Targeting FSTL1 for Multiple Fibrotic and Systemic Autoimmune Diseases

Abstract

Follistatin-like 1 (FSTL1) is a matricellular protein that is upregulated during development and disease, including idiopathic pulmonary fibrosis (IPF), keloid, and arthritis. The profibrotic and pro-inflammatory roles of FSTL1 have been intensively studied during the last several years, as well as in this report. We screened and identified epitope-specific monoclonal neutralizing antibodies (nAbs) to functionally block FSTL1. FSTL1 nAbs attenuated bleomycin-induced pulmonary and dermal fibrosis in vivo and transforming growth factor (TGF)-β1-induced dermal fibrosis ex vivo in human skin. In addition, FSTL1 nAbs significantly reduced existing lung fibrosis and skin fibrosis in experimental models. FSTL1 nAbs exerted their potent antifibrotic effects via reduced TGF-β1 responsiveness and subsequent myofibroblast activation and extracellular matrix production. We also observed that FSTL1 nAbs attenuated the severity of collagen-induced arthritis in mice, which was accompanied by reduced inflammatory responses in vitro. Our findings suggest that FSTL1 nAbs are a promising new therapeutic strategy for the treatment of multiple organ fibrosis and systemic autoimmune diseases.

Graphical Abstract

Li et al. demonstrate that blocking FSTL1 with neutralizing antibodies ameliorates pulmonary and skin fibrotic diseases and attenuates systemic autoimmune diseases such as rheumatoid arthritis. These findings reveal that FSTL1 neutralizing antibodies may be a potential multifunctional therapeutic approach for organ fibrosis and systemic autoimmune diseases.

Introduction

Follistatin-like 1 (FSTL1) is a small secreted glycoprotein that is primarily produced by cells of mesenchymal origin.¹ Based on the presence of a follistatin (FS) domain and an extracellular calcium-binding (EC) domain, FSTL1 belongs to the SPARC family of matricellular proteins, whose members participate in dynamic matrix-cell interactions and fine-tune cellular functions.² Although the functions and mechanisms of FSTL1 are not completely understood, a fast moving body of literature has described the involvement of FSTL1 in embryonic development, tissue remodeling/repair, and inflammatory processes and has identified the regulatory role of FSTL1 in cell growth, differentiation, apoptosis, and migration.^3,4 FSTL1 was first discovered as a transforming growth factor (TGF)-β1-induced protein⁵ and has also been shown to function largely via regulation of the TGF-β/BMP signaling pathway at the level of membrane receptors.^6,7 However, the activation of different signaling pathways, such as phosphorylated AKT (pAKT), phosphorylated AMP-activated protein kinase (pAMPK), phosphorylated extracellular signal-regulated kinase (pERK), and mitogen-activated protein kinase (MAPK), and of the inflammatory response via the Toll-like receptor 4 (TLR4)/CD14 pathway, has been linked to FSTL1 in various disease models.8, 9, 10 Therefore, FSTL1 is an important modulator of cell-matrix interactions and the integration of multiple proteins and pathways that are fundamental to many cell functions.

Organ fibrosis is defined as an unremitting deposition of extracellular matrix (ECM) components due to excessive fibroblast activity.¹¹ It is a final common pathological feature of many diseases and leads to end-stage organ failure with consequent high morbidity and mortality. Examples of these diseases are idiopathic pulmonary fibrosis (IPF), systemic sclerosis (SSc), hypertrophic scars and keloids, endomyocardial fibrosis, and liver cirrhosis.12, 13, 14, 15, 16 No effective medical therapies are currently available. TGF-β1 is probably the most critical profibrotic cytokine in the initiation and perpetuation of organ fibrosis due to its roles in regulating epithelial injury, myofibroblast differentiation, and collagen production.^17,18 However, few candidates that directly target the TGF-β pathway have reached even early phase clinical trials.^19,20 Alternative strategies are being investigated. Recent studies have shown that matricellular protein Fstl1 is upregulated in lungs from patients with IPF and in a bleomycin mouse model.⁷ Fstl1 is a downstream effector of TGF-β1-induced fibrotic responses, and it facilitates TGF-β1 signal transduction into cells.⁷ These findings suggest an important FSTL1/TGF-β1 signaling axis underlying the profibrotic responses in IPF pathogenesis and a potential therapeutic target of FSTL1 for the treatment of progressing pulmonary fibrosis.

Numerous reports have also suggested the involvement of FSTL1 in human systemic autoimmune diseases, particularly in the pathogenesis of rheumatoid arthritis (RA).²¹ Elevation of Fstl1 has also been reported in the synovium and/or serum from patients with RA and in a collagen-induced arthritis (CIA) mouse model. FSTL1 has been identified as a novel pro-inflammatory protein, and it enhanced the synthesis of proinflammatory cytokines and chemokines in vitro and in vivo.22, 23, 24 Therefore, a strategy to target FSTL1 may be ideal for human fibrotic and immune diseases.

In the present study, we screened and identified epitope-specific monoclonal neutralizing antibodies (nAbs) to functionally block FSTL1. We evaluated the effects of these nAbs on the development of fibrosis in vitro using primary human lung or skin fibroblasts, ex vivo using human skin explants, and in vivo using bleomycin-induced pulmonary and skin fibrosis. We also tested the effects of these nAbs on RA using an in vivo CIA mouse model. We provide evidence that increased FSTL1 expression is critical in the pathogenesis of fibrotic and systemic autoimmune diseases and that anti-FSTL1 nAbs may be a potential multifunctional therapeutic approach for patients with pulmonary or skin fibrosis and for patients with RA.

Results

Generation and Characterization of FSTL1 nAbs

Based on the profibrotic⁷ and pro-inflammatory⁴ effects of FSTL1, we hypothesized that blocking FSTL1 function with specific nAbs would alleviate fibrotic and inflammatory responses. To this end, we generated FSTL1 nAbs (22B6) via the immunization of mixed recombinant FSTL1 protein fractions⁷ and generated 38 anti-FSTL1 monoclonal antibodies (mAbs) targeting different epitopes of mouse FSTL1 protein using the Surface Epitope Antibody Library (SEAL) technique.²⁵ The initial selection of positive FSTL1 nAbs was based on the ability of mAbs to block FSTL1-facilitated TGF-β1 signaling using a (CAGA)₁₂-luciferase reporter in NIH 3T3 cells (Figure S1). Further selection of FSTL1 nAbs was based on the ability of mAbs to block FSTL1-facilitated TGF-β1-increased type I collagen (Col1a1/Col1a2) and fibronectin (Fn) mRNA expression in mouse embryonic fibroblasts (MEFs) (Figure S2). Because of the pleiotropic effects of the TGF signaling pathway on tissue homeostasis and because of the serious side effects of excessive inhibition of the TGF-β1 signaling pathway,^19,20 we selected antibodies with experimental values between a 100% (the solid line) and 50% blockage rate (the dotted line). A panel of 15 antibodies was selected based on these criteria (Table 1). The final determination for FSTL1 nAbs was based on their epitope position, titer value, and blockage ability. Three candidate antibodies (4D22, 2K6, and 3I2) with distinct epitopes, higher titer, and striking blocking effects were finalized (Table 1; Figure 1A). Unfortunately, the candidate hybridoma line (3I2) died in the follow-up culture.

Table 1.

Epitope Position, ELISA Titer, and Blockage Activity Information for 15 Selected Antibodies

Epitope Domain	Antibody	Epitope Position	ELISA Titer (K)	Luciferase Blockage Activity (%)	ECM Product Blockage Activity (%)
Linker (SP-FK)	1F10	19–28	100	13.04	15.63
Linker (FK-EC)	1F5	116–125	100	4.47	32.54
	1H7	127–136	100	10.34	34.89
	1K15	127–136	100	14.99	46.56
	3G16	127–136	128	2.50	42.29
	4D22	132–141	128	28.02	47.74
	4N3	132–141	100	10.13	25.28
EC domain	1M7	147–156	100	40.93	46.84
	1P10	147–156	100	30.55	43.96
	4K21	162–171	100	16.26	51.12
	2E20	165–174	100	9.66	26.99
	3I2	201–210	128	41.89	72.26
	4E1	201–210	100	3.82	67.18
	4N18	201–210	100	25.06	47.31
VWC domain	2K6	270–279	128	18.69	49.05

Domain structure of FSTL1 protein: SP, signal peptide; FK, follistatin-like and Kazal-like domain; EF, EF hands of the extracellular calcium-binding (EC) domain; and VWC, von Willebrand factor type C domain.

Figure 1.

Epitope Location and Specificity of FSTL1 nAbs

(A) Epitope location of FSTL1 nAbs in different domains of the FSTL1 molecule, and the framed antibodies are finalized candidate antibodies (4D22, 2K6, and 3I2). SP, signal peptide; EF, EF hands of the extracellular calcium-binding (EC) domain; VWC, von Willebrand factor type C domain. (B and C) Kinetic analyses of the interaction of 2K6 (B) and 4D22 (C) nAbs with mouse FSTL1 protein by MST. (D and E) Pull-down assays οf 2K6 (D) and 4D22 (E) nAbs with FSTL1 (top), TGF-β1 (middle), or follistatin (FS; bottom). The protein-nAbs complex was immunoprecipitated by protein G agarose and subjected to western blot analysis of FSTL1, TGF-β1, and follistatin. The left lane is the positive control by recombinant protein. (F) Pull-down assays to verify whether FSTL1 nAbs interfere with the binding of FSTL1 and TGF-β1. The His-tagged FSTL1 protein, TGF-β1 protein, and FSTL1 nAbs (or control isotype antibody IgG2b) complex was pulled down using Ni-NTA agarose and then subjected to western blot assays with anti-TGF-β1 antibody to confirm the presence of TGF-β1. The presence of FSTL1 was detected with anti-His antibody. (G) Pull-down assays to verify whether FSTL1 nAbs interfere with the binding of FSTL1 and type II receptor of TGF-β1 (TβRII). HEK293 cells were transfected with Myc-His-tagged Fstl1 and HA-tagged TβRII, then treated with FSTL1 nAbs or control isotype antibody IgG2b. Myc-His-tagged FSTL1 was pulled down using Ni-NTA agarose and then subjected to western blot analyses with anti-HA antibody to confirm the presence of TβRII. The presence of FSTL1 was detected with anti-Myc antibody. TCL, total cell lysate. (See also Figures S1–S5.)

The 2K6 and 4D22 nAbs were monitored for their abilities to bind to immobilized mouse FSTL1 protein using the microscale thermophoresis (MST) assay. The observed binding curves revealed a dissociation rate constant (K_d) of 106.8 nM for 2K6 and a K_d of 113.1 nM for 4D22 (Figures 1B and 1C); however, these two nAbs did not bind to FS, another FS-SPARC family member protein (Figures S3A and S3B). Similar results were obtained using the human FSTL1 protein (K_d of 321.2 nM for 2K6 and K_d of 65.6 nM for 4D22, Figures S3C and S3D). Pull-down assays showed that 2K6 and 4D22 specifically recognized the FSTL1 protein, but not FS or TGF-β1 (Figures 1D and 1E), and then blocked the interaction of FSTL1 with TGF-β1 (Figure 1F) and its type II receptor (TβRII) (Figure 1G). These results confirmed the high specific affinity of both nAbs to FSTL1 and suggested the potential mechanism underlying the downstream signaling of FSTL1.

Biological assays revealed that both 2K6 and 4D22 nAbs blocked the ability of FSTL1 to induce the production of interleukin (IL)-6 in COS-7 cells (Figure S4A),²² to attenuate hypertrophy to phenylephrine (PE) treatment in primary rat cardiac myocytes (Figure S4B),²⁶ and to antagonize BMP4-inhibited epithelial-mesenchymal transition (EMT) in A549 cells, as determined by the inhibition of BMP4-increased E-cadherin (E-cad) and promotion of BMP4-decreased N-cadherin (N-cad) (Figures S4C and S4D).⁶ These data suggest that 2K6 and 4D22 nAbs were biologically active in these in vitro assays. Taken together, we concluded that both 2K6 and 4D22 were FSTL1 nAbs with high specific affinity for, and strong ability to block, FSTL1 protein.

We further evaluated the toxicity of 2K6 and 4D22 nAbs in C57BL/6J mice. In the acute toxicity study, mice were intraperitoneally (i.p.) administered 2K6 and 4D22 nAbs in a single dose (1 mg/mouse), and toxicity was evaluated 24 h after injection. All mice tolerated the nAbs well, and the body weights did not change (Figure S5A). The appearance and internal morphology of the main organs were not damaged (Figures S5B and S5C). A 7-day chronic toxicity study of 2K6 and 4D22 nAbs was performed at 500 μg/mouse (dosed every other day), and no animal exhibited any systemic symptoms during the observation period. The body weight and organ morphology in 2K6 and 4D22 nAb-treated mice were not changed compared with those in the control group (Figures S5D–SDF). These data suggest that 2K6 and 4D22 nAbs are safe and applicable for further investigation in vivo.

Elevated Expression of FSTL1 in IPF and a Bleomycin Lung Fibrosis Model

We previously reported that Fstl1 expression was increased in the lungs and fibroblasts from patients with IPF and in a bleomycin mouse model.⁷ Because FSTL1 is a secreted glycoprotein, we collected serum samples from 19 IPF patients and 20 healthy controls and measured elevated levels of circulating FSTL1 in IPF serum samples (Figure 2A). Similar elevations of FSTL1 levels were detected in the peripheral blood from bleomycin-injured mice (Figure 2B). FSTL1 was upregulated in all samples from patients with IPF and in the bleomycin model, which indicates its clinical relevance to IPF.

Figure 2.

FSTL1 nAbs Block TGF-β1 Responsiveness in IPF Fibroblasts and Attenuate Bleomycin-Induced Lung Fibrosis In Vivo

(A) FSTL1 levels in serum of IPF patients (n = 19) and normal control individuals (n = 20) were determined by ELISA. (B) FSTL1 protein levels in serum of C57BL/6J mice (n = 3) at the different time-point after bleomycin injury was determined by western blot. β-tubulin was used as a loading control. Relative density of FSTL1 normalized β-tubulin is represented. (C-D) Primary lung fibroblasts from IPF patients (n = 2) were treated with nAbs or IgG2b 1 h before TGF-β1 treatment. (C) Western blot analysis of p-Smad2, Smad2, p-Smad3, and Smad3. Relative density of p-Smad2 normalized to Smad2 and p-Smad3 normalized to Smad3 are represented. (D) Western blot analysis of α-SMA, Col1. β-actin was used as a loading control. Relative density of α-SMA normalized to β-actin is represented. (E) Prevention dosing regimen of lung fibrosis model. C57BL/6J mice were pre-intraperitoneally injected with nAbs or IgG2b (50 μg/mouse/time point; n = 10) on days -3, -2, and -1 before bleomycin treatment, and lungs were harvested at day 14. (F) Hydroxyproline contents in lung tissues. (G) Masson trichrome staining of collagen on lung sections; bars, 100 μm. (H) Lung fibrotic score analysis of the lung sections. (I) Intervention dosing regimen at early stage of lung fibrosis model. C57BL/6J mice were intraperitoneally injected with nAbs or IgG2b (50 μg/mouse/time point; n = 10) on days 5, 8, and 11 after bleomycin treatment, and lungs were harvested at day 14. (J) Hydroxyproline contents in lung tissues. (K) Masson trichrome staining of collagen on lung sections; bars, 100 μm. (L) Lung fibrotic score analysis of the lung sections. (M) Intervention dosing regimen at late stage of lung fibrosis model. C57BL/6J mice were intraperitoneal injected with nAbs or IgG2b (50 μg/mouse/time point; n = 10) on days 8, 11, 14, and 17 after bleomycin treatment, and lungs were harvested at day 21. (N) Hydroxyproline contents in lung tissues. (O) Masson trichrome staining of collagen on lung sections; bars, 100 μm. (P) Lung fibrotic score analysis of the lung sections. Horizontal lines indicate mean. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001. (See also Figure S6.)

FSTL1 nAbs Attenuated TGF-β1 Responsiveness in IPF Fibroblasts

We previously showed that FSTL1 had a profibrotic role via facilitation of the TGF-β1 signal in lung fibrosis using a genetic model of Fstl1^+/– mice.⁷ To evaluate the effects of FSTL1 nAbs (2K6 or 4D22) during fibrosis, we used an in vitro model of TGF-β1-induced myofibroblast differentiation and ECM production. IPF fibroblasts were treated with 5 ng/mL TGF-β1 with/without 2K6 or 4D22 nAbs, and total protein was isolated from fibroblasts for western blot analyses. As shown in Figure 2C, 2K6 and 4D22 nAb treatment attenuated TGF-β1-induced canonical Smad signaling, as determined by the lower phosphorylation levels of Smad2/3. The 2K6 and 4D22 nAbs attenuated TGF-β1-induced myofibroblast differentiation and ECM production, as determined by the decrease in α-smooth muscle actin (α-SMA) (a marker of myofibroblast) expression and type I collagen (a well-known TGF-β-stimulated ECM component) production (Figure 2D). The similar blocking ability of these two nAbs was also confirmed using a human lung fibroblast cell line (HFL1) (Figures S6A and S6B) and primary mouse lung fibroblasts (Figures S6C and S6D). Therefore, FSTL1 nAbs recognized human and murine FSTL1 and blocked FSTL1-promoted TGF-β responsiveness in vitro.

FSTL1 nAbs Attenuated Bleomycin-Induced Lung Fibrosis in Mice

To further investigate the therapeutic role of FSTL1 nAbs in lung fibrosis in vivo, prophylactic and therapeutic lung fibrosis mouse models were used. Neutralizing FSTL1 with the 2K6 or 4D22 nAbs did not prophylactically prevent bleomycin-induced lung fibrosis (Figures 2E–2H). We then examined the potential of FSTL1 nAbs by administration of bleomycin into mice to initiate lung fibrosis, followed by i.p. injections of 2K6 or 4D22 nAbs from day 5 to day 11 (Figure 2I). FSTL1 nAb treatment significantly reduced lung fibrosis, as determined by hydroxyproline content, Masson’s trichrome staining, and the statistical percentage of fibrotic area (Figures 2J–2L).

We also examined whether FSTL1 nAbs prevented the further progression of established fibrosis. We injected mice with bleomycin to establish fibrotic disease and then injected 2K6 or 4D22 nAbs i.p. from day 8 to day 17 after bleomycin injury (Figure 2M). The mice were sacrificed on day 21, and samples were collected for fibrosis analysis. FSTL1 nAbs significantly reduced lung fibrosis after fibrotic disease was established (Figures 2N–2P). Collectively, these data confirmed the potent antifibrotic activity of the FSTL1 nAbs.

Elevated Expression of FSTL1 in Keloid and the Bleomycin Skin Fibrosis Model

An altered expression of FSTL1 is found in skin injuries.²⁷ Therefore, we hypothesized that FSTL1 was a broad profibrotic factor in multiple organs, such as skin. To test this hypothesis, we examined FSTL1 expression in patients with keloid, which is an overgrowth of fibrotic scar tissue at the site of skin injury, particularly a wound or a surgical incision.²⁸ We collected skin biopsies from seven keloids and three healthy controls and identified a remarkable increase in FSTL1 mRNA and protein expression in all disease samples compared with that in control subjects (Figures 3A and 3B). The increased FSTL1 mRNA and protein expression was also confirmed in keloid fibroblasts isolated from the above biopsies (Figures S7A and S7B) and in a published gene-profiling dataset (Figure S7C).²⁹ We also used a bleomycin model of skin fibrosis³⁰ and observed that bleomycin-induced injury stimulated Fstl1 mRNA and protein expression in skin (Figures 3C and 3D). Therefore, Fstl1 is critical for the pathogenesis of skin scarring and fibrosis.

Figure 3.

Fstl1 Depletion Protects Mice from Bleomycin-Induced Skin Fibrosis

(A and B) mRNA levels (A) and protein levels (B) of FSTL1 in healthy skin tissues (n = 3) and keloid skin tissues (n = 7) were measured by RT-PCR and western blot analysis. Relative density of FSTL1-normalized β-actin is represented. (C and D) mRNA levels (C) and protein levels (D) of FSTL1 in skin tissues of C57BL/6J mice after PBS or bleomycin treatment (n = 5 per group) were measured by RT-PCR and western blot analysis. Relative density of FSTL1-normalized β-actin is represented. (E–I) Fstl1^+/− and WT mice were treated with PBS or bleomycin for 14 d (n = 5 per group), and skin tissues were harvested for the following analyses: (E) qRT-PCR analysis of Fstl1; (F) H&E staining and (G) relative graphical presentation of dermal thickness data (scale bars, 100 μm); (H) qRT-PCR analysis of Col1a1 and α-SMA; and (I) western blot analysis of Col1 and α-SMA (β-actin was used as a loading control); relative densities of Col1 normalized to β-actin and α-SMA normalized to β-actin are represented. (J) Primary skin fibroblasts from C57BL/6J mice were treated with 5 ng/mL TGF-β1 for the indicated time points. FSTL1 protein expression in conditioned medium (supernatant [SN]) or fibroblasts (Cell) was determined by western blot. β-Tubulin was used as a loading control. Relative density of FSTL1-normalized β-tubulin is represented. (K) Primary skin fibroblasts from C57BL/6J mice were treated with 5 ng/mL TGF-β1 and 100 ng/mL FSTL1. Protein expression of α-SMA in cell extracts (Cell), and protein expression of Col1 in medium (supernatant) were determined by western blot 24 h after TGF-β1 treatment. β-Actin was used as a loading control. Relative densities of p-Smad3 normalized to Smad3 and α-SMA normalized to β-actin are represented. (L) Primary skin fibroblasts from Fstl1^+/− mice and WT littermates were treated with 5 ng/mL TGF-β1. p-Smad3 and Smad3 were analyzed by western blot 30 min after TGF-β1 treatment. Protein expression levels of α-SMA in cell extracts and Col1 in medium were detected by western blot 24 h after TGF-β1 treatment. β-Actin was used as a loading control. Relative densities of p-Smad3 normalized to Smad3 and α-SMA normalized to β-actin are represented. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (See also Figure S7.)

Fstl1 Depletion Protected Mice from Bleomycin-Induced Skin Fibrosis

The profibrotic role of Fstl1 in skin fibrosis was assessed using a genetically manipulated model of Fstl1^+/– mice, which exhibited a 47.04% decrease in Fstl1 mRNA levels in Fstl1^+/– skin tissue and a 72.75% decrease in Fstl1 mRNA levels in bleomycin-treated Fstl1^+/– skin tissue (Figure 3E). Fstl1 depletion protected mice from bleomycin injury, and Fstl1^+/– mice showed reduced skin fibrosis, as determined by the decreased dermal thickness (Figures 3F and 3G) and reduced type I collagen production (Figures 3H and 3I) compared with wild-type (WT) controls. The attenuated fibrosis was further supported by a decrease in α-SMA expression, which indicated reduced accumulation of myofibroblasts in Fstl1^+/– skin (Figures 3H and 3I).

To evaluate whether Fstl1-modulated TGF-β1 signaling was involved in fibroblast activation and collagen production in skin fibroblasts, we isolated primary fibroblasts from mouse skin and treated the cells with TGF-β1. Figure 3J shows that TGF-β1 induced the expression and secretion of FSTL1 in a time-dependent manner. Gain-of-function experiments revealed that recombinant FSTL1 protein promoted TGF-β1-induced myofibroblast activation and type I collagen production (Figure 3K). As expected, the loss-of-function experiment using fibroblasts isolated from Fstl1^+/– skin (Figure 3L) confirmed the decrease in TGF-β1-induced Smad3 phosphorylation, α-SMA expression, and type I collage production. Taken together, our data indicate that Fstl1 is induced in response to skin injury and involved in skin fibrosis via TGF-β1 signal activation. Fstl1 may be causally involved in driving skin fibrosis as a profibrotic factor.

FSTL1 nAbs Attenuated TGF-β1 Responsiveness In Vitro and Bleomycin-Induced Skin Fibrosis In Vivo

To determine the anti-fibrotic effect of FSTL1 nAbs on skin scar and fibrosis, we first used an in vitro model of TGF-β1-induced skin myofibroblast differentiation and ECM production.³¹ Figures 4A and 4B confirm the decrease in FSTL1-promoted TGF-β1-induced Smad3 phosphorylation, α-SMA expression, and type I collage production with the administration of FSTL1 nAbs (2K6 or 4D22).

Figure 4.

FSTL1 nAbs Attenuate TGF-β1 Responsiveness In Vitro and Bleomycin-Induced Skin Fibrosis In Vivo

(A and B) Primary skin fibroblasts from C57BL/6J mice were treated with 1 μg/mL FSTL1 nAbs (2K6 or 4D22) or control IgG2b 1 h before 5 ng/mL TGF-β1 treatment. (A) p-Smad3 and Smad3 were analyzed by western blot 30 min after TGF-β1 treatment. (B) Protein expression levels of α-SMA in cell extracts and Col1 in medium were detected by western blot 24 h after TGF-β1 treatment. Relative densities of p-Smad3 normalized to Smad3 and α-SMA normalized to β-actin are represented in the graphs. (C) Bleomycin-induced skin fibrosis model in C57BL/6J mice. Bleomycin (20 mg) was subcutaneously injected into the upper back of mice daily for 14 consecutive days in a total volume of 100 μL, and FSTL1 nAbs (10 μg/ml) were administered in the same total volume combined with bleomycin every day (n = 5 per group). (D) Representative images of H&E stained (upper) and α-SMA immunohistochemical-stained (lower) sections of bleomycin-treated mice skin. Scale bars, 100 μm. (E) Graphical presentation of dermal thickness data shown in (D). (F and G) Relative mRNA expression of α-SMA (F) and Col1a1 (G) in skin tissues of bleomycin-treated mice. (H) Intervention dosing regimen of bleomycin-induced skin fibrosis model in mice. Bleomycin (20 mg) was subcutaneously injected into the upper back of mice daily for 14 consecutive days in a total volume of 100 μL, and FSTL1 nAbs (10 μg/mL) were administered in the same total volume from day 7 to day 14 combined with bleomycin (n = 5 per group). (I) Representative images of H&E-stained (upper) and α-SMA immunohistochemical-stained (lower) sections of mice skin. Scale bars, 100 μm. (J) Graphical presentation of dermal thickness data shown in (I). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

We intradermally injected 2K6 or 4D22 nAbs daily in combination with bleomycin, and skin samples were collected on day 14 (Figure 4C).³⁰ The prevention mouse model showed that blockade of FSTL1 function with the 2K6 or 4D22 nAbs protected mice from bleomycin injury-induced skin fibrosis, as indicated by the decreased dermal thickness, α-SMA-positive myofibroblast population (Figures 4D and 4E), α-SMA expression, and type I collagen production (Figures 4F and 4G). We also used a therapeutic model of daily intradermal injections of 2K6 or 4D22 nAbs from day 7 to day 14 after the initiation of the bleomycin injections (Figure 4H). Figures 4I and 4J show that FSTL1 nAbs also significantly ameliorated bleomycin-induced dermal fibrosis even after fibrosis was established.

FSTL1 nAbs Attenuated TGF-β1-Induced Fibrosis in Human Skin Explants

The anti-fibrotic effects of FSTL1 nAbs was further assessed to test whether the nAbs were effective in an ex vivo model of TGF-β1-induced fibrosis in human skin explants (Figure 5A),³⁰ in which TGF-β1 intradermal injections significantly increased the dermal thickness of human skin explants in a dose-dependent manner (Figures 5B and 5C). Coinjection of 2K6 or 4D22 nAbs in combination with TGF-β1 (10 ng/mL) (Figure 5D) reduced TGF-β1-induced skin fibrosis, as determined by dermal thickness and Masson’s trichrome staining (Figures 5E and 5F). Injections of 2K6 or 4D22 nAbs 3 days after TGF-β1 administration (Figure 5G) significantly ameliorated TGF-β1-induced dermal fibrosis (Figures 5H and 5I). Collectively, our data suggest that FSTL1 nAbs (2K6 and 4D22) are a viable therapeutic option for fibrosis in several organs because these antibodies reversed ongoing fibrosis in the lungs and skin.

Figure 5.

FSTL1 nAbs Attenuate TGF-β1-Induced Fibrosis in Human Skin Explants

(A) TGF-β1-induced ex vivo human skin fibrosis model. Different doses of TGF-β1 (1 ng/mL, 10 ng/mL, 50 ng/mL) were injected intradermally in cultured human skin (n = 3), and skin tissues were collected after 1 week. PBS was used as negative control. (B) Representative images of H&E-stained (upper) and Masson’s trichrome-stained (lower) sections of TGF-β1-treated human skin. Scale bars, 100 μm. (C) Graphical presentation of dermal thickness data shown in (B). (D) Treatment model of FSTL1 nAbs in TGF-β1-induced human skin fibrosis model. 2K6 and 4D22 nAbs (10 μg/mL) in combination with TGF-β1 (10 ng/mL) were injected intradermally in cultured human skin (n = 3), and skin tissues were collected after 1 week. PBS was used as negative control. (E) Representative images of H&E-stained (upper) and Masson’s trichrome-stained (lower) sections of TGF-β1-treated human skin. Scale bars, 100 μm. (F) Graphical presentation of dermal thickness data shown in (E). (G) Intervention dosing regimen of TGF-β1-induced ex vivo human skin fibrosis model. TGF-β1 (10 ng/mL) was injected intradermally in cultured human skin (n = 3), FSTL1 nAbs (10 μg/mL) were injected intradermally on day 3 after TGF-β1 treatment, and skin tissues were collected after 1 week. PBS was used as negative control. (H) Representative images of H&E-stained (upper) and Masson’s trichrome-stained (lower) sections of TGF-β1-treated human skin. Scale bars, 100 μm. (I) Graphical presentation of dermal thickness data shown in (H). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

FSTL1 nAbs Attenuated the Inflammatory Response In Vitro and the Severity of CIA In Vivo

Elevated FSTL1 expression and high serum concentrations of FSTL1 in patients with RA and in a CIA model have been reported,^22,23 and the pro-inflammatory role of Fstl1 has been investigated using a genetic mouse model that was hypomorphic for Fstl1.³² To test the anti-inflammatory effect of FSTL1 nAbs (2K6 or 4D22) in RA, we first used an in vitro model of IL-1β-induced production of inflammatory chemokines in human synovial fibroblasts (SW982), which are a vital stromal cell population in the pathogenesis of RA. As expected, IL-1β stimulated FSTL1 protein expression in SW982 cells^33,34 (Figure 6A), and 2K6 and 4D22 nAb treatment attenuated IL-1β-induced IL-6 and MCP-1 production (Figures 6B and 6C). The similar blocking abilities of these two nAbs was also confirmed using the newly isolated mouse mesenchymal stromal cells (MSCs)³² (Figures 6D–6F). Therefore, FSTL1 nAbs recognized FSTL1 and blocked FSTL1-promoted inflammatory responses in vitro.

Figure 6.

FSTL1 nAbs Decrease the Severity of Collagen-Induced Arthritis in Mice

(A) SW982 cells were cultured for 2 days in the absence or presence of IL-1β (10 ng/mL), and supernatants and cells were collected and assayed for FSTL1 by western blotting. β-Tubulin was used as a loading control. Relative density of FSTL1-normalized β-tubulin is represented. (B and C) SW982 cells were treated with 1 μg/mL 2K6/4D22/control IgG2b with or without 10 ng/mL IL-1β and cultured for 3 days, and supernatants were collected and assayed for IL-6 (B) and MCP-1 (C) by ELISA. Each bar represents the mean ± SEM of triplicate independent experiments. (D) MSCs were cultured for 3 days in the absence (medium) or presence of IL-1β (10 ng/mL), and supernatants and cells were collected and assayed for FSTL1 by western blotting. β-Tubulin was used as a loading control. Relative density of FSTL1-normalized β-tubulin is represented. (E and F) MSCs were treated with 1 μg/mL 2K6/4D22/IgG2b with or without 10 ng/mL IL-1β and cultured for 3 days, and supernatants were collected and assayed for IL-6 (E) and MCP-1 (F) by ELISA. Each bar represents the mean ± SEM of triplicate independent experiments. (G) Intervention dosing regimen of CIA model. DBA/1 mice were immunized with type II collagen on day 0 and day 21 to induce CIA. 2K6 and 4D22 mAbs and their isotype-matched control IgG2b (50 μg/mouse/time point; n = 10 per group) were administered twice a week after day 21 and mice were sacrificed at day 60. (H) The mean arthritic index is shown. (I and J) The representative paws (I) and mean paw swelling (J) are shown. (K) Knee joints and elