Regulation of Endothelial-to-Mesenchymal Transition by microRNAs in Chronic Allograft Dysfunction

Abstract

Fibrosis is a universal finding in chronic allograft dysfunction and is characterized by an accumulation of extracellular matrix. The precise source of the myofibroblasts responsible for matrix deposition is not understood and pharmacological strategies for prevention or treatment of fibrosis remain limited. One source of myofibroblasts in fibrosis is endothelial-to-mesenchymal transition, a process first described in heart development and involving endothelial cells undergoing a phenotypic change to become more like mesenchymal cells. Recently, lineage tracing of endothelial cells in mouse models allowed studies of EndMT in vivo and reported 27-35% of myofibroblasts involved in cardiac fibrosis and 16% of isolated fibroblasts in bleomycin-induced pulmonary fibrosis to be of endothelial origin.

Over the last decade, miRNAs have increasingly been described as key regulators of biological processes through repression or degradation of targeted mRNA. The stability and abundance of miRNAs in body fluids makes them attractive as potential biomarkers and progress is being made in developing miRNA targeted therapeutics.

In this review we will discuss evidence of miRNA regulation of EndMT from in vitro and in vivo studies and the potential relevance of this to heart, lung and kidney allograft dysfunction.

Introduction

Organ transplantation is the gold standard and in many cases the only treatment for patients with end-stage organ failure.¹ A limitation to successful transplantation is chronic allograft dysfunction where one of the processes leading failure of the transplanted organ is fibrosis.^2–4 Increasing graft survival remains a challenge; over the last two decades there have been substantial improvements in short-term outcomes after transplantation, but a reduction in late graft failure has been harder to achieve.^1,4,5

Although fibrosis is common, fundamental questions about the mechanisms of fibrosis remain, including the origins of matrix producing myofibroblasts. EndMT is the process by which endothelial cells undergo a phenotypic change to become more like mesenchymal cells. The transition may be partial and reversible but in undergoing EndMT endothelial cells loose markers such as CD31 and VE-cadherin and gain mesenchymal markers including αSMA. Cell adhesion properties are lost and morphology changes from a closely abutting cobblestone pattern to elongated cells with increased potential for migration, therefore resembling myofibroblasts.^6–9

MiRNAs have been shown to be regulators of many cellular processes¹⁰ and the purpose of this review is to examine the current evidence on their regulation of endothelial cell phenotype in fibrosis of heart, lung and kidney. Better understanding of this process could lead to new therapies to reduce fibrosis in allografts and therefore increase graft survival.

Origin of Myofibroblasts

Myofibroblasts have been shown to be the key effector cells in fibrosis and whilst it is agreed they have multiple sources, their exact origins remain unclear and may vary with tissue type.^{6,7,9,11–19} The current candidates are shown in Figure 1.

Figure 1.

Allograft injury results in release of cytokines, chemokines and increase in inflammatory and immune cells in the allograft which induce differentiation of different cells, including tissue-resident fibroblasts, bone marrow progenitor cells, endothelial cells, epithelial cells and pericytes, into myofibroblasts. Myofibroblasts produce collagen and collagen deposition results in fibrosis associated with chronic allograft failure.

Using endothelial cell lineage tracing and double-labelling techniques in mouse models of cardiac and renal fibrosis, approximately 30% of all fibroblasts were found to be of endothelial origin.²⁰ Conversely, lineage tracing in an ischaemic model of cardiac fibrosis found <1% of myofibroblasts to be of endothelial origin.²¹ This discrepancy may relate to the models of fibrosis and markers used.

Studies in pulmonary fibrosis models have provided further evidence of EndMT; co-localisation of αSMA with an endothelial marker has been demonstrated^22,23 and lineage tracing in bleomycin-induced pulmonary fibrosis identified 16% of isolated fibroblasts to be of endothelial origin.²⁴

Staining on human allografts undergoing rejection have also provided evidence for the process of EndMT in these tissues.^25,26 Compared to healthy controls, kidneys from patients diagnosed with chronic allograft dysfunction show increased expression of mesenchymal markers and a decreased expression of endothelial markers. Furthermore, double-staining found colocalisation of endothelial and mesenchymal markers in the chronic allograft dysfunction group providing further support for the role of EndMT in renal chronic allograft dysfunction.²⁷

Similarly, in endothelial cells from human coronary arteries colocalisation of CD31 with the mesenchymal marker Notch3 was significantly greater in chronically rejecting heart allografts (evident in 80% of cases) when compared to normal human coronary arteries.²⁸

Mechanism of EndMT in Fibrosis

The cell signalling mechanisms that control EndMT are covered in detail elsewhere.^7,9 In brief, TGFβ signalling is the most clearly defined but several other pathways have been shown to be involved, a summary of which is provided in Figure 2.²⁹

Figure 2.

Summary of main signalling pathways involved in changing gene expression in EndMT and how microRNAs that are upregulated (orange) or downregulated (purple) interact with these pathways. Red arrows indicate inhibition. Green arrows indicate activation.

Downstream of TGFβ signalling the Smad proteins are key effectors of changes in gene expression along with other transcriptional regulators such as Snail, Twist and Slug.⁷ Other pathways downstream of TGFβ include those acting through Ras, ROCK and RhoA-GTPases.³⁰ ET-1 and inflammatory cytokines IL-1β, TNF-α and INF-ϒ have also been implicated in driving EndMT signalling, whereas FGF signalling and PAI-1 inhibit the TGFβ pathway. In the case of FGF signalling the interaction with TGFβ is a reciprocal inhibitory one via FGFR1 and its intracellular subunit FRS2α.^9,28,31–35

Mitogen activated protein kinases and PI3 kinase are also modulated during EndMT although their roles are less well defined. There are inconsistencies between the studies on PI3 kinase and the impact of phosphorylated Akt levels on gene expression is not fully understood.^7,9,36–39

Notch signalling, endoplasmic reticular stress, reactive oxygen species and hypoxia have also been found to be involved in EndMT.^40–42 In hypoxia, HIF-1α is activated and in turn affects gene expression either directly or by interacting with the TGFβ pathway at the level of the transcription factor Snail1.⁴³

How miRNAs Work

MiRNAs are short single-stranded non-coding RNA sequences around 22 nucleotides in length that regulate protein expression at the post-transcriptional level.^10,44 Given >60% of protein coding genes are thought to be under miRNA control,¹⁰ these RNA sequences likely play a role in regulating almost every cellular process and there is increasing evidence for their involvement in fibrosis and EndMT.^7,45

Mature, single-stranded miRNAs are produced by processing of their double-stranded precursors (Figure 3). The miRNA gene is transcribed in the nucleus to produce a double stranded hairpin-shaped pri-miRNA. Pri-miRNA is then processed by endonucleases to form the shorter pre-miRNA which is exported into the cytoplasm. Here, further endonucleases cleave the structure to leave a double-stranded segment that RISC separates into the mature miRNA strand and its opposing arm.^46–48

Figure 3.

In the nucleus, the miRNA gene is transcribed by RNA polymerase II into the double stranded pri-miRNA. The enzyme Drosha RNase III endonuclease works with cofactor DiGeorge syndrome critical region 8 (DGCR8) to cleave both strands of pri-miRNA near the base of the primary stem loop to produce a shorter 60-70 nucleotide precursor known as pre-miRNA. Pre-miRNA is then actively exported into the cytoplasm via Exportin5 channel. Here, a second RNase III endonuclease, Dicer, cuts both strands of pre-miRNA near the base of the stem loop to leave a duplex of the mature miRNA and a similar sized complementary fragment of the opposing arm (miRNA*). The duplex is separated by incorporation of miRNA strand into RISC by binding to argonaute proteins.¹⁰⁷

Nucleotides 2-8 of the miRNA are referred to as the seed region and bind the target sequence on mRNA through Watson-Crick pairing to direct the miRNA-induced silencing complex. Predominantly, it is the three prime untranslated region of target mRNA to which the seed region binds but less frequently target sequences have also been identified in the open reading frame and the five prime untranslated region of mRNA.^47,49

Evidence of miRNA Involvement in EndMT in Heart, Lung, and Kidney Fibrosis

In Vitro Studies

EndMT has been demonstrated in vitro on exposure of endothelial cells to fibrotic stimuli such as TGFβ^36,50,51 high glucose concentrations,^52,53 radiation⁵⁴ or hypoxia.^43,55 MiRNA arrays have been used to screen for differentially expressed miRNAs in the presence or absence of fibrotic stimuli.^36,52,55 A summary of miRNAs whose levels have been found to be differentially expressed in EndMT is shown in Table 1.

Table 1. miRNAs Identified to be Upregulated or Downregulated in EndMT in Different Cells and Tissues.

MiRNAs	Upregulated	Downregulated
let-7	Mouse cardiac endothelial cells36	Mouse transplanted aorta28 Mouse diabetic kidney66 HUVEC28 HUAEC28 HMEC35
18a-5p		HAVEC57 Mouse diabetic heart57
20a		HUVEC63
21	Irradiated mouse lung54 Human pulmonary endothelial cells54 HUVEC42,56 Mouse cardiac endothelial cells36 Mouse transverse aortic constriction operated hearts56
29		HUVEC56 HMVEC35,67 Mouse diabetic kidney35,67
30b	Mouse cardiac endothelial cells36
122a		Mouse cardiac endothelial cells36
125b	Mouse cardiac endothelial cells36
126	Chronic hypoxic rat lung39 RPMEC39	EPC37
127		Mouse cardiac endothelial cells36
130a	Lung microvascular endothelial cells64
133a		Mouse diabetic heart59
155	MEEC,55 HCAEC51
195	Mouse cardiac endothelial cells36
196		Mouse cardiac endothelial cells36
200a		HAEC30
200b		Mouse diabetic heart 53 Primary mouse heart endothelial cells53
216a	HUVEC56
320		HUVEC52
375		Mouse cardiac endothelial cells36
483		HUVEC58

A role in activating EndMT has been suggested for certain miRNAs as blocking their activity is effective in preventing phenotypic changes. MiR-21 is expressed at higher levels in human cardiac fibroblasts than in endothelial cells and is upregulated in response to TGFβ stimulation in HUVEC.^42,56 MiR-21 blockade partly rescues endothelial marker expression that was suppressed by TGFβ in these cells and prevents changes in gene expression associated with EndMT.⁵⁶ Furthermore, increasing miR-21 levels in endothelial cells directly stimulated EndMT in the absence of TGFβ, suggesting that miR-21 alone is sufficient to drive EndMT. In contrast, in HPECs, although miR-21 was upregulated, it was found to be dispensable in EndMT occurring in response to radiation as its inhibition did not affect this process.⁵⁴ In human lung fibroblasts miR-21 expression increased and a miR-21 antagomir inhibited expression of collagen type IIIA supporting involvement in fibrosis although not necessarily in EndMT.⁵⁴

Upregulating other miRNAs can prevent EndMT indicating that they are important to maintain normal endothelial phenotype. Inhibition or downregulation of these miRNAs is therefore permissive for EndMT. For instance, in high glucose conditions, miR-18a-5p,⁵⁷ miR-200b,⁵³ and miR-320⁵² are downregulated and their upregulation appears to protect against EndMT and maintain normal endothelial function.

Serum from patients with fibrotic disease has been shown to modulate miRNA levels. Incubating HUVEC with serum from patients with Kawasaki disease was associated with lower expression of miR-483 when compared with cells incubated with healthy sera. Overexpression of miR-483 attenuated the EndMT response to disease sera.⁵⁸

Certain miRNAs may play a role in negative feedback within the signalling pathway for EndMT. For example, although miR-155 is upregulated under fibrotic conditions, overexpression with pre-miR-155 blocks the ability of TGFβ to induce EndMT in mouse embryonic endothelial cells. In support of an inhibitory role, antagomirs to miR-155 were able to induce αSMA expression in the absence of TGFβ.⁵⁵ In the case of miR-200a, which has been shown to oppose EndMT, overexpression can increase endothelial marker expression above baseline in the absence of TGFβ.³⁰

Preclinical Models

As with in vitro studies, miRNA array data on animal tissue has also demonstrated differential expression of certain miRNAs between organs that have undergone fibrosis in response to transplantation, radiation, chronic hypoxia or diabetes and control uninjured tissues. Although this does not necessarily indicate involvement in the process of EndMT, it can help to focus the direction of further studies.^35,39,54 Table 2 shows changes in miRNA expression identified in different models of fibrosis, the vast majority of which have also been implicated in EndMT in vitro.

Table 2. Differentially Expressed miRNAs in EndMT in Different In Vivo Animal Models of Fibrosis.

Fibrosis Model		MiRNAs
Transplantation	Mouse	Down: Let-728 (aorta)
Transverse aortic constriction	Mouse	Up: miR-2156 (heart)
Diabetes	Mouse	Down: miR-200b53 (heart), miR-18a-5p57 (heart), miR-133a59, let-735,66 (kidney), miR-2935,67 (kidney),
	Rat	Down: miR-32052 (kidney)
Radiation	Mouse	Up: miR-2154 (lung)
Monocrotaline-induced pulmonary hypertension	Rat	Up: miR-130a64 (lung)
Chronic hypoxia-induced pulmonary hypertension	Rat	Up: miR-126a-5p39 (lung)

Upregulated Up. Downregulated Down. Tissue studied in brackets.

Down, downregulated; Up, upregulated.

Involvement of miRNA in the process of EndMT in vivo has been suggested through correlation between expression of specific miRNAs, extent of EndMT and amount of fibrosis in the studied tissues. For instance, the CD1 strain of mice experience more renal fibrosis in response to diabetes than 129Sv mice. The mechanism of fibrosis in CD1 mice is at least partially driven by EndMT as demonstrated by increased colocalisation of endothelial and mesenchymal markers in diabetic kidneys compared to controls. This pattern was seen to a much lesser extent in 129Sv mice. Interestingly, in CD1 mice but not in 129Sv mice, diabetes was associated with suppression of let-7 and miR-29 in the kidney.³⁵

Consistent with the finding that miR-21 promotes EndMT in vitro,⁵⁶ miR-21 levels were found to be increased compared to controls in lung tissue from a mouse model of radiation-induced pulmonary fibrosis. In situ hybridisation showed this increase to be specific to radiation-damaged areas of lung⁵⁴ and EndMT is reported in radiation-induced lung injury.²³ This raises the possibility that miR-21 is involved in EndMT in the lung, but further research is required to determine whether this link exists.

The impact of specific miRNAs on the process of EndMT has been investigated in vivo through manipulation of their levels. An inhibitory effect of let-7 on EndMT was supported by the increased expression of mesenchymal markers in endothelial cells, associated with administering let-7 antagomirs in adult mice.²⁸ This role was further examined using a transplant model of aorta allograft in transgenic mice that allow fate mapping of endothelial cells by labelling with green fluorescent protein. Tissue was examined 2 weeks after transplantation, when rejection was occurring. In mice given let-7 antagomirs the proportion of luminal endothelial cells undergoing EndMT had increased from 61% to 80-90%. Conversely, let-7 mimic reduced collagen deposition and the proportion of endothelial cells undergoing EndMT to 33.7%. Together these results support a role for miRNA in modulating EndMT.²⁸

miR-133a is another inhibitor of EndMT in vivo as transgenic mice that upregulate miR-133a specifically in heart were shown to be partially protected from EndMT seen in diabetic cardiac fibrosis.⁵⁹ Similarly, overexpression of miR-200b in endothelial cells reduced cardiac fibrosis in diabetic mice and prevented EndMT. Interestingly, the downregulation of miR-200b seen in diabetic hearts compared to controls occurred only in endothelial cells, not cardiomyocytes, further supporting involvement in the process of EndMT.⁵³

Consistent with findings in lung that miR-21 levels increase in pulmonary fibrosis,⁵⁴ Kumarswamy et al.⁵⁶ saw the same trend in hearts that had undergone transverse aortic constriction. Furthermore, administration of miR-21 inhibitors was shown to reduce the proportion of cells undergoing EndMT supporting a function of miR-21 in activating EndMT.⁵⁶

Changes in expression observed in vitro do not consistently translate to in vivo models (Table 1). Although TGFβ treatment was associated with downregulation of miR-29b and upregulation of miR-216a in HUVEC, these changes were not seen in endothelial cells sorted from hearts of transverse aortic constriction operated mice when compared to sham-operated controls. There was, however, evidence of EndMT occurring in these hearts.⁵⁶

Human Studies

Human studies on the involvement of miRNA in EndMT during fibrosis of heart, lung and kidney remain sparse. EndMT has been shown to occur in heart and kidney chronic allograft dysfunction, as discussed earlier, but these studies did not investigate changes in miRNA.^28,34 Others have found differential expression of miRNAs in human fibrotic disease but not identified a direct link with EndMT.^39,58

MiRNA levels have been shown to correlate with disease states. A strong correlation was shown between miR-126a-5p serum levels and severity of disease in neonates with pulmonary hypertension.³⁹ Likewise, miR-21^60,61 has been demonstrated to increase with decreasing GFR in renal transplantation whereas miR-200b falls.⁶¹ In lung transplantation, miR-155 was upregulated and miR-195 downregulated in serum during acute rejection and these changes resolved with treatment.⁶² In Kawasaki disease there was an inverse correlation between serum levels of miR-483 and severity of disease. Interestingly, it seems that miR-483 comes from endothelial cells as there was also an inverse correlation between CD31-positive microparticles containing miR-483 and degree of coronary artery damage.⁵⁸

Certain miRNAs, previously shown to have a role in EndMT through in vitro or animal studies, have also been implicated in conditions relevant to transplantation in humans so it could be hypothesized that these miRNA changes affect endothelial function. The changes reported in these miRNAs is summarised in Table 3. There are some inconsistencies in the direction in which a particular miRNA is modulated between human and experimental studies and also between organs and disease states. Promisingly, for the two most studied miRNAs in this field, miR-21 and miR-155, results were far more consistent across all groups with miR-21 being upregulated and miR-155 downregulated in transplant pathology. Perhaps further study will identify additional miRNAs with a consistent response that could be developed as useful targets in transplantation.

Table 3. Evidence in Humans of miRNA Regulation in Disease States Relevant to Transplantation.

MiRNA	Ischaemia Reperfusion Injury	Acute Rejection	Chronic Rejection	Drug Toxicity	Infection
↓let-7		Tx kidney: ↓let 7b-5p84 ↓let 7b85 ↑let-7i84,85 ↓let-7c86		native kidney: ↑let-7i-3p ↓let-7f-5p ↓let7e-5p87	Tx kidney: ↑let-7a,7b,7c,7e 84
↓18a-5p	↑Tx kidney85
↓20a	↑Tx kidney85 ↑ native heart88			native kidney: ↓20a-5p87
↑21	↑native kidney89,90 ↑Tx kidney85 ↑native heart91	↑Tx heart:65,92 21-5p93 ↑Tx kidney65,85,86	↑Tx lung94 ↑Tx kidney60,61	↑native kidney87	↑native heart95
↓29	native kidney:↓29a-3p96				native heart:↓29c95
↑30b		↓Tx kidney86		↓native kidney87	↓native heart95 Tx kidney:↑30b-5p84
↓122a	native heart:↓12297
↑125b		↓Tx kidney84,85		↑125b-1-3p native kidney87
↑126	native kidney:↓126-3p96	↓ Tx kidney84,86
↓127	native kidney:↓127-3p96
↑130a	↑native heart88	↓Tx kidney84			Tx kidney:↑130a-3p84
↓133a	↑native heart97,98				↓native heart95,99
↑155	↓native kidney90	↑Tx heart65,92,100 ↑Tx lung62 ↑Tx kidney65,85,86,101,102	↑Tx kidney60,61 ↑ Tx lung62		↑native heart95,99,103 ↑native lung104
↑195	↑native heart88	↓Tx lung62	↓Tx lung62,105
↓200a		↓Tx kidney86
↓200b			↓Tx kidney60,61		Tx kidney:↑200b-3p84
↓320		↓Tx kidney106
↓483		↓Tx kidney106		native kidney: ↑483-3p87	native heart:↓483-3p95
↓196 ↑216a ↓375

miRNAs that have been shown to be upregulated (↑) or downregulated (↓) in EndMT through in vitro and in vivo models are listed in the first column. Evidence of these miRNAs being involved in different disease processes in human native and transplanted (Tx) kidney, heart and lung are indicated in subsequent columns. Where a specific miRNA within a family has been studied this has been listed.

How MiRNAs Regulate EndMT in Fibrosis

Modulating expression of specific miRNAs can impact on EndMT and has been used to investigate their function. Current knowledge on the ways in which miRNAs interact with the pathways regulating EndMT in fibrosis are summarized in Figure 2 with more detail provided in Table 4. Various mechanisms of action have been implicated and miRNAs may be effective at modulating EndMT only within a precise time window. For instance, miR-20a mimics reduced the number of cells that entered EndMT in vitro but had no effect on cells that had already started the program.⁶³

Table 4. Actions of miRNAs on Pathways Implicated in Regulation of EndMT.

MiRNA	Factors Modulating miRNA Levels	Actions of miRNA on Pathway
FGF and TGFβ
let-728,35	Upregulated by FGF signaling which is inhibited by inflammatory cytokines in fibrosis	Direct inhibition of TGFβR1 expression28 upregulates FGFR1 and miR-2935
miR-20a63	Upregulated by FGF signaling	Inhibits TGFβ signaling: directly targets TGFβR1, TGFβR2 and SARA
miR-2935	Upregulated by FGFR1 signaling	Directly inhibits IFN-ϒ expression Upregulates FGFR1 and let-7
miR-133a59		Inhibits expression of ET-1 which acts synergistically with TGFβ. Prevented TGFβ-induced increased in pSmad2 and pERK1/2 expression.
miR-14550	TGFβ1 signaling reduces processing of pre-miR-145 into mature form by upregulating MALAT1 which inhibits DICER expression	Directly inhibits MALAT1 expression and TGFβR2 and Smad3 expression to inhibit TGFβ signaling
Ras/RhoA
miR-200a30	Downregulated by TGFβ signaling	Directly inhibits growth factor receptor bound protein 2 (GRB2) expression. GRB2 is downstream of TGFβ and activates Ras to GTP-bound form.
NFκB
miR-130a64	Upregulated by NFκB downstream of TGFβ	Promotes NFκB expression so part of a positive feedback loop
PI3 kinase
miR-2156		Directly inhibits PTEN, an inhibitor of PI3K, to increase PI3K activity downstream of TGFβ. Downregulated CD31 and endothelial nitric oxide synthase and upregulated matrix metalloproteinases 2 and 9.
miR-126a-5p39		Prevents fall in proportion of phosphorylated Akt seen in hypoxia. Exact target is not known.
miR-12637	Downregulated by TGFβ1 signaling	Prevents TGFβ-induced fall in proportion of phosphorylated Akt by directly inhibiting expression of PI3KR2, a PI3K inhibitor.
Dipeptidyl peptidase-4
miR-2967		Directly inhibits DPP-4 expression
Acetylation
miR-133a59		Prevents upregulation of the acetylating transcription coactivator P300
miR-200b53		Prevents upregulation of the acetylating transcription coactivator P300
Angiotensin
Let-766	Levels restored by increasing AcSDKP directly or through administration of ACE inhibitor
miR-133a59		Prevents increase in angiotensinogen mRNA expression
Notch
miR-18a-5p57		Directly inhibits Notch2 expression
KLF/CTGF
miR-18358	Upregulated by KLF which is downregulated in inflammation	Directly inhibits CTGF (connective tissue growth factor) expression

Pathways indicated by subtitles. MiRNAs that are downregulated in EndMT or fibrosis are listed in italics, those that are upregulated are in normal script. The second column lists factors and pathways that have been shown to influence miRNA expression. miRNAs affecting FGF and TGFβ signaling pathways have been grouped together as a reciprocal inhibitory interaction exists between these two pathways which is at least partly mediated by interaction with miRNAs.

MiRNAs can be the effectors of both positive and negative feedback loops in the pathways that lead to EndMT. An example of a positive feedback loop is found in the interaction between miR-130a and NFκB. miR-130a is upregulated in a NFκB dependent fashion in both fibrotic mouse lungs and cells treated with TGFβ1 in vitro. MiR-130a mimics were found to increase NFκB-linked luciferase activity whereas antagomirs decreased it indicating a function in promoting NFκB activity. As miRNAs primarily inhibit gene expression it is possible that there are other steps downstream of miR-130a leading to promotion of NFκB activity.⁶⁴

For other miRNAs there is more conflicting data on their modes of action. miR-155 appears to function as a negative regulator of EndMT as it is upregulated by fibrotic stimuli but inhibits EndMT when overexpressed and prevents the activation of RhoA downstream of TGFβ when studied in mouse embryonic endothelial cells.⁵⁵ However, this may be a tissue specific role as it is not in keeping with findings in human and mouse heart transplantation where, although miR-155 was similarly upregulated with fibrosis, graft survival was increased in miR-155 knockout mice so supporting a function of miR-155 to promote fibrosis.⁶⁵ In vitro work on human coronary artery endothelial cells found miR-155 to promote EndMT as the addition of its inhibitor hindered TGF-β induced changes in fibrotic markers.⁵¹

miRNAs most likely affect protein expression in many, if not all, of the signalling pathways involved in EndMT. In general, the results of in vitro and in vivo studies are consistent with the predicted function of miRNAs based on in silico analysis and would suggest that miRNAs are important in maintaining normal endothelial phenotype.^37,39,56

Clinical Relevance

Some studies have demonstrated that existing treatments for diseases associated with fibrosis influence miRNA expression and so inhibit EndMT.^58,66,67 Specifically, atorvastatin was shown to act on the KLF4-miR-483-CTGF axis in Kawasaki’s Disease⁵⁸ and the downregulation of miR-29 and let-7 observed in the diabetic mouse kidney was prevented by linagliptin⁶⁷ and imidapril⁶⁶ respectively. However, these observations do not extend into the realms of allograft rejection.

MiRNA levels can be manipulated in vivo^28,56 to effectively alter a disease process. Activity can be increased with synthetic miRNA mimics⁶⁸ or vectors carrying a pre-miRNA sequence.⁶⁹ miRNA blockade is achieved with antagomirs (chemically modified antisense oligonucleotide sequences to a particular miRNA), miRNA sponge⁷⁰ to competitively bind miRNAs or miRNA mask⁷¹ to competitively bind the target sequence. Promising progress is being made in the area of hepatitis C where a miR-122 inhibitor (Miravirsen) effectively reduced hepatitis C virus RNA levels beyond the duration of treatment in a phase II clinical trial, without developing viral resistance.⁷² Drugs targeting miRNAs to treat cancer have also reached clinical development.⁷³

Aside from therapeutics, understanding changes in miRNA levels could contribute to clinical care through diagnosis and monitoring of disease. At this stage, only a handful of papers discuss the potential for using miRNAs as biomarkers in chronic allograft dysfunction^45,74–80 but there is more work on acute allograft rejection. The stable nature of miRNAs make them an attractive target to investigate as biomarkers. So far this quality has facilitated their study in various samples including tissue, urine, serum and plasma.⁸⁰

It is worth considering also the implications of losing endothelial cell phenotype through EndMT, rather than purely the gain of mesenchymal features as contributing to the disease process. A logical consequence of endothelial cells losing their phenotype in this way would be a reduction in the density of capillaries. Particularly in the kidney, capillary rarefaction is one of the main features of progressive fibrotic disease^81,82 in both native and transplanted organs. In targeting EndMT it may therefore be possible to reduce both gain of fibrotic tissue and loss of capillaries in order to improve organ function.

Concluding Remarks and Future Perspectives

The predominant preclinical model used for fibrosis is diabetes (Table 2) and more work is required at the preclinical level to specifically look at the process of chronic allograft dysfunction. Although levels of some miRNAs have been shown to be altered in serum,⁵⁸ other studies have suggested an autocrine or paracrine action.⁵³ If local mechanisms of action do predominate, a more invasive approach to human studies would be required using tissue samples rather than serum to detect changes.

The study of miRNA function is complicated by their expression across multiple cell lines and the variability in their modulation between cell types. The challenge of identifying their precise mechanism of action is further added to by the multitude of target mRNAs for a single miRNA. Combined, these factors contribute to the apparent slow progress in our understanding of the role of miRNA in human transplantation.^47,49,83

Given the dearth of research into miRNA regulation of fibrosis in human chronic allograft dysfunction a key next step would be to investigate whether findings from animal models reliably translate into humans. As summarized in Tables 1, 2 and 3 there is some variability in miRNA expression changes between in vitro and in vivo models depending on cell types used; particularly in the case of the let-7 which is perhaps the most widely studied miRNA in this field. It is therefore essential to correlate findings from preclinical models to those in human disease and so far this has produced promising results for miR-21 and miR-155. Ultimately the goal should be for a better understanding of miRNA in fibrosis to enable early detection and treatment of chronic allograft dysfunction so improving transplant survival and patient quality of life.

Abbreviations

αSMA

α smooth muscle actin

CD31

platelet endothelial cell adhesion molecule

CTGF

connective tissue growth factor

EndMT

endothelial-to-mesenchymal transition

EPC

bone marrow derived endothelial progenitor cells

ER stress

endoplasmic reticular stress

ERK

extracellular signal regulated kinase

ET-1

endothelin-1

FGF

fibroblast growth factor

FGFR

fibroblast growth factor receptor

GFR

glomerular filtration rate

GTP

guanosine triphosphate

HAEC

human aortic endothelial cells

HAVEC

human aortic valve endothelial cells

HCAEC

human coronary artery endothelial cells

HIF-1α

hypoxia inducible factor-1α

HMVEC

human dermal microvascular endothelial cells

HUAEC

human umbilical artery endothelial cells

HUVEC

human umbilical vein endothelial cells

IL-1β

interleukin-1β

INF-ϒ

interferon-ϒ

KLF

krüppel-like factor

MEEC

mouse embryonic endothelial cells

miRNA

mature microRNA

mRNA

messenger RNA

NECD

notch extracellular domain

NICD

notch intracellular domain

NOX

NADPH oxidases

PAI-1

plasminogen activator inhibitor-1

PI3K

phosphoinositide 3 kinase

Pri-miRNA

primary transcript of microRNA

HPEC

human pulmonary endothelial cells

RISC

RNA-induced silencing complex

ROCK

rho-associated protein kinase

ROS

reactive oxygen species

SARA

Smad anchor for receptor activation

TGFβ

transforming growth factor β

TGFβR

transforming growth factor β receptor

TNF-α

tumour necrosis factor-α