The promising role of noncoding RNAs in cancer-associated fibroblasts: an overview of current status and future perspectives

Tumor progression is not an independent event related only to the biological characteristics of tumor cells but instead is a process affected and modulated by many factors. One of these factors, the tumor microenvironment (TME), is vital for tumor initiation, progression, metastasis, immunosuppression and therapeutic resistance [1, 2]. The TME consists of cancer-associated fibroblasts (CAFs), macrophages, myeloid-derived suppressor cells (MDSCs), lymphocytes, antigen-presenting cells and other tumor-associated stromal cells, as well as microvessels, lymphatic vessels, nerves, infiltrating biomolecules and extracellular matrix (ECM) components [3, 4]. As the key components among these constituents, CAFs are abundant in the TME (especially in pancreatic cancer) and interact closely with tumor cells [5]. Fibroblasts are a population of spindle-shaped cells of mesenchymal origin that exhibit distinct phenotypes according to the microenvironment. Fibroblasts activated during wound healing are called “myofibroblasts,” while those activated in tumor tissues are defined as CAFs [6, 7]. Unlike normal fibroblasts (NFs), most CAFs have a myofibroblastic phenotype (termed myCAFs) with characteristics such as overexpression of α-smooth muscle actin (α-SMA) and fibroblast activation protein (FAP) and are in a state of active proliferation and metabolism [8]. Additionally, some CAFs are defined as inflammatory CAFs (iCAFs) due to their secretory properties and inflammatory regulatory function [9]. CAFs contribute substantially to tumor progression in multifarious manners. As the main source of collagen-producing cells in the TME, CAFs provide mechanical support for tumor tissues and regulate the growth and invasion of tumor cells by synthesizing or remodeling the structure of the ECM [10]. Importantly, CAFs can also mediate the interplay between tumor cells and stromal cells by secreting a variety of growth factors, cytokines, chemokines and exosomes, thus affecting the key steps in tumor malignant progression [11, 12]. However, the specific mechanisms through which CAFs play these roles need further exploration.

Only < 2% of the transcripts in the human genome encode proteins; the remaining 98% are noncoding RNAs (ncRNAs). Because they lack complete open reading frames (ORFs), ncRNAs cannot be templates for protein synthesis, but they contribute significantly to the regulation of epigenetic modification levels. NcRNAs are classified into two main categories according to their length: small noncoding RNAs (sncRNAs) and long noncoding RNAs (lncRNAs). SncRNAs are usually less than 200 nucleotides (nt) in length and include microRNAs (miRNAs), piwi-interacting RNAs (piRNAs), small interfering RNAs (siRNAs), and so on. MiRNAs are a class of sncRNAs with a length of approximately 22 nt that mediate inhibition of gene translation or degradation of mRNA by binding to the 3′ untranslated region (3′ UTR) or ORFs of their target mRNA, thereby regulating the expression of target proteins at the posttranscriptional level [13]. NcRNAs with a transcript length of more than 200 nt are defined as lncRNAs; lncRNAs regulate gene expression at the transcriptional or posttranscriptional level through interactions with proteins or nucleic acids, thus affecting many biological processes [14]. Circular RNAs (circRNAs) are a new class of endogenous ncRNAs with a single-stranded, closed-loop structure that lacks 5′ and 3′ ends and a poly(A) tail, which makes circRNAs more stable than linear RNAs [15]. CircRNAs act as miRNA sponges or transcriptional regulators or bind to RNA-binding proteins (RBPs) to regulate gene expression [16]. NcRNAs participate in many physiological and pathological processes in the body, and ncRNA dysregulation is closely related to the development of diseases such as cardiovascular diseases, neurological diseases and cancer [17‐20]. During tumor progression, ncRNAs can regulate the proliferation, stemness, metabolism, differentiation, apoptosis, invasion and drug resistance of tumor cells, while their dysregulation can be affected by gene mutations or epigenetic changes in tumor cells [21‐23]. More importantly, ncRNAs can mediate the interplay between tumor cells and stromal cells via various mechanisms to promote or suppress tumor progression.

The role of ncRNAs in tumor cells has been confirmed by extensive research, and their effects on the TME have received increasing attention. Accumulating evidence shows that miRNAs are involved in the transformation of NFs to CAFs and modulate the crosstalk between CAFs and tumor cells. Although the research on the role of lncRNAs in CAFs is not as extensive and in-depth as that on the role of miRNAs, it has achieved remarkable advances. Importantly, tumor cells and CAFs can communicate more directly and effectively by producing exosomes packed with ncRNAs, which can be taken up by adjacent or distant cells to further contribute to tumor progression. To date, no study has investigated the function of other types of ncRNAs, such as circRNAs, in CAFs. Thus, this review focuses on the studies of miRNAs and lncRNAs. In this review, we introduce the origins and heterogeneity of CAFs; focus on the roles of miRNAs and lncRNAs in the formation and metabolic reprogramming of CAFs as well as on the contribution of miRNAs and lncRNAs in CAFs to aspects of tumor progression, such as proliferation, stemness, angiogenesis, metastasis and therapeutic resistance; and discuss the possible therapeutic targets among these mechanisms.

