What can we learn from Fli1-deficient mice, new animal models of systemic sclerosis?

Abstract

Systemic sclerosis is a complex multifactorial disease characterized by autoimmunity, vasculopathy, and selective organ fibrosis. A series of genetic and epidemiological studies have demonstrated that environmental influences play a central role in the onset of systemic sclerosis, while genetic factors determine the susceptibility to and the severity of this disease. Therefore, the identification of predisposing factors related to environmental influences would provide us with an informative clue to better understand the pathological process of this disease. Based on this concept, the deficiency of transcription factor Friend leukemia virus integration 1, which is epigenetically suppressed in systemic sclerosis, seems to be a potential candidate acting as the predisposing factor of this disease. Indeed, Fli1-mutated mice serve as a set of useful disease models to disclose the complex pathology of systemic sclerosis. This article overviews the recent advancement in systemic sclerosis animal models associated with Friend leukemia virus integration 1 deficiency.

Introduction

Systemic sclerosis (SSc) is a multisystem autoimmune disease characterized by initial vascular injury and resultant fibrosis of the skin and various internal organs. Although the pathogenesis of SSc still remains largely elusive, it is generally accepted that this disease is caused by a complex interaction between genetic factors and environmental influences. The critical contribution of genetic factors is quite obvious because the highest risk factor for SSc development is family history (1). However, a twin study revealed that concordance for SSc is low in the twins (4.7%) and similar in monozygotic and dizygotic twins (4.2% vs 5.6%), while concordance for the presence of autoantibodies against nuclear and cytoplasmic antigens is significantly higher in the healthy monozygotic twin sibling than in the healthy dizygotic twin sibling of an SSc patient (95% vs 60%, p < 0.05) (2). These two studies suggest that genetic factors are linked to the induction of autoimmunity but not enough for the development of clinically definite SSc. In line with this notion, most of susceptibility genes for SSc are human leukocyte antigen (HLA) haplotypes and non-HLA genes related to immunity and inflammation which are shared by other collagen diseases, such as rheumatoid arthritis and systemic lupus erythematosus (3). In addition to these epidemiological data, several independent case–control and genome-wide association studies exhibit the association between single nucleotide polymorphisms of certain disease-susceptibility genes and disease severity in SSc patients. Taken together, it is speculated that a certain set of environmental factors triggers SSc onset in individuals genetically predisposed to autoimmune diseases, and that genetic factors determine disease severity once patients develop SSc (Fig. 1).

Fig. 1.

The role of genetic factors and environmental influences in the development of SSc. Accumulation of predisposing factors, such as genetic factors and environmental influences, alters the phenotypes of various kinds of cells, including immune cells, vascular cells, and fibroblasts. Environmental factors seem to determine the onset of SSc, while genetic factors affect the susceptibility to and the severity of this disease.

Environmental factors, including silica dust, solvents, chemicals, drugs, and infectious agents, potentially affect the behavior of various cell types through the modification of gene expression profiles by directly acting on cell signaling pathways and/or by epigenetic mechanisms, such as histone modifications and DNA methylation. DNA methylation occurs at CpG islands under the control of a number of DNA methyltransferases. Methylated DNA is recognized by a family of proteins with a highly conserved methyl-CpG DNA binding domain. Proteins with this domain regulate the recruitment of chromatin-modifying enzymes, including histone deacetylases. Thus, DNA methylation results in histone deacetylation leading to the transcriptional repression of target genes through chromatin condensation. In SSc, DNA hypermethylation of the FLI1 and KLF5 genes in dermal fibroblasts (4, 5) and the BMPRII gene in microvascular endothelial cells (6) and DNA hypomethylation of the ACTA2 gene in lung fibroblasts (7) and the CD40L and CD70 genes in CD4⁺ T cells (8, 9) have been reported (3). Also, hypoacetylation of histones H3 and H4 in the FLI1 and KLF5 promoters is exhibited in SSc dermal fibroblasts (4, 5). Consistent with these data, the expression levels of DNA methylation and histone acetylation–related enzymes, such as DNA methyltransferase 1, methyl-CpG DNA binding proteins 1 and 2, and histone deacetylases 1 and 6, are significantly increased in SSc dermal fibroblasts compared with normal dermal fibroblasts (4). These genes with epigenetic modifications may be predisposing factors of SSc, possibly reflecting the influence of environmental factors.

Based on these backgrounds, it is postulated that animals with predisposing factors reflecting environmental influences may spontaneously develop SSc-like manifestations and serve as a new SSc disease model. Supporting this idea, Fli1-mutated mice in which Fli1 expression is haploinsufficient in all of the cells or deficient in certain cell types have provided us a variety of useful clues to better understand the pathological process of SSc. This article reviews the recent advance of Fli1-mutated animal models of SSc.

Why has Fli1 attracted much attention in SSc?

Fli1 is a member of the ETS transcription factor family, which was initially identified as a proto-oncogene in Friend virus–induced erythroleukemia in mice (10). Fli1 is expressed at high levels in endothelial cells and hematopoietic cells according to comparative gene expression profile analysis of various human cell lines (11) and also detectable in dermal fibroblasts and keratinocytes at relatively lower levels by immunohistochemistry with human skin samples (12-14). Consistently, Fli1 plays pivotal roles in the development and/or differentiation of megakaryocytes, myelomonocytes, erythrocytes, and natural killer (NK) cells (15, 16); in vascular development and angiogenesis (17, 18); in extracellular matrix (ECM) remodeling (12, 19-22); and in the phenotypical alteration of keratinocytes (14).

The involvement of Fli1 in ECM metabolism was initially investigated in the regulation of TNC gene expression (19). Then, Fli1 was shown to serve as a potent transcriptional repressor of the type I collagen genes (COL1A1 and COL1A2) in dermal fibroblasts (12, 21). The first study regarding the role of Fli1 in SSc was documented by Kubo et al. (13), demonstrating a remarkable decrease in Fli1 expression in dermal fibroblasts, dermal microvascular endothelial cells, and perivascular inflammatory cells in the involved and non-involved skin of SSc patients, particularly to a greater extent in the involved skin. After that, an excellent study by Wang et al. (4) regarding epigenetic modifications of the FLI1 gene was published in 2006, disclosing the hypermethylation of CpG islands and the hypoacetylation of histones H3 and H4 in the FLI1 promoter using bulk skin and/or cultured dermal fibroblasts of SSc patients. Therefore, Fli1 deficiency is a potential predisposing factor of SSc, possibly reflecting environmental influences.

The impact of Fli1 deficiency on dermal fibroblasts, dermal microvascular endothelial cells, macrophages, and CD4⁺ T cells

Fli1^+/− mice

To elucidate the contribution of Fli1 deficiency to the induction of SSc-like features, Fli1^+/− mice have been investigated. Fli1^+/− mice look macroscopically normal but show some abnormalities molecularly and microscopically. Dermal thickness of the back skin is comparable between wild type (WT) and Fli1^+/− mice, but the levels of soluble type I collagen protein and Col1a2 messenger RNA (mRNA) are much higher in Fli1^+/− mice than in WT mice (23). However, Ctgf mRNA levels are not altered in the skin of Fli1^+/− mice. Consistently, human dermal fibroblasts treated with FLI1 small interfering RNA (siRNA) oligonucleotide express COL1A2 mRNA at higher levels than human dermal fibroblasts treated with non-silencing scrambled RNA, while both cells produce Ctgf mRNA at comparable levels (5, 23). Thus, Fli1 haploinsufficiency promotes type I collagen production but not the expression of connective tissue growth factor (CTGF) in dermal fibroblasts, leading to the accumulation of type I collagen with no change in dermal thickness in vivo. With respect to dermal blood vessels, Fli1^+/− mice exhibit a mild distortion of arterioles and increased vascular permeability of small vessels (5) but not arteriolar stenosis, capillary dilation, and capillary loss, which are characteristically seen in SSc vasculopathy (24-26). Thus, Fli1 deficiency induces a part of SSc-like phenotypes at least in dermal fibroblasts and dermal microvascular endothelial cells, but Fli1^+/− mice do not show any clinical features of SSc. This notion supports the idea that Fli1 deficiency is a predisposing factor of SSc, and some additional factors are required for the induction of SSc-specific disease process.

Bleomycin-treated Fli1^+/− mice

It is further investigated whether Fli1 haploinsufficiency exacerbates three cardinal features of SSc in mice treated with bleomycin (BLM), an established animal model of SSc.

Dermal fibrosis and the activation of dermal fibroblasts

The 4-week BLM challenge increases dermal thickness; collagen content; mRNA expression of the Col1a1, Col1a2, and Ctgf genes; and the number of α-smooth muscle actin (SMA)-positive myofibroblasts in Fli1^+/− mice to a greater extent than in WT mice (23). An important feature of SSc dermal fibroblasts is the constitutive activation of transforming growth factor (TGF)-β signaling even in the situation when TGF-β expression is not altered. This property is at least partially attributable to the upregulated expression of integrins αVβ3 and αVβ5, which recruit latent TGF-β and subsequently promote the release of active TGF-β on cell surface of SSc dermal fibroblasts. Of note, this important feature of SSc dermal fibroblasts is reproduced in dermal fibroblasts of BLM-treated Fli1^+/− mice. Furthermore, human dermal fibroblasts transfected with FLI1 siRNA oligonucleotide activate latent TGF-β on cell surface through the upregulated expression of these integrins, resulting in the induction of CTGF expression (23). These results collectively indicate that Fli1 haploinsufficiency augments the BLM-induced SSc-like phenotype in dermal fibroblasts.

Vascular activation and predominant infiltration of Th2/Th17 cells

Vascular activation contributes to the development of tissue fibrosis through the induction of pro-fibrotic inflammation and endothelial-to-mesenchymal transition (EndoMT) in SSc (27). The correlation of Th1/Th2/Th17 balance with tissue fibrosis has been well studied in diffuse cutaneous systemic sclerosis (dcSSc). In the early and sclerotic phases of dcSSc, Th2/Th17 immune response is predominant, while immune polarization shifts from Th2/Th17 to Th1 in parallel with the resolution of skin sclerosis in dcSSc (28). This is plausible because Th2 cytokines, such as interleukin (IL)-4 and IL-13, exert a pro-fibrotic effect on dermal fibroblasts (29, 30), and IL-17A is required for the development of skin fibrosis in BLM-treated mice (31, 32), while interferon (IFN)-γ, a Th1 cytokine, reduces type I collagen expression (33, 34). The infiltration of Th1/Th2/Th17 cells in the lesional skin of BLM-treated mice is tightly regulated by cell adhesion molecules. For instance, intercellular adhesion molecule-1 (ICAM-1) and glycosylation-dependent cell adhesion molecule-1 (GlyCAM-1) promote the infiltration of Th2/Th17 cells, while P-selectin and E-selectin facilitate the infiltration of Th1 cells (35). Of note, the relative fold-induction of ICAM-1 and GlyCAM-1 to E-selectin and P-selectin in response to BLM is elevated in Fli1^+/− mice compared with WT mice. Furthermore, gene silencing of FLI1 results in the induction of ICAM-1 and GlyCAM-1 and the suppression of E-selectin and P-selectin in human dermal microvascular endothelial cells (23). Therefore, Fli1 deficiency directly causes the increased ratio of ICAM-1 and GlyCAM-1 to E-selectin and P-selectin in BLM-treated mice, leading to the predominant infiltration of Th2/Th17 cells in the skin. With regard to EndoMT, Fli1 haploinsufficiency increases the number of fibroblast-specific protein 1 (FSP1)/vascular endothelial (VE)-cadherin double positive cells in phosphate-buffered saline (PBS)- and BLM-treated mice. Importantly, Fli1 binds to the promoter of SNAIL1 gene, a master regulator of EndoMT, and gene silencing of FLI1 enhances the expression of SNAIL1 mRNA in human dermal microvascular endothelial cells, indicating that Fli1 deficiency directly induces EndoMT (23). Collectively, Fli1 haploinsufficiency potentially enhances the BLM-dependent induction of a pro-fibrotic phenotype in dermal microvascular endothelial cells.

The polarization of macrophages

A clinical trial of tocilizumab, an anti-IL-6 receptor antibody, has revealed that this treatment resolves skin sclerosis along with the suppression of a cluster of M2 macrophage–associated genes in early dcSSc patients, suggesting a critical role of M2 macrophages in SSc skin fibrosis. Consistent with this notion, mRNA levels of M2 macrophage markers, Arg1, Fizz1, and Ym1, and the number of arginase 1-positive macrophages in the lesional skin are significantly elevated in BLM-treated Fli1^+/− mice compared with BLM-treated WT mice while comparable at baseline. Of note, IL-4 and IL-13 induce polarization of macrophages toward the M2 phenotype to a greater extent in Fli1^+/− peritoneal macrophages than in WT peritoneal macrophages, indicating that Fli1 haploinsufficiency directly contributes to the differentiation of M2 macrophages (23). Therefore, Fli1 haploinsufficiency also seems to serve as a predisposing factor to exemplify the BLM-dependent induction of an SSc-like phenotype in macrophages.

Endothelial cell–specific Fli1 knockout mice (Fli1 EcKO mice)

As described above, Fli1^+/− mice show mild vascular distortion and increased vascular leakage, and BLM-treated mice represent endothelial cell activation similar to SSc endothelial cells. However, both of these models do not display vascular structural abnormalities characteristic of SSc, such as arteriolar stenosis and capillary dilation (24-26). A possible reason underlying this issue is that haploinsufficient loss of the Fli1 gene in endothelial cells is not enough to induce SSc-like vascular changes. This notion is supported by the establishment of endothelial cell–specific Fli1 knockout mice (Fli1^flox/flox;Tie2-Cre mice [Fli1 EcKO mice]). This animal model exhibits arteriolar stenosis and capillary dilation as well as remarkable vascular leakage. Importantly, the expression of VE-cadherin, platelet endothelial cell adhesion molecule-1 (PECAM-1), platelet-derived growth factor subunit B (PDGF-B), and sphingosine-1-phosphate receptor 1 (S1P₁), all of which regulate endothelial cell–cell interaction and/or endothelial cell–pericyte/vascular smooth muscle cell interaction, is decreased, while matrix metalloproteinase-9, an enzyme degrading vascular basement membrane, is upregulated in endothelial cells of Fli1 EcKO mice. Since endothelial Fli1 expression is much more decreased in Fli1 EcKO mice than in Fli1^+/− mice, endothelial Fli1 deficiency is enough to induce vascular structural changes characteristic of SSc. Therefore, Fli1 deficiency may be directly related to the development of SSc vasculopathy (17).

The identification of another potential predisposing factor of SSc, KLF5

As described so far, Fli1 deficiency induces SSc-like properties in various cell types under a physiological condition or in response to some triggers, and mice with endothelial Fli1 deficiency partially mimic the structural abnormalities of SSc vasculopathy. However, Fli1 deficiency alone is not sufficient to spontaneously develop SSc clinical symptoms in mice. Under this situation, the best way to confirm the contribution of Fli1 deficiency to the development of SSc is to investigate whether the addition of other predisposing factors to Fli1 deficiency promotes the spontaneous development of SSc-like clinical features in mice. Based on this idea, we extensively explored previous literature reporting the critical transcription factor in terms of tissue fibrosis regulation, and KLF5 caught our attention as a potential candidate. KLF5 is a member of the SP/KLF transcription factor family, which has been shown to contribute to the pathological tissue fibrosis of the heart and kidney in mice (36, 37). Furthermore, DNA microarray analysis with bulk skin revealed a significant downregulation of KLF5 gene in SSc lesional skin compared with healthy control skin. Of note, our study demonstrated the following findings: (a) KLF5 expression is decreased in dermal fibroblasts of SSc lesional skin at least partially by an epigenetic mechanism, (b) BLM injection induces extensive dermal fibrosis in Klf5^+/− mice, (c) KLF5 serves as a potent repressor of the CTGF gene in dermal fibroblasts, and KLF5 deficiency is enough to enhance CTGF expression, (d) Fli1 deficiency alone does not affect the transcription of the CTGF promoter but simultaneous downregulation of KLF5 and Fli1 synergistically stimulates the CTGF promoter activity, and (e) KLF5 and Fli1 make transcription factor complex. Therefore, KLF5 deficiency seems to be a potential predisposing factor of SSc which is related to environmental influences.

A new SSc animal model, Klf5^+/−;Fli1^+/− mice

Based on the data described above, we hypothesized that simultaneous downregulations of KLF5 and Fli1 are likely to resemble the hallmark of SSc dermal fibroblasts, namely, the increased expression of type I collagen and CTGF. Therefore, we generated mice with double heterozygous loss for Klf5 and Fli1 genes and investigated whether this animal model mimics clinical features of SSc (Fig. 2).

Fig. 2.

The rationale for the generation of a new SSc animal model, Klf5^+/−;Fli1^+/− mice. In the skin, Klf5^+/− mice and Fli1^+/− mice highly express connective tissue growth factor (CTGF) and type I collagen, respectively. Since the hallmark of SSc dermal fibroblasts is the upregulated expression of CTGF and type I collagen, Klf5^+/−;Fli1^+/− mice were generated to mimic skin fibrosis of SSc.

Klf5^+/−;Fli1^+/− mice mimic dermal and pulmonary fibrosis characteristic of SSc

Klf5^+/−;Fli1^+/− mice spontaneously develop remarkable dermal fibrosis at the age of 3 months. Dermal thickness, the amount of collagen content, and mRNA levels of the Col1a1, Col1a2, and Ctgf genes are markedly increased in Klf5^+/−;Fli1^+/− mice compared with WT littermates. Furthermore, when observed with electron microscopy, dermal collagen fibrils are irregular and thick in longitudinal sections and highly variable in diameter in transverse sections, which are reminiscent of collagen structure in SSc skin (38). Importantly, the skin of Klf5^+/−;Fli1^+/− mice and SSc patients shares similar expression profiles of fibrillogenesis-associated genes, including decreased decorin expression, increased levels of lumican and ADAMTS-2, and comparable levels of fibromodulin, biglycan, bone morphogenetic protein-1, and lysyl oxidase. Therefore, it is likely that KLF5 and Fli1 make transcription factor complex and coordinately regulate the expression of fibrosis- and fibrillogenesis-related molecules, leading to the induction of an SSc-like skin phenotype at the ultrastructural level. In addition, Klf5^+/−;Fli1^+/− mice exhibit non-specific interstitial pneumonia, the main histological pattern of interstitial lung disease (ILD) associated with SSc characterized by diffuse uniform expansion of alveolar septa with patchy septal lymphocytic infiltration, which becomes evident at the age of 8 months.

Klf5^+/−;Fli1^+/− mice develop vascular changes similar to SSc

In addition to SSc-like organ fibrosis, Klf5^+/−;Fli1^+/− mice exhibit stenosis of arterioles and bushy capillaries as early as the age of 1 month and a progressive decrease in the number of subcutaneous vascular density from 2 months of age. Reduced arteriolar blood flow velocity and interstitial hypoxia are also evident in the skin of Klf5^+/−;Fli1^+/− mice. Furthermore, pulmonary arterioles develop markedly thickened vascular walls with expansion of α-SMA-positive cells at 8 months of age, a typical histologic feature of pulmonary arterial hypertension, and interseptal venules show intimal fibrosis, focal luminal narrowing, and perivascular lymphocytic infiltrates, all suggestive of pulmonary veno-occlusive disease. Thus, the haploinsufficiency of Klf5 and Fli1 genes is enough to reproduce vascular alterations characteristic of SSc, such as destructive vasculopathy (capillary loss) and proliferative obliterative vasculopathy (stenosis of arterioles with the proliferation of vascular smooth muscle cells) (5).

Klf5^+/−;Fli1^+/− mice represent aberrant immune activation

Klf5^+/−;Fli1^+/− mice mimic B-cell activation of SSc, which is a part of immune abnormality, characteristically seen in this disease (39). CD19 is a positive regulator of B cells, and the increased expression of CD19 is associated with the aberrant activation of B cells which is characterized by the production of autoantibody and excessive IL-6 (40). Of note, splenic Klf5^+/−;Fli1^+/− B cells exhibit the increase in cell surface CD19 expression and IL-6 secretion compared with WT B cells, and serum levels of IL-6 and antinuclear antibodies are elevated in Klf5^+/−;Fli1^+/− mice compared with WT mice. In line with evidence that ILD associated with SSc frequently accompanies diffuse or aggregated B-cell infiltrates (41), Klf5^+/−;Fli1^+/− lungs demonstrate perivascular B-cell accumulation starting from the age of 3 months, which progresses to prominent B-cell lymphoid aggregates and diffuse interstitial infiltrates at the age of 8 months. Notably, both KLF5 and Fli1 interact with the Cd19 promoter in murine B cells, suggesting that transcription factor complex intrinsically regulate CD19 expression in B cells (5).

Significance of Klf5^+/−;Fli1^+/− mice in SSc research

In addition to tissue fibrosis of the skin and lung, Klf5^+/−;Fli1^+/− mice display SSc-like cardiac and intestinal involvement (unpublished data). Of note, Klf5^+/−;Fli1^+/− mice reproduce SSc-specific disease cascade, starting with immune abnormalities, leading to vasculopathy, and finally resulting in tissue fibrosis (Fig. 3). Considering the high reproducibility of the SSc-like organ involvement and the SSc-specific disease cascade in Klf5^+/−;Fli1^+/− mice, the establishment of this animal model supports the canonical idea that environmental factors determine SSc onset. On the other hand, it is also confirmed that genetic factors affect the severity of SSc by a series of experiments with Irf5^–/– mice and human clinical samples (42).

Fig. 3.

Clinical and pathological features of Klf5^+/−;Fli1^+/− mice. Klf5^+/−;Fli1^+/− mice spontaneously develop the three cardinal pathological features of SSc, namely, starting with immune abnormalities, followed by vasculopathy, and finally resulting in tissue fibrosis.

The role of Fli1 deficiency in keratinocytes

Altered phenotype of SSc keratinocytes

Since SSc is characterized by autoimmunity, vasculopathy, and tissue fibrosis, previous studies focused on the role of immune cells, vascular cells, and fibroblasts in the development of this disease. However, recent studies have demonstrated the altered phenotype of SSc keratinocytes. For example, SSc keratinocytes persistently express wound-associated cytokeratins, keratin 6 (K6) and keratin 16 (K16), not only in the sclerotic skin but also in the non-lesional skin (43). SSc keratinocytes stimulate fibroblasts in cell culture with excessively secreted IL-1α (43), which is a major alarmin released from the epithelial cells triggering an inflammatory response in fibroblasts (44). Increased expression of the key pro-fibrotic growth factor, CTGF, is also evident in SSc epidermis (45, 46). In addition, epithelial-to-mesenchymal transition (EMT), a central mechanism in fibrosis development driven by TGF-β1 (47), is enhanced in SSc epidermis with the increased expression of its cardinal regulator SNAIL1 (48-50). Of particular interest is the evidence that specific keratin expression signatures correlate with the presence of ILD in the analysis of global gene profiling with SSc lesional skin (51). Therefore, it seems that the epithelial phenotype is not only related to dermal fibrosis but rather profoundly associated with SSc development itself.

Fli1 deficiency induces SSc-like gene expression profiles in keratinocytes

Fli1 expression is decreased in keratinocytes as well as in dermal fibroblasts, dermal microvascular endothelial cells, and inflammatory cells in the lesional skin of SSc patients, especially diffuse cutaneous SSc patients. In the epidermal sheets, FLI1 mRNA is significantly downregulated in dcSSc patients compared with healthy controls, while the increase in K6, K16, IL1A, CTGF, and SNAIL1 mRNAs is evident in dcSSc patients. Furthermore, FLI1 siRNA-treated normal human keratinocytes express K6 and K16, as well as IL-1α, CTGF, and SNAIL1 to a greater extent than scrambled non-silencing RNA-treated control cells. Moreover, K6 and K16 are highly expressed in keratinocytes of Fli1^+/− mice compared with those of WT mice in vivo. Thus, Fli1 deficiency plays a critical role in the induction of SSc-like properties in keratinocytes.

Epithelial cell–specific Fli1 knockout mice reproduce selective organ fibrosis and autoimmunity similar to those of SSc

A recent study with epithelial cell–specific Fli1 knockout mice (Fli1^flox/flox, K14-Cre mice [Fli1 KcKO mice]) has disclosed an unknown role of epithelial cells in the development of SSc. Since K14 is exclusively expressed by dermal and esophageal stratified squamous epithelia and medullary thymic epithelial cells (mTECs), Fli1 gene is selectively deleted from those cells in this animal model. In Fli1 KcKO mice, the upregulated expression of K6, K16, IL-1α, CTGF, and SNAIL1 is strictly reproduced in keratinocytes. In addition, IL-1α expression is increased in stratified squamous epithelia of esophagus. Correspondingly, Fli1 KcKO mice spontaneously develop tissue fibrosis in the skin and esophagus. Another important clinical feature of Fli1 KcKO mice is the development of ILD similar to that of SSc. This is induced by the impairment of the negative selection of autoreactive T cells to lung antigen(s) and the generation of regulatory T cells in the thymus, which is partially mediated by the downregulation of autoimmune regulator (AIRE), a molecule regulating the expression of tissue-restricted antigens in mTECs (52). Importantly, transfer of NK1.1–CD3⁺ T cells from Fli1 KcKO mice to Rag1^–/– mice reproduces ILD, while Fli1 KcKO;Rag1^–/– mice spontaneously develop dermal and esophageal fibrosis but not ILD. Therefore, Fli1 deficiency in stratified squamous epithelia and mTECs is sufficient to induce dermal and esophageal fibrosis and autoimmune ILD, respectively. Overall, these results suggest that the phenotypical alteration of epithelial cells contributes to selective organ fibrosis and autoimmunity in SSc (Fig. 4).

Fig. 4.

A new hypothesis explaining selective organ fibrosis and autoimmunity in SSc. Fli1 deficiency induces the phenotypical alteration of epithelial cells. Fli1-deficient stratified squamous epithelia secrete pro-fibrotic factors and undergo partial epithelial-to-mesenchymal transition (EMT), contributing to the development of dermal and esophageal fibrosis. Fli1-deficient medullary thymic epithelial cells result in the impairment of the negative selection of autoreactive T cells to lung antigen(s) and the generation of regulatory T cells in the thymus, leading to autoimmunity and interstitial lung disease.

Fli1 EcKO mice provide us with a clue to elucidate the molecular mechanism by which bosentan exerts a disease-modifying effect on SSc vasculopathy

As demonstrated by two high-quality randomized clinical trials, bosentan prevents the development of digital ulcers in SSc patients. Now, bosentan is broadly used as the first-line treatment for the prevention of digital ulcers associated with SSc, but the detailed molecular mechanism underlying this clinical effect still remains unknown. However, the studies with Fli1 EcKO mice provide us with a clue to elucidate a potential mechanism explaining the clinical efficacy of bosentan for SSc digital ulcers. As already described, Fli1 EcKO mice mimic vascular fragility of SSc, which can be evaluated by the leakage of Evans blue dye from the vessels. Of note, 2-week administration of bosentan improves vascular fragility in parallel with the increased expression of molecules regulating endothelial cell–cell interaction and endothelial cell–pericyte/vascular smooth muscle cell interaction, such as VE-cadherin, PECAM-1, PDGF-B, and S1P₁. Also, endothelial Fli1 expression of Fli1 EcKO mice, which is decreased by 50%–80% compared with WT mice, is increased after the administration of bosentan. In dermal microvascular endothelial cells, endothelin-1 induces the phosphorylation of Fli1 at threonine 312 through the sequential activation of c-Abl and protein kinase C-δ (PKC-δ), which results in the loss of DNA binding ability and the proteasomal degradation of Fli1, and bosentan increases the expression and the DNA binding ability of Fli1 by blocking this pathway. Since endothelial cells maintain the homeostatic condition by autocrine and paracrine effects of endothelin-1, bosentan can increase the expression of endothelial Fli1 in Fli1 EcKO mice. Consistent with this notion, endothelial Fli1 expression is increased in SSc patients treated with bosentan compared with those untreated with bosentan (53). Importantly, bosentan exerts a reverse remodeling effect on SSc vasculopathy (54, 55). Therefore, bosentan improves SSc vasculopathy at least in part via reversing the expression of endothelial Fli1.

Conclusion

In the pathogenesis of SSc, environmental influences have been believed to play a central role in the onset of SSc, while genetic factors determine the susceptibility to and the severity of SSc. This canonical idea is strongly supported by evidence that all the three cardinal pathological features of SSc are recapitulated in mice by simultaneous downregulation of Klf5 and Fli1 genes, both of which are epigenetically suppressed in SSc dermal fibroblasts. Fli1 EcKO mice and Fli1 KcKO mice resemble vascular structural abnormalities and selective organ fibrosis with autoimmunity characteristic of SSc, respectively (Tab. I). The results of these studies strongly suggest that Fli1 deficiency is a potential predisposing factor of SSc, and further studies on Fli1 deficiency would help us better understand the complicated pathological process of this disease.

Table I.

Summary of phenotypical features of Fli1-mutated mice

	Immune abnormalities and inflammation	Vasculopathy	Fibrosis
Fli1+/− mice	–	Mild vascular distortion	Elevated type I collagen deposition in the skin
BLM-treated Fli1+/− mice	Elevation of Th2/Th17 cytokines and increased infiltration of mast cells and M2 macrophages	Increased vascular disintegrity	Skin fibrosis (other organs are not investigated)
Klf5+/−;Fli1+/− mice	B-cell activation, autoantibody production, and IL-6 elevation	Capillary loss, capillary dilation, and arteriolar stenosis	Skin fibrosis and interstitial lung disease (other organs are not investigated)
Fli1 EcKO mice	–	Capillary dilation and arteriolar stenosis	–
Fli1 KcKO mice	Elevation of Th2/Th17 cytokines, B-cell activation, autoantibody production, IL-6 elevation, and increased infiltration of mast cells and macrophages	–	Dermal and esophageal fibrosis and interstitial lung disease

BLM = bleomycin; Th = T helper; EcKO = endothelial cell–specific knockout; KcKO = K14-expressing cell-specific knockout.