of this dormancy state of tumor cells [ ]. Once again, miRNAs have been recently shown to be implicated in this process.

MSCs (mesenchymal stem cells) from BM release exosomes that contain miR-23b toward metastatic breast tumor cells. Once in the cytoplasm, miR-23b initiates breast tumor cells dormancy through the inhibition of MARCKS (myristoylated alanine-rich C-kinase substrate) gene expression, a protein that normally promotes cell cycling progress and motility. In vitro cocultures between metastatic breast tumor cells (previous selected for their bone tropism and considered as breast CSCs) with BM-MSCs induce breast tumor cells dormancy, just like coculture between breast CSCs and BM-MSCs culture media and breast CSCs and exosomes derived from BM-MSCs. Dormancy of bone metastatic breast tumor cells is confirmed by the observations of their proliferation suppression and by stem cell–like surface markers underexpression. Cocultures also inhibit their invasion capacity and promote docetaxel resistance. Analysis on metastatic breast tumor cells extracted from patients' BM shows an increased miR-23b expression in these cells. Exosomal transfer of miR-23b from BM-MSCs to metastatic breast tumor cells impairs MARCKS gene expression, leading to dormancy state establishment of these tumor cells and subsequent chemoresistance development [ ].

Despite all these recent discoveries about miRNAs' implication in bone metastasis development process, their role must continue to be studied to find even more efficient and adapted diagnostic, prognostic, and therapeutic elements/targets, the final goal being to improve the survival of patients affected by bone metastasis.

MicroRNAs and chemoresistance

Among all the cancer treatments available, chemotherapy alone or in association with radiotherapy or surgical resection is currently one of the most commonly used. In the last decades, a wide panel of chemotherapeutic drugs has been developed to improve the survival rate of patients with cancer, and some new ones are still in development today. Cisplatin, ifosfamide, and methotrexate are the three main agents used in the treatment of osteosarcoma [ ], whereas anthracyclins, taxanes (paclitaxel and docetaxel), and endocrine-therapy molecules such as tamoxifen are rather administered in a breast and prostate cancer context. Both compounds act by limiting rapid tumor cell growth and allow a control over disseminated disease. Used as adjuvant therapy, they aimed to limit residual cancerous lesions after surgery and as a consequence, the disease recurrence. In the case of a neoadjuvant use, their goal is to reduce the tumor burden in order to facilitate the surgical resection. However, despite these last decade's advances in the field of therapeutic protocol optimization and personalization, chemoresistance is one of the major obstacle hampering patient's good clinical outcome. There are two main origins of drug treatment failure: intrinsic resistance, due to the inherent early genetic mutations of a subpopulation of heterogeneous cancer cells independently of the drug's presence, and acquired resistance, which is subsequent to the treatment. The introduction of multi-agents chemotherapy is often sufficient to counter intrinsic resistance but it is reported that acquired ones are responsible of more than 90% of advanced cancer relapses [ ]. Evidence of miRNAs-mediated drug resistance in both intrinsic and acquired mechanisms is accumulating. Numerous studies have underlined the importance of miRNAs as essential regulators of various normal cellular processes including proliferation [ ], cell cycle control, apoptosis, or DNA repair in several cell types. As such pathways are specifically hampered by chemotherapeutic agents, deregulation of miRNAs is a mean used by cancer cells to bypass drug's effects. Overexpression of oncogenic miRNAs or loss of function of tumor suppressors' ones is not without consequences and can result in chemoresistance, leading to poor clinical outcome and therapeutic failure. The aim of this second part is to provide an overview of recent findings that reveal the implication of miRNAs in the chemoresistance mechanisms related to osteosarcoma, breast cancer, and prostate cancer. Here, we provide a systematic discussion of several mechanisms regulated by miRNAs that account for the drug-resistant phenotype of bone sarcoma cells by classifying them in (1) alterations in the intracellular drug concentration mediated by efflux pumps, (2) reduced apoptosis, (3) alterations of the cell cycle and the proliferative potential, (4) a facilitated EMT process, or (5) an increased repair of DNA damages. Furthermore, miRNAs are relatively stable in body fluids, poorly degraded by circulating ribonucleases, and these small molecules are able to support severe physicochemical conditions. So, in a clinical extent, we also present the use of miRNAs as relevant predictive biomarkers of the chemotherapeutic drugs response in the context of these pathologies.