Transforming growth factor‐β signalling in renal fibrosis: from Smads to non‐coding RNAs

Abstract

Transforming growth factor‐β (TGF‐β) is the key player in tissue fibrosis. However, antifibrotic therapy targeting this multifunctional protein may interfere with other physiological processes to cause side effects. Thus, precise therapeutic targets need to be identified by further understanding the underlying mechanisms of TGF‐β1 signalling during fibrogenesis. Equilibrium of Smad signalling is crucial for TGF‐β‐mediated renal fibrosis, where Smad3 is pathogenic but Smad2 and Smad7 are protective. The activation of TGF‐β1/Smad signalling triggers extracellular matrix deposition, and local myofibroblast generation and activation. Mechanistic studies have shown that TGF‐β/Smad3 transits the microRNA profile from antifibrotic to profibrotic and therefore promotes renal fibrosis via regulating non‐coding RNAs at transcriptional levels. More importantly, disease‐specific Smad3‐dependent long non‐coding RNAs have been recently uncovered from mouse kidney disease models and may represent novel precision therapeutic targets for chronic kidney disease. In this review, mechanisms of TGF‐β‐driven renal fibrosis via non‐coding RNAs and their translational capacities will be discussed in detail.

Introduction

Renal fibrosis, characterized by excessive deposition of extracellular matrix (ECM), is recognized as a common pathological feature of chronic kidney diseases (CKD), accompanied by a progression of renal malfunctions (Eddy & Neilson, 2006; Eddy, 2014). While effective therapy for renal fibrosis is still lacking, a number of studies have demonstrated that transforming growth factor‐β (TGF‐β) is the key mediator that drives glomerular and tubulointerstitial fibrosis (Wang et al. 2005b), and considerable evidence has revealed that it is substantially up‐regulated in the injured kidney in both CKD patients and animal disease models (Yamamoto et al. 1993; Sato et al. 2003). This underlies the essential regulatory role of TGF‐β in tissue fibrosis (Border et al. 1990; Yamamoto et al. 1993; Russo et al. 2007). However, TGF‐β1 is not an ideal therapeutic target for fibrotic diseases due to its importance in normal physiological processes (Rodrigues‐Diez et al. 2015). Hence, precise therapeutic targets should be further identified by resolving the underlying molecular mechanisms of TGF‐β1‐driven fibrosis. In renal fibrosis, TGF‐β binds to the plasma membrane receptor; Smad signalling is then initiated by Smad2/3 phosphorylation and complexed with Smad4 to translocate into the nucleus (Yan et al. 2016). The Smads complex regulates transcription of target genes in order to trigger ECM accumulation, and myofibroblast generation and activation in the injured sites (Lan & Chung, 2012). Follow‐up mechanistic studies revealed that TGF‐β/Smad3 signalling promotes renal fibrosis through downstream transcriptomes, including non‐coding RNAs (Chung et al. 2010a; Putta et al. 2012; Zhong et al. 2013; Lin et al. 2014; Chung & Lan, 2015). Recently, a number of disease‐specific Smad3‐dependent long non‐coding RNAs have been identified in mouse CKD models. In this review, current findings regarding TGF‐β‐driven renal fibrosis and potential targeted therapy will be extensively discussed.

Diversity of TGF‐β family

Transforming growth factor‐β (TGF‐β) is not only the prototype of its highly homologous isoforms but also represents a large family of over 40 different related proteins. Three distinct isoforms of TGF‐β (TGF‐β1, TGF‐β2 and TGF‐β3) have been identified in mammals which share 70–82% homology at the amino acid level. Other proteins in the family, including the bone morphogenetic proteins (BMPs), show unique biological activities (Yu et al. 2003; Meng et al. 2013). Interestingly, the three mammalian isoforms of TGF‐β, TGF‐β1, 2 and 3, originated from different chromosomes (human chromosomes 19q13, 1q41 and 14q24, respectively) (Romeo et al. 1993). The promoter of TGF‐β1 is distinguished from those of TGF‐β2 and 3 by replacing the classic TATAA box with multiple regulatory sites at its promoter region (Kim et al. 1990). These features enable the TGF‐β1 promoter to be activated directly by transactivating proteins, such as reactive oxygen species (ROS), plasmin and acid, providing a mechanistic basis for specifically overexpressing this isoform during tissue repair, stress, pathogenesis of viral‐mediated diseases, carcinogenesis and organ fibrogenesis (Roberts, 1998; Wang et al. 2005b). As a result of the special promoter construction among its family, TGF‐β1 is well known as a key profibrotic mediator in fibrotic diseases, particularly in renal fibrosis.

Roles of TGF‐β1 in renal fibrosis

TGF‐β1 is the primary factor driving tissue fibrosis in organs such as bone, lung, placental tissue and, in particular, kidney. TGF‐β1 is significantly up‐regulated in the injured kidney as an initial step of tissue scarring. TGF‐β1 and its isoforms are synthesized and secreted as latency‐associated peptide (LAP), which binds to latent TGF‐β binding proteins (LTBP) (Annes et al. 2003; Robertson et al. 2015). The latent TGF‐β complex is then cleaved by a wide range of proteases, e.g. plasmin, matrix metalloproteinase (MMP)‐2 and ‐9, and thrombospondin, to provide active TGF‐β (Yuan & Varga, 2001; Wang et al. 2013a). The active TGF‐β binds to the extracellular domain of TGF‐β receptor type II (TβRII), which activates TGF‐β receptor type I (TβRI) kinase, and thus further triggering downstream signalling pathways for exerting its biological functions (Derynck & Zhang, 2003). In general, the release and activation of TGF‐β1 promotes renal fibrosis through both direct and indirect actions on various cell types.

TGF‐β1 acts directly on different types of renal cells to induce profibrotic responses, including cell proliferation, migration, activation and transcription of various extracellular matrix proteins, and inhibits their degradation (Lopez‐Hernandez & Lopez‐Novoa, 2012). Activation of TGF‐β1 signalling can also induce podocyte apoptosis, which leads to podocytopenia, therefore promoting the development of progressive glomerulosclerosis (Bottinger & Bitzer, 2002; Lopez‐Hernandez & Lopez‐Novoa, 2012). The direct effect and the underlying mechanisms of TGF‐β1 on renal fibroblasts have been well studied. There is evidence illustrating that TGF‐β1 stimulates the production of types I and IV collagen, laminin, fibronectin and heparan sulfate proteoglycans, but inhibits the degradation of ECM (Samarakoon et al. 2012). In addition, TGF‐β1 also promotes renal fibrosis by inducing the transformation of tubular epithelial cells to myofibroblasts through epithelial–mesenchymal transition (EMT) (Lan, 2003).

Nevertheless, TGF‐β1 also mediates the profibrotic response through a number of indirect mechanisms. For example, some studies have demonstrated that TGF‐β1 induces apoptosis in cultured tubule epithelial cells (Bhaskaran et al. 2003; Docherty et al. 2006). However, the extent of TGF‐β1‐induced cell apoptosis is low in these experiments. It is believed that either the pro‐apoptotic effect of TGF‐β1 only takes place with the cooperation of other mediators or cytokines under pathological circumstances, or the damage‐amplifying mechanisms are only present in vivo but not in vitro. The role of TGF‐β1 is a prototypic multifunctional cytokine in apoptosis regulation, which induces not only collagen synthesis but also autophagy and collagen degradation at different stages of the disease development (Ding & Choi, 2014).

TGF‐β1/Smad signalling: the canonical pathway in renal fibrosis

TGF‐β1 signalling acts through both canonical and non‐canonical pathways (Derynck & Zhang, 2003). The canonical signalling pathway involves the phosphorylation of Smad2 and Smad3 by activating TβRI (type I TGF‐β receptor) to convey intracellular signals via the phosphorylation of receptor‐regulated Smads (R‐Smads) (Wrana et al. 1994). Consistent results from different studies indicated that the TGF‐β1/Smad‐dependent mechanisms are essential for induction of EMT and the accumulation of ECM. Overexpression of Smad3 inhibits MMP‐1 activity in fibroblasts dependent on the ubiquitin E3‐ligase‐mediated Smad7 suppression, which imbalances the Smads’ signalling and therefore triggers fibrogenesis locally in the injured tissues (Yan et al. 2016).

In the context of TGF‐β1/Smad signalling, Smad3 is confirmed to be pathogenic, because deletion of Smad3 inhibits renal fibrosis induced by almost all aetiologies, including obstructive, diabetic, hypertensive and acute toxic nephropathy (Sato et al. 2003; Wang et al. 2006; Yang et al. 2009; Zhou et al. 2010; Chung et al. 2010b; Lan, 2012). TGF‐β1 also induces tissue inhibitor of MMP‐1 (TIMP‐1) through Smad3 to inhibit ECM degradation, and overexpression of Smad3 inhibits MMP‐1 activity in fibroblasts, suggesting a pathogenic role of Smad3 during renal fibrosis (Yuan & Varga, 2001).

In contrast, researchers found that Smad2 physically complexes with Smad3 and works in a non‐redundant manner during embryonic development. Unexpectedly, Smad2 plays a protective role in renal fibrosis, although it shares more than 90% structural similarity with Smad3 (Massague & Wotton, 2000). Our studies demonstrated that deletion of Smad2 significantly enhances renal fibrosis, which is associated with enhancement of Smad3 activities including phosphorylation, nuclear translocation, mRNA expression and transcriptional regulation (Meng et al. 2010). Indeed, TGF‐β1 is also a well‐known mediator for angiogenesis, where impaired angiogenesis leads to the progressive renal fibrosis (Sekiguchi et al. 2012). Smad2, but not Smad3, acts as an anti‐angiogenic factor for mediating the expressions of thrombospondin‐1 (TSP‐1) and a truncated form of the VEGF receptor Flt‐1, soluble fms‐like tyrosine kinase‐1 (sFlt‐1), in response to TGF‐β1 (Meng et al. 2010).

Smad4 is a common Smad for both TGF‐β signalling and BMP signalling. It serves as a key molecule for shuttling Smad2/3 into the nucleus (Meng et al. 2013). Interestingly, deletion of Smad4 does not alter phosphorylation nor nuclear translocation of Smad2/3, but largely inhibits the DNA binding activity of Smad3 to the promoter region of collagen (Tsuchida et al. 2003) and ECM deposition both in vivo and in vitro (Meng et al. 2012b). Smad4 is also reported to be an anti‐inflammatory mediator; conditional deletion of Smad4 in tubular epithelial cells enhances the inflammatory response in the obstructed kidney demonstrating an important regulatory role of TGF‐β/Smad signalling in renal inflammation (Meng et al. 2012b). Collectively, Smad4 specifically regulates the transcriptional activity of Smad3 to initiate expression of its targeted genes, instead of the nuclear shuttling of Smad3.

Smad7 is an inhibitory Smad induced by TGF‐β1, which prevents the hyperactivation of TGF‐β1 signalling by inhibiting TβRI and Smad2/3 (Yan & Chen, 2011). In chronic kidney disease, TGF‐β1 not only up‐regulates the Smad7 expression level but also activates the Smurf‐ and Arkadia‐dependent ubiquitin–proteasome pathways, which in turn degrade Smad7 protein via a post‐transcriptional mechanism thereby enhancing TGF‐β/Smad3‐mediated renal fibrosis (Fukasawa et al. 2004). Smurf1, Smurf2 and Arkadia are E3 ubiquitin ligases targeting Smad7 via physical interactions. Smad7 acts as an adaptor protein to recruit E3 ubiquitin ligases to the TGF‐β1 receptor complex to promote its degradation through the proteasomal–ubiquitin degradation pathway (Yan et al. 2016). In addition, Smad7 is able to inhibit renal inflammation (Fukasawa et al. 2004), since TGF‐β/Smad3 signalling also enhances renal inflammation by activating the nuclear factor‐κB (NF‐κB) ‐dependent inflammatory response (Liu et al. 2013). Thus, the imbalance of Smads is a key process for TGF‐β‐driven renal fibrosis (summarized in Fig. 1).

Figure 1. Overview of canonical TGF‐β1/Smads signalling in tissue fibrosis.

For initiating the canonical TGF‐β1/Smads signalling, latency‐associated peptide (LAP) binds to latent TGF‐β1 binding proteins (LTBP), then the complex is cleaved by proteases to release the active TGF‐β1. The active form of TGF‐β1 binds to its receptor II (TβRII), which activates the TGF‐β receptor type I (TβRI) kinase that further phosphorylates Smad2 and Smad3. The activated Smad2 and Smad3 then complex with Smad4 and translocate into the nucleus. Meanwhile, TGF‐β1 also activates Smad ubiquitin regulatory factor (Smurf) to degrade Smad7 via post‐transcriptional modification thereby enhancing TGF‐β/Smad3 signalling. Thus, imbalance of Smads signallings is a key for fibrogenesis, which ultimately accelerates the accumulation of extracellular matrix (ECM), myofibroblasts to epithelial–mesenchymal transition (EMT), endothelial–mesenchymal transition (EndoMT), macrophage–myofibroblast transition (MMT), apoptosis and angiogenesis.

Controversy of targeting TGF‐β1/Smad signalling

The TGF‐β1 signalling pathway is a well‐documented pathogenic promoter of renal fibrosis, suggested by the beneficial effects of TGF‐β1 blockade in a number of animal kidney disease models. Several strategies that target the upstream signalling of TGF‐β1 have been developed by using neutralizing antibodies (anti‐TGF‐βIgG4), antisense TGF‐β1 oligodeoxynucleotides, soluble human TβRII fragment‐crystallizable region (sTβRII.Fc) and specific inhibitors for TβR kinases (GW788388 and IN‐1130); they have shown antifibrotic effects in experimental models (Lan & Chung, 2012; Meng et al. 2012a). Although the critical role of TGF‐β1 in renal fibrosis has been well studied, tremendous attention has been paid to its role in renal inflammation. TGF‐β1 is also suggested as an anti‐inflammation cytokine (Rodrigues‐Diez et al. 2015). TGF‐β1‐deficient mice resulted in lethal inflammation and death at 3 weeks of age (Letterio & Roberts, 1997; Levéen et al., 2002). In contrast, overexpression of latent TGF‐β1 protected mice against progressive renal inflammation and fibrosis induced by obstructive and immunologically induced crescentic glomerulonephritis (Wang et al. 2005a; Huang et al. 2008a, b ). The mechanism of activation is more complex than initially proposed. A study showed that the LAP/TGF‐β complex binds to a regulatory T cell receptor GARP (also known as LRRC32), and has an important role in regulatory T cell phenotype and function (Robertson et al. 2015). Conditional deletion of the TβRII or TGF‐β1 gene from T cells leads to autoimmune diseases. This interaction might partially explain why transgenic overexpression of latent TGF‐β1 reduces both tissue inflammation and fibrosis in kidney disease mouse models (Huang et al. 2008a, b ). Similarly, deletion of TβRII protects against renal fibrosis but enhances NF‐κB signalling, along with the up‐regulation of renal pro‐inflammatory cytokines, which may be associated with the impairment of TGF‐β1 signalling (Meng et al. 2012a).

Indeed, Smad3 is one of the most studied downstream regulators of the TGF‐β1 signalling cascades in renal fibrosis. Solid evidence showed that targeting Smads ameliorates renal fibrosis. Smad3‐deficient mice protected against kidney injury associated with suppression in collagen deposition compared to the wild‐type mice (Bottinger & Bitzer, 2002; Huang et al. 2008a). In contrast, deletion of Smad2 in tubular epithelial cells accelerated unilateral ureteral obstruction (UUO)‐induced renal fibrosis in mice by enhancing Smad3 activation. Gene transfer‐mediated overexpression of Smad7 attenuated renal fibrosis in mouse models of kidney disease including aristolochic acid nephropathy, hypertensive nephropathy and acute kidney disease (Dai et al. 2015; Fu et al. 2017; Liu et al. 2017). Smad3‐specific inhibitors, e.g. SIS3 and GQ5, effectively suppress renal fibrosis in mouse models with diabetic and UUO‐induced nephropathy (Lan, 2012).

However, diverse roles of TGF‐β1/Smads in fibrosis and immunity have hampered the development of anti‐TGF‐β1 treatment. A series of animal studies have demonstrated that plasmid‐based gene transfer‐mediated overexpression of Smad7 can suppress renal fibrosis and reduce NF‐κB‐driven inflammation, but this approach is limited by the fact that Smad7 also promotes podocyte apoptosis (Schiffer et al. 2001). Furthermore, mice lacking either TGF‐β1 or Smad3 show impaired immunity including fatal multi‐organ inflammation and autoimmune disease (Kulkarni et al. 1993). Nevertheless, Smads also interact with other signalling pathways, such as the mitogen‐activated protein kinase (MAPK) and NF‐κB pathways, to regulate renal inflammation and fibrosis (Li et al. 2004; Ka et al. 2007; Huang et al. 2008b). In CKD, other factors like advanced glycation end products (AGEs) and angiotensin II (Ang‐II) can bind to their respective receptors and then activate the downstream TGF‐β signalling by crosstalking with Smads via ERK/p38 MAPK‐dependent mechanisms (Li et al. 2004; Wang et al. 2006). All of these studies revealed the complexity of TGF‐β1/Smads signalling during disease development. More importantly, these findings implied that targeting TGF‐β/Smad signalling may not be an ideal solution for kidney diseases (Lan & Chung, 2012). Furthermore, direct targeting of TGF‐β1 or its receptor is unlikely to be a therapeutically feasible option due to the involvement of TGF‐β1 in systems such as the immune system. Thus, further research is needed to identify a precision therapeutic target to inhibit TGF‐β1‐induced tissue fibrosis.

Non‐coding RNAs in TGF‐β1/Smad3‐driven renal fibrosis

MicroRNAs (miRNAs) are short RNA molecules of 20–22 nucleotides in length encoded from the genomic DNA sequence without protein‐coding activity. The miRNAs bind to their respective target mRNAs and recruit the RNA‐induced silencing complex (RISC). There are multiple steps in the biogenesis of miRNAs (Filipowicz, 2005). TGF‐β1 regulates a number of miRNAs to promote renal fibrosis in kidney diseases. The regulatory roles of TGF‐β1/Smad3 in miRNA‐mediated renal fibrosis have been demonstrated in an experimental mouse model of obstructive nephropathy by miRNA microarray and real‐time polymerase chain reaction (qPCR) (Chung & Lan, 2015).

Early studies by microarray assays demonstrated the abundance of miR‐192, ‐194, ‐204, ‐215 and ‐216 in the kidney compared with other organs. These studies suggested the biological role of miRNAs in kidney function. So far, the physiological roles of a number of miRNAs have been uncovered in normal as well as diseased kidneys. TGF‐β1 is able to up‐regulate miR‐21, miR‐93, miR‐192, miR‐216a, miR‐217, miR‐377, miR‐382 and miR‐433, but down‐regulate the miR‐29, miR‐200 families, miR‐491‐5p, and so on (Kriegel et al. 2010; Krupa et al. 2010; Liu et al. 2010; Kantharidis et al. 2011; Lan, 2011; Chung et al. 2013a; Li et al. 2013; Wang et al.2013b). All these TGF‐β1‐mediated miRNAs participate in renal fibrosis (Lan & Chung, 2012; Chung et al. 2013b). The miR‐200 family contains miR‐200a, miR‐200b, miR‐200c, miR‐429 and miR‐141(Choi & Ng, 2017). Studies revealed the expression of miR‐21, miR‐29, miR‐192, miR‐200 and miR‐433 via Smad3‐dependent mechanisms (Chung et al. 2010a; Qin et al. 2011; Zhong et al. 2011); so far Smad3‐binding sites have been identified in the promoter regions of miR‐21, miR‐29, miR‐192 and miR‐200. Conversely, the TGF‐β1‐mediated miRNAs also interact with Smad3 to mediate its activity and functions.

Interestingly, mice lacking Smad3 are protected against renal fibrosis associated with miR‐21 and miR‐192 reduction (Chung et al. 2010a; Zhong et al. 2011). By contrast, severe renal fibrosis in the obstructive nephropathy is associated with the loss of miR‐29, which is prevented in Smad3‐knockout mice. It is also reported that overexpression of miR‐200a decreases Smad3 activity and therefore attenuates TGF‐β1‐induced fibrosis (Wang et al. 2011). Clinically, the presence of miRNAs in blood and urine have been examined as early biomarkers for renal diseases. In line with this notion, targeting miRNAs showed therapeutic effects on experimental animal studies with CKD (Ben‐Dov et al. 2014; Van Craenenbroeck et al. 2015) (summarized in Table 1).

Table 1.

TGF‐β1/Smad3‐regulated microRNAs (miRNAs) in renal fibrosis

Antifibrotic	Targetgene	Profibrotic	Targetgene	Profibrotic or antifibrotic	Targetgene
miR‐29	Col, MMP, Fos, Adams, TGF‐β1/2, HDAC4	miR‐21	ERK/MAPK,PTEN, Smad7, PPARα, Spry1	miR‐145	Latent TGF‐β1, KLF4, TGFR2
miR‐let‐7	TGFR1	miR‐433	Azin1	miR‐192	P53,Zeb1/2/E‐cadherin
miR‐15b	TGFR1	miR216a	PTEN	miR‐200	TGF‐β2,Zeb1/2/E‐cadherin
miR‐19b	TGFR2	miR‐217	PTEN
miR‐26a	Smad4	miR‐377	SIRT1
miR‐30	TGF‐β2,Snail	miR‐382	HSPD1,SOD2
miR‐101	TGFR1	miR‐491‐5p	Par‐3
miR‐130b	TGFR1	miR‐17‐5p	Smad7

Although emerging evidence shows that miRNAs play a critical role in kidney diseases, the off‐target effects, avoidance of internal nucleases, and toxicity hinder the clinical application of miRNA‐based therapy. This is because miRNAs non‐specifically act as cofactors instead of downstream effectors in the signalling pathways. In contrast, long non‐coding RNAs (lncRNAs), defined as RNAs longer than 200 nucleotides without protein‐coding activity, show high specificity in their biological functions. Recently, we have revealed the involvement of Smad3‐dependent lncRNAs during renal injury in mouse models of immunologically induced anti‐glomerular basement membranous glomerulonephritis (anti‐GBM GN) and non‐immune disease of unilateral ureteral obstructive nephropathy (UUO) by using high‐throughput RNA sequencing (RNA‐seq). Of these, 21 lncRNAs were significantly regulated in both UUO and anti‐GBM GN models, while all being mediated in a Smad3‐dependent manner. Among these novel lncRNAs (Zhou et al. 2014), Arid2‐IR (np_28496) is a Smad3‐associated lncRNA, as the Smad3 binding site was identified in the promoter region of Arid2‐IR. Knockdown of Arid2‐IR significantly inhibited renal inflammation via suppressing the NF‐κB‐dependent inflammatory response in tubular epithelial cells as well as mice with kidney injury (Zhou et al. 2015). Another recently identified Smad3‐dependent lncRNA Erbb4‐IR (np_5318) has been demonstrated to promote renal fibrosis in both UUO‐induced and diabetic nephropathy mouse models through suppressing renoprotectants Smad7 and miR‐29b, respectively. More importantly, silencing of renal Erbb4‐IR significantly inhibited the expression of collagen I in both diabetic‐ and obstructive‐injured kidneys in vivo (Sun et al. 2017b; Feng et al. 2018). These studies revealed that lncRNA may serve as a competing endogenous RNA (ceRNA) or a molecular sponge for modulating the expression and biological functions of miRNAs.

In addition, Shi's group compared the transcriptome profiles of renal tissues from UUO‐injured and normal rats by RNA‐seq. In total, 24 lncRNAs were up‐regulated and 79 lncRNAs were down‐regulated in the renal tissues of the UUO rats, where 19 of them may contain conserved Smad3‐binding motifs on the promoter regions. Among them, lncRNAs with a putative promoter containing more than four conserved Smad3‐binding motifs were demonstrated to be induced by TGF‐β significantly in normal rat renal tubular epithelial NRK‐52E cells. The researchers further confirmed that lncRNA TCONS_00088786 (neighbour of CtsD) and TCONS_01496394 (neighbour of ATF3) were tightly regulated by TGF‐β stimulation and could also influence the expression of some fibrosis‐related genes through a feedback mechanism (Sun et al. 2017a). Nevertheless, Xie et al. revealed that lncRNA‐H19 expression was significantly up‐regulated in TGF‐β2‐induced HK‐2 cell fibrosis and UUO‐induced renal fibrosis in vivo (Xie et al. 2016).

Nevertheless, lncRNA research is still a challenge as the methods used are mostly based on the existing platforms. In situ hybridization and northern blot are the most commonly used setups to detect the expression level of a lncRNA in vitro and in vivo. Our recent study demonstrated that the origin of a novel lncRNA could be identified by a combination of fluorescent in situ hybridization (FISH) and immunohistochemistry (IHC) staining on diabetic injured kidney section (Sun et al. 2017b) (Fig. 2). Besides, RNA‐seq is always conducted for uncovering novel lncRNAs (Zhou et al. 2015), but increasing evidence suggests the presence of non‐polyadenylated lncRNAs, which were lost during the traditional library preparation where only transcripts with the poly(A) tail were preserved. Furthermore, regulatory protein bindings on lncRNAs at transcriptional level can be routinely predicted by software (e.g. ECR browser; Zhou et al. 2015; Sun et al. 2017b; Feng et al. 2018), but platforms specific for lncRNA interactions with other biomolecules are still not available.

Figure 2. A novel Smad3‐dependent lncRNA Erbb4‐IR is highly expressed in glomerular mesangial cells of diabetic injured kidney in vivo .

The lncRNA Erbb4‐IR was highly expressed in the glomerular mesangial cells and tubular epithelial cells of diabetic injured kidney in the 12‐week‐old db/db mice compared to the normal kidney of db/m mice. The formalin‐fixed paraffin‐embedded (FFPE) kidney sections of db/m and db/db mice were subjected to Erbb4‐IR‐specific FISH assay followed by immunofluorescent co‐staining of epithelial cell marker keratins (FITC‐conjugated) and DAPI. Representative images of 3 independent experiments are shown. Magnification 400x. g; glomerulus.

Indeed, it has been demonstrated that change of renal lncRNA profiles are associated with diabetic nephropathy (DN) progression, including inflammatory and fibrotic responses (Ding et al. 2012). For example, we demonstrated that Erbb4‐IR is pathogenic in DN (Sun et al. 2017a). ASncmtRNA‐2 is upregulated by ROS and may promote glomerular fibrosis in DN via positively regulating the expression of TGF‐β1 (Gao et al. 2017). Non‐coding RNA CJ241444 (CJ24) and Atg2a are originated from the upstream genomic sequence of miR‐192; Smad‐binding elements (SBEs) were identified in the 3′UTR of Atg2a and two potential Smad binding sites and five potential Ets‐1‐binding elements (EBEs) (four EBE clusters) at the upstream region of the CJ24 gene. The study further confirmed that upstream Ets‐1 binding sites are negative regulators of the CJ24 upstream region, whereas the EBE‐1300 site appears to be critical for the TGF‐β response (Kato et al. 2013). A number of new lncRNAs have been uncovered from kidney disease animal models; intensive characterizations should be done to further identify their biological roles in the development of renal fibrosis (summarized in Table 2).

Table 2.

TGF‐β/Smads‐mediated lncRNAs

lncRNA	Kidney disease model	Pathological output(s)	Reference
Erbb4‐IR	UUO mice db/db mice	Fibrosis, miR‐29b	Feng et al. (2018) Sun et al. (2017b)
Arid2‐IR	UUO	Inflammation	Zhou et al. (2014)
TCONS_00088786	UUO rats	Fibrosis	Sun et al. (2017b)
TCONS_01496394	UUO rats	Fibrosis	Sun et al. (2017b)
lncRNAH19	UUO mice	Fibrosis, miR‐17	Xie et al. (2016)
lncRNA‐MGC	db/db mice	Fibrosis, miR‐192	Kato et al. (2016)

Prospects of lncRNA‐targeted therapy for kidney diseases

Targeting TGF‐β/Smad3 may impair host immunity and cause autoimmune diseases. Lessons learnt from the Smad3‐knockout mice strongly suggested that development of therapeutics based on the TGF‐β/Smad3 signalling should target the downstream transcriptomes that are specific for renal inflammation and fibrosis without affecting the immune system. Thus, we proposed to target the downstream TGF‐β/Smad pathways by identifying Smad3‐dependent pathogenic non‐coding RNAs specific for tissue fibrosis (Lan et al. 2003). A number of Smad3‐dependent miRNAs have been identified, such as let‐7, antifibrotic miR‐29, as well as profibrotic miR‐21 and miR‐433. Down‐regulation of miR‐21 or overexpression of miR‐29 may represent alternative therapeutic strategies for CKD (Qin et al. 2011; Zhong et al. 2013). However, we should be aware of the complexity of the regulatory and working mechanisms of miRNAs, as different genes can be regulated by a single miRNA and vice versa (Singh et al. 2011). It is also possible that miRNAs can regulate each other during the pathophysiological processes. Within a given cluster, miRNAs may show the same pattern of expression but some of the cluster members may give different expression patterns (Khella et al. 2012). Some miRNAs, like miR‐29b, are encoded from more than one genomic locus with distinct promoter contexts (Kriegel et al. 2012). The regulatory mechanisms for controlling the expression of miRNA from more than one genomic locus are still largely unknown. Nevertheless, the functional roles of some miRNAs in tissue fibrosis are still controversial due to the contradictory findings obtained from different studies. For instance, the role of miR‐192 is still undefined in fibrosis. It is reported that miR‐192 is elevated in mouse tissue fibrosis models in vivo and in vitro (Kato et al. 2007; Chung et al. 2010a; Putta et al. 2012). Knockout or knockdown of miR‐192 largely attenuated renal fibrosis possibly through induction of ZEB1/2 (Krupa et al. 2010). However, a new study indicated that TGF‐β1 reduces miR‐192 expression in human tubular epithelial cells and deficiency of miR‐192 accelerates renal fibrosis in diabetic nephropathy (Wang et al. 2010).

Fortunately, lncRNA is highly disease‐ and tissue‐specific and may represent an ideal therapeutic target for kidney disease. lncRNAs were previously suggested as a ceRNA or a molecular sponge in modulating miRNA expression and biological functions. Compared to protein‐coding transcriptomes, lncRNAs are far more specific to organs, tissues, cell types, developmental stages and disease conditions, making them promising candidates as diagnostic and prognostic biomarkers (Nguyen & Carninci, 2016). Our recent studies provided encouraging findings that targeting renal lncRNAs (e.g. Erbb4‐IR) resulted in improved renal function and inhibited renal fibrosis in the mouse kidney fibrosis model with UUO using a non‐invasive ultrasound microbubble‐mediated technique (Feng et al. 2018). Nevertheless, silencing of renal Erbb4‐IR also improved renal function and inhibited renal fibrosis in db/db mice with type 2 diabetes phenotypes (Sun et al. 2017b). These findings supported the functional importance of lncRNAs in the pathogenesis of renal fibrosis and suggested targeting renal lncRNAs may represent a new precision strategy for renal fibrosis. Meanwhile, we should note that there is still a lack of an effective method to identify lncRNA homologues between different species; the translational studies of lncRNAs identified from animal models are still in difficulty. Evolutionary conservation is currently a challenge for lncRNA research (Chen et al. 2016). There will be a huge breakthrough when the animal‐based experimental findings can be smoothly transferred into a pre‐clinical setting in the future.

Conclusion

The role of TGF‐β/Smad signalling in renal fibrosis is inextricably complex, because it can induce pathogenesis via both direct and indirect mechanisms from proteomic crosstalk to transcriptional regulation. Increasing evidence demonstrated the interaction of lncRNAs between proteins, genomes and miRNAs under diseased conditions. The progress of TGF‐β signalling research has been rapidly accelerated during the last decade, where discovery of pathogenic lncRNAs opens up a new area for developing effective gene‐based therapies for kidney diseases. Thus, identification of renal lncRNAs with therapeutic potential will be one of the major goals for us in the coming decade.

Additional information

Competing interests

None declared.

Author contributions

P.M.‐K.T. and Y.‐Y.Z. contributed equally to preparing the first draft and revision of this manuscript. T.S.‐K.M., P.C.‐T.T. and X.‐R.H. proofread the manuscript. H.‐Y.L. supervised the design and arrangement of the manuscript. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This study was supported by grants from the Lui Che Woo Institute of Innovative Medicine (CARE program), the Research Grants Council of Hong Kong (GRF 14117815, 14121816, 14163317, C7018‐16G, TRS T12‐402/13N) and the Health and Medical Research Fund (03140486, 14152321).

Biographies

Patrick Ming‐Kuen Tang is a research assistant professor working with Ying‐Ying Zhang on studies of Smad3‐dependent lncRNAs in the development of diabetic kidney injury.

Ying‐Ying Zhang is a nephrologist from the Tongji University School of Medicine in Shanghai. She is doing her PhD study under the mentorship of Hui‐Yao Lan.

Hui‐Yao Lan is a Choh‐Ming Li Professor of Biomedical Sciences jointly at the Department of Medicine and Therapeutics, and the Li Ka Shing Institute of Health Sciences at the Chinese University of Hong Kong. He has published more than 320 peer‐reviewed papers with h‐index over 86 and has specialized in the pathogenic activities of TGF‐β/Smad signalling in tissue fibrosis, inflammation and cancer microenvironments. The authors are a research group from the Chinese University of Hong Kong which intensively focuses on the TGF‐β1/Smads signalling‐mediated pathogenesis in kidney physiology. They are also the first group to investigate the biological roles of Smad3‐dependent lncRNAs in the development of kidney diseases including renal inflammation, kidney fibrosis and diabetic kidney injury.