CAFs are a heterogeneous population of cells, and their heterogeneity might be attributable to their numerous potential cellular precursors (Fig. 1) [6, 9]. Tissue-resident NFs can be activated as CAFs under the stimulatory effects of the TME. Transforming growth factor-β (TGF-β), epidermal growth factor (EGF), platelet-derived growth factor (PDGF), fibroblast growth factor 2 (FGF2), sonic hedgehog (SHH), bone morphogenetic protein (BMP) and reactive oxygen species (ROS) are pivotal regulators of fibroblast activation [6, 24‐26]. For example, quiescent pancreatic stellate cells (PSCs) and hepatic stellate cells (HSCs), tissue-resident fibroblasts of the pancreas and liver, can acquire a myofibroblast-like phenotype including characteristics such as α-SMA expression when activated by TGF-β and PDGF and can then transform into CAFs [27, 28]. Mesenchymal stem cells (MSCs) might be another source of CAFs [29]. By fluorescent protein labeling, Raz et al. found that bone marrow-derived MSCs can be recruited and reprogrammed into different CAF subsets [30]. In breast cancer, osteopontin mediates the MZF1-TGF-β1-dependent differentiation of MSCs into CAFs [31]. HOXA9 expressed in epithelial ovarian cancer cells induces adipose and bone marrow-derived MSCs to acquire the CAF phenotype through transcriptional activation of the gene encoding TGF-β2 [32]. These studies indicated that TGF-β plays an important role in this process. In prostate cancer, MSCs can differentiate into CAFs after activation by C-X-C motif chemokine receptor 6 (CXCR6) and its ligand CXCL16 [33]. Additionally, the recruitment and transformation of MSCs also depend on CXCL12, CCL2 and CCL5 secreted in the TME [29, 34]. CAFs might also be differentiated from nonfibroblast lines such as epithelial cells [35, 36], endothelial cells [37], adipocytes [38], pericytes [39] and smooth muscle cells [40]. Epithelial cells and endothelial cells transdifferentiate into CAFs through epithelial-to-mesenchymal transition (EMT) or endothelial-to-mesenchymal transition (EndoMT), after which they express an abundance of mesenchymal markers such as α-SMA and FAP [41, 42].

In summary, the CAF population is composed of multiple cell subsets and may have diverse phenotypes and functions. However, overall, the exact origins of CAFs have not been fully clarified. CAFs express a variety of proteins, such as α-SMA, PDGF receptor-α/β (PDGFRα/β), FAP, fibronectin, kindlin-2, tenascin C and integrin-α11 [43‐46]. However, the expression levels of these proteins vary among different CAF subsets, and none of these proteins is specific to NFs or CAFs [47]. To further identify the origins of CAFs, researchers have used mouse models with genetic lineage tracing and fluorescence tagging, which can suitably describe the models of disease progression [30, 48]. Although these models have undergone technical progress, identifying the exact biological origin of CAFs remains difficult because of the lack of specific markers for NFs and CAFs [49, 50].

As previously described, the CAF population is composed of multiple cell subsets, which typically exhibit heterogeneity. In pancreatic cancer, a study grouped CAFs into two subsets: myCAFs and iCAFs [9]. MyCAFs are located adjacent to cancer cells, are characterized by a high level of α-SMA expression and exhibit a matrix-producing contractile phenotype. In contrast, iCAFs are located further from cancer cells and are recognized by their high expression of tumor-promoting cytokines and chemokines such as interleukin-6 (IL-6). Further, Elyada et al. confirmed the existence of myCAFs and iCAFs in pancreatic cancer by single-cell RNA sequencing [51]. Interestingly, in that study, the researchers also identified a new subset of CAFs and defined these cells as “antigen-presenting CAFs” (apCAFs), which express MHC-II and CD74 but do not express classical costimulatory molecules. ApCAFs can activate CD4+ T cells in an antigen-specific manner, indicating their immunoregulatory ability. In another study, CAFs in pancreatic cancer were divided into at least four subtypes based on a transcriptomic analysis [52]. Subtype A cells labeled with periostin were found in the invasive front of the primary tumor, which is important for the formation of tumor capsule and metastasis. Subtype B cells labeled with myosin-11 were enriched in larger tumors and were associated with lymph node metastasis and poor prognosis. Subtype C cells labeled with podoplanin were enriched in a variety of immune-related pathways and seem to have an immunogenic profile and are related to a better prognosis. In addition, in breast cancer, a recent study identified four CAF subpopulations (S1-4) with different distributions and unique properties among breast cancer subtypes [53]. Among these subpopulations, S1-CAFs, which overexpress FAP, can attract T cells and promote their differentiation into Treg cells, thus contributing to an immunosuppressive TME. In contrast to promoting tumor growth, some subtypes of CAFs may be able to suppress tumor progression. For example, Rhim et al. found that the number of α-SMA + myofibroblasts was decreased in the TME of mice with Shh-deficient pancreatic ductal adenocarcinoma (PDAC), resulting in poor differentiation and stronger invasiveness [54]. Additionally, Ozdemir et al. demonstrated that depletion of α-SMA + fibroblasts leads to more invasive and undifferentiated tumors [50]. In some cases, different subtypes of CAFs in the same TME may have opposite effects on cancer progression. For instance, two CAF subtypes that differentially express CD146 play conflicting roles in the responsiveness of breast cancer to endocrine therapy [55]. Importantly, Givel et al. grouped CAFs into four subpopulations in mesenchymal high-grade serous ovarian cancer based on CD29, FAP, FSP1, and SMA protein levels [56]. CAF-S1 and CAF-S4 fibroblasts were defined as activated CAFs because of their high SMA expression levels, while CAF-S2 and CAF-S3 fibroblasts were considered to be nonactivated CAFs with low levels of SMA. CAF-S1 fibroblasts can improve the survival of CD25⁺FOXP3⁺ T lymphocytes and result in immunosuppression, which is mediated by the high expression of CXCL12β in CAF-S1. The different levels of CXCL12β in CAF-S1 and CAF-S4 fibroblasts lead to different functions among CAF subpopulations, which ultimately affects the prognosis of cancer patients [56, 57].

Due to the heterogeneous phenotypes, functions and spatial distribution of CAFs, researchers began to consider whether different subtypes can be interconverted—that is, whether CAFs exhibit plasticity. CAFs were found to be able to transition from the α-SMA+ and IL-6-producing state through TGF-β-dependent manipulation [58]. Additionally, prostate CAFs can acquire a tumor-like phenotype via a novel lipid-microtubule-organizing center (MTOC) axis regulated by pigment epithelium-derived factor (PEDF) [59]. In pancreatic cancer, the activation of the JAK/STAT signaling pathway induces the iCAFs phenotype, while JAK inhibitors can transform iCAFs into myCAFs; TGF-β signaling can also antagonize IL1-induced JAK/STAT signaling, thus inhibiting the iCAFs phenotype [60]. The phenotypic and functional transformation of CAFs reflects their instability and plasticity, which is a potential target for cancer therapy. However, considering the profound adaptability of fibroblasts to direct, local and long-term stimulatory factors, it is difficult to determine whether targeting specific CAF subsets will have lasting effects [61]. Many signaling pathways, such as EGFR signaling, Wnt signaling, Hippo signaling and TGF-β signaling, have been shown to play a key role in the regulation of CAF heterogeneity [57]. These signaling pathways may be another important direction for CAF-targeted therapy, and further studies of these pathways may help improve our understanding of CAF heterogeneity.

The importance of the TME, as a key regulator of tumor progression, has received increasing attention. A deeper understanding of CAFs, a kind of stromal cell that is the main component of the TME, is very helpful for us to further clarify the mechanisms by which the TME contributes to tumor progression. CAFs are a heterogeneous population of cells of multiple origins that mainly promote the development of tumors. However, some recent studies have shown that they can also have tumor-suppressive effects. Although some studies have reported that gene mutations are present in CAFs, such mutations are generally believed to be uncommon. Therefore, the phenotypic and functional changes in CAFs are regulated mainly by other factors, among which ncRNAs play a crucial role. As reviewed herein, miRNAs participate in the formation and activation of CAFs and modulate their metabolic reprogramming, which is one of the important characteristics that distinguishes CAFs from NFs and is a prerequisite for the role of CAFs. This process can be accomplished through miRNA dysregulation in CAFs or mediated by exosomal miRNAs secreted from tumor cells. Subsequently, through miRNAs, activated CAFs can regulate numerous biological functions in tumor cells, including proliferation, stemness, metabolism, angiogenesis, invasion, metastasis and the acquisition of therapeutic resistance. These regulatory effects can be exerted through miRNA dysregulation in CAFs, which leads to the transformation of the matrix-regulatory function of CAFs and the secretion of related molecules to regulate tumor cells. Similarly, through secreted exosomes, exosomal miRNAs can be transferred to tumor cells and act directly on the corresponding targets to regulate a series of biological characteristics. It can be seen that miRNAs are involved in almost all aspects of communication between CAFs and tumor cells. Moreover, studies on lncRNAs in CAFs are being carried out gradually and have shown the expected results, which reveal that lncRNAs also contribute to tumor progression via many mechanisms. Interestingly, some studies have shown that lncRNAs act as miRNA sponges to regulate the level of miRNAs, linking the study of lncRNAs with the study of miRNAs. However, most of the studies mentioned in our review have focused on the mechanism of a single signaling pathway mediated by a single ncRNA, and further research on the complex regulatory networks in the TME is lacking. Therefore, the conclusions of our studies may be one-sided and address only part of a large and complex network. As research on ncRNAs intensifies, we should conduct more in-depth studies on the regulatory networks mediated by ncRNAs to identify both the overall function of a single ncRNA and the synergistic functions of multiple ncRNAs and to validate the driving factors and causal relationships. This knowledge will help us to identify the optimal single or multiple ncRNA targets in order to design more effective and safer treatments. Last but not least, the in-depth study of ncRNAs in CAFs can enable us to further understand the nature and biological characteristics of CAFs and may facilitate the identification of additional treatments targeting CAFs, rather than just treatments targeting ncRNAs.