Targeting miR-155 to Treat Experimental Scleroderma

Scleroderma is a refractory autoimmune skin fibrotic disorder. Alterations of microRNAs in lesional skin could be a new approach to treating the disease. Here, we found that expression of miR-155 was up regulated in lesional skin tissue from patients with either systemic or localized scleroderma, and correlated with fibrosis area. Then we demonstrated the potential of miR-155 as a therapeutic target in pre-clinical scleroderma models. MiR-155−/− mice were resistant to bleomycin induced skin fibrosis. Moreover, topical antagomiR-155 could effectively treat mice primed with subcutaneous bleomycin. In primary skin fibroblast, miR-155 silencing could inhibit collagen synthesis function, as well as signaling intensity of two pro-fibrotic pathways, Wnt/β-catenin and Akt, simultaneously. We further showed that miR-155 could regulate the two pathways via directly targeting casein kinase 1α (CK1α) and Src homology 2-containing inositol phosphatase-1 (SHIP-1), as previous reports. Mice with miR-155 knockout or topical antagomir-155 treatment showed inhibited Wnt/β-catenin and Akt signaling in skin upon bleomycin challenge. Together, our data suggest the potential of miR-155 silencing as a promising treatment for dermal fibrosis, especially in topical applications.

Topical antagomiR-155 effectively treated bleomycin induced skin fibrosis. We first demonstrated that transcutaneous absorption of topical antagomiR-155 worked in C57/BL6 mice, as shown in the trace of conjugated Cy3 ( Supplementary Fig. S2). Then we treated the mice with antagomiR-155 two weeks after preliminary bleomycin induction (The procedure is depicted in Fig. 3a). After two-week treatment, miR-155 expression in skin, but not in liver, bone marrow or blood cells, could be effectively down regulated by topical antagomiR-155 (Fig. 3b, Figure S3). No overt side effects were observed. Topical antagomiR-155 significantly decreased the dermal thickening (Fig. 3b,c), collagen deposition (Fig. 3d), as well as density of α -SMA+ activated fibroblasts in bleomycin challenged skin tissue (Fig. 3e).
MiR-155 silencing inhibited collagen production in primary skin fibroblasts. We isolated primary mouse skin fibroblasts and transfected them with miR-155 mimic or inhibitor, which successfully regulated miR-155 expression 24 hours after transfection ( Supplementary Fig. S4). The mRNA expressions of type I collagen were elevated with miR-155 mimic and decreased with miR-155 inhibitor, and similar changes were observed on α -SMA (Fig. 4a). Further, miR-155 inhibitor could also remarkably decreased collagen released to the supernatant (Fig. 4b).
MiR-155 regulated Wnt/β-catenin and Akt signaling in vitro. Several major signaling pathways have been found to promote fibrosis and SSc development 2 . We screened these pathways using western blot analysis in primary skin fibroblast challenged with TGF-β . Among these pathways, we noticed that β -catenin and Akt signaling intensity could be regulated by miR-155. MiR-155 mimic could strongly decrease the degradation of β -catenin and increase the phosphorylation of Akt, while miR-155 inhibitor did the opposite to the two pathways (Fig. 5).
MiR-155 regulated Wnt/β-catenin and Akt signaling by directly targeting CK1α and SHIP-1, respectively. We conducted a bio-informatics search and identified casein kinase 1α (CK1α ), a negative modulatory protein on β -catenin pathway, as a predicted target of miR-155 in human and mouse. Meanwhile we identified a negative regulator on Akt signal pathway, Src homology 2-containing inositol phosphatase-1 (SHIP-1), as another target of miR-155 (Fig. 6a). Then we developed a luciferase reporter construct consisting of CK1α 3′ -UTR miR-155 binding region (LucCK1), and used a mutated construct (LucCK1mu) and vehicle plasmid (Luc) as controls (mutated sequences are depicted in Fig. 5a). MiR-155 mimic significantly decreased luciferase activity in HEK293cells transfected with LucCK1 reporter, compared with cells that transfected with control vectors (Fig. 6b), which suggested that CK1α is a direct target of miR-155. The direct interaction between miR-155 and 3′ -UTR of mouse SHIP-1 mRNA was similarly demonstrated (Fig. 6c). Further Western blot analysis showed that miR-155 silencing could increase both CK1α and SHIP-1 protein levels, with inhibited β -catenin degradation and Akt phosphorylation simultaneously (Fig. 6d,e). Both systemic and topical miR-155 targeting regulated Wnt/β-catenin and Akt signaling in vivo. Protein level of β -catenin and pAkt both showed decreasing trend in miR-155 −/− mouse skin tissue, compared with WT mice (Supplementary Fig. S5). Topical antagomiR-155 application lead inhibited staining of β -catenin and phosphorylated Akt in multiple cell sets from skin tissue, including but not limited to fibroblasts (Fig. 7a). Average optic density (AOD) value of dermal layer from treatment group was significantly less than control (Fig. 7b). These in vivo findings were consistent with what we observed in the in vitro study above.

Discussion
SSc is a heterogeneous disease whose pathogenesis is characterized by three hallmarks: excessive deposition of extracellular matrix, small vessel vasculopathy and production of autoantibodies 2 . Though the clinical manifestations of SSc vary, most of the patients have skin thickening and variable involvement of internal organs. Compared with vasculopathy or autoimmunity, fibrosis of skin and other organs still lacks approved treatments. Tyrosine kinase and TGF-β inhibitors have shown potential anti-fibrotic effects in experimental research recently; however none of them succeed in randomized clinical trials 16 . Here we describe that miRNA could be a novel potential treatment for fibrosis.
Morphea is also a disorder characterized with excessive collagen deposition in dermis or subcutaneous tissue 1 . However, unlike SSc, it lacks vasculopathy features such as sclerodactyly, Raynaud phenomenon, nailfold capillary changes, and telangiectasias 17 . In our study, miR-155 was upregulated both in SSc and morphea skin samples. This indicated that the work of miR-155 in scleroderma might not depend on vascular injury. On the other side, the tendency that patients with larger fibrosis area or inner organ sclerosis had higher miR-155 expression in their skin further supported possible relation of miR-155 with fibrosis.
Recent data indicated that the miR-155 expression is up-regulated in many inflammatory fibrosis syndromes other than scleroderma, including idiopathic pulmonary fibrosis 18,19 , cystic fibrosis 20,21 and alcoholic/nonalcoholic liver fibrosis [22][23][24] , as well as in animal models of these diseases 20,22,[24][25][26] . Besides, loss of miR-155 in mice can significantly inhibit pressure-overload 27 or diabetes 28 induced cardiac fibrosis and remodeling, suggesting the potential of miR-155 as a treatment target in fibrotic conditions.
In this study, targeting miR-155 could inhibit Wnt/β -catenin and Akt pathways, which are necessary for fibrosis development. Our data indicate miRNA as a novel approach to touch the two pathways simultaneously. As one of the most well accepted pro-fibrotic signaling pathways, Wnt/β -catenin pathway is demonstrated to be involved in SSc development and the experimental models [29][30][31] . Meanwhile, Akt is also activated in SSc fibroblasts, and blocking Akt by siRNA, small molecular inhibitor 32 or its upstream protein 33 can treat experimental skin fibrosis effectively. Moreover, the two pathways can crosstalk with canonical TGF-β signaling; they both can be activated by canonic pro-fibrotic cytokine TGF-β in fibroblast 34 and other cell types 35 .
CK1α is a serine/threonine kinase leading phosphorylation and degradation of multiple components of β -catenin pathway, and can be up-regulated by β -catenin as a negative feedback 36 . Previous study on human liposarcoma has demonstrated that miR-155 impacts β -catenin signaling through directly targeting CK1α 37 . To our knowledge, it is the first time to show the potential role of CK1α in scleroderma treatment. While another member of Casein kinase family, CK II, has also recently emerged as a possible therapeutic target for scleroderma 38 .
Similarly, SHIP-1 is a phosphatase that can abolish phosphorylation of Akt, which promote cell proliferation and survival 39,40 . SHIP-1 has also been proved as a direct target of miR-155 41 . In fact, regulation of SHIP-1 by miR-155 is critical to autoimmunity or inflammation in animal studies, such as arthritis 42 and lupus 43 . Present researches have also shed some light on SHIP-1 and fibrosis. SHIP-1 is essential for proliferation, survival, migration and collagen production of fibroblasts [44][45][46] ; and SHIP-1 deficiency attenuates airway fibrosis in allergy mouse model 47 .
Our study revealed a potentially novel treatment approach to target miRNA. To our knowledge, this study first reports that a cholesterol-conjugated antagomiR has succeeded to treat skin lesion epicutaneously. It is plausible to hypothesize that epicutaneous antagomiR-155 could be especially beneficial to patients with local scleroderma such as morphea. Further in vivo study on human skin tissue is warranted.   For topical treatment, B6 mice were first injected subcutaneously with bleomycin every other day for two weeks; from day 15, the mice were applied antagomiR-155 or scramble control (RiboBio Co., Ltd., Guangzhou, China) epicutaneously every other day for another two weeks with continuous bleomycin injection. The antagomiR-155 and scramble control were 3′ -cholesterol and 2′ -OMe modified and dissolved in 95% acetone at concentration of 0.67 nmol/ml. Each mouse was administrated 2.6 nmol of antagomiR-155 or scramble control on the lesional skin area each time. The mice were sacrificed on day 28.
Tissue Fibrosis assessment. For histologic assessment, mouse skin samples were fixed and stained with hematoxylin and eosin and and Sirius red (Chondrex Inc., WA, USA) staining according to the manufacturer's instructions. Thickness of dermis of each mouse was calculated as mean value of two distinct Sirius red staining sections, with five measurements at different positions in each section. Skin collagen content was measured using Sircol collagen dye-binding assay (Biocolor, Belfast, Northern Ireland) according to the instruction of manuscript, where each mouse was sampled by skin punch biopsies (6-mm in diameter) from the bleomycin injection site.
For skin fibroblast counting, α -SMA positive fibroblast (the staining is detailed in the next passage) was counted as the mean value of two distinct sections for each mouse. Each section included five random fields at 400 times magnification.
Similarly for immunofluorescence assay, the sections were secondly stained with Alex488 labeled goat-anti-mouse IgG or CY3 labeled goat-anti-rabbit IgG antibodies. Each fluorescent staining was recorded by the OlyVIA system (Olymphus, Southend-on-Sea, UK) under one condition. Average optic density (AOD) of each sample was calculated as the mean value of two distinct fields at 100 times magnification by using Image Pro Plus 6.0 (Media Cybernetics, Rochville, MD, USA). Skin layers only between muscle and epidermis were counted in AOD calculation.
Primary skin fibroblast isolation and culture. We digested fresh skin tissue from juvenile B6 mice (younger than one week, 6 mice for one time) with 0.1% dispase II overnight, removed departed epidermis, treated the dermis with 0.1% collagenase I, and filtered digested cell suspension with nylon membrane. Then the cells were cultured with Dulbecco's modified eagle medium (DMEM) supplemented with 10% fetal calf serum and passage 3 to 7 was used for experiments. Supernatant collagen concentration was detected by Sircol collagen dye-binding assay according to manuscript (Biocolor).
Quantitative PCR. Total RNA from cultured cell was isolated using the TRIzol reagent; RNA from paraffin sections was isolated using RNeasy PPFE kit (Qiagen, Hilden, Germany). The reverse transcription kit (Qiagen) was used for cDNA preparation. TaqMan miRNA assay (Applied Biosystems) was used for determination of the expression level of mouse miR-155 (MS00001701) and human miR-155 (MS00003605). The expression of U6B small nuclear RNA was used as endogenous control. SYBR Green Master Mix (Applied Biosystems) was used for determination of mRNA expression of mouse col1a1, col1a2, and α -SMA. The expression of GAPDH was used as endogenous control (primer sequences seen in Table S2). Each sample in the chain reaction was amplified in triplicate.
Western blot analysis. Primary mouse skin fibroblasts were lysed by RIPA solution and fresh tissue was lysed using T-PER reagent with proteinase and phosphatase inhibitor (Thermo Fisher). After gel electrophoresis and electrotransferation, proteins were detected with antibodies against SHIP-1, CK1α , pAkt, Akt (pan), β -catenin, pErk, pJNK, pSmad2/3, pp38 and GAPDH. HRP conjugated goat anti-rabbit or anti-rat secondary antibodies were used. Semiquantitative analysis based on densitometry was performed using Image J software (National Institute of Health, Bethesda, MD, USA).
Luciferase Activity Assay. The mouse CK1α miRNA target site and its mutation were amplified by primers; the target site was predicted by bioinformatics database including miRBase, PicTar and Target Scan Human. These PCR products were both cloned downstream of the luciferase gene in psiCHECK-2 luciferase vector (Promega, WI, USA), and the constructs were named "Luc-CK1α " and "Luc-CK1α (mu)". These constructs were transfected together with miR-155 mimic or scrambled miRNA into HEK293 cells. Luciferase activity was measured using the Dual-Luciferase Reporter Assay (Promega) 24 h after transfection. Each treatment was performed in triplicate.

Statistical analysis.
All continuous variables were expressed as means ± SD. Comparisons between two groups were tested for statistical significance with unpaired t test or Mann-Whitney U test, as appropriate. Comparison among three or more groups was performed with analysis of variance (ANOVA) followed by Bonferroni correction. Correlation between two groups of continuous variables was analyzed with linear regression. All statistical analysis was performed using SAS 11.0 (SAS Institute Inc., Cary, NC, USA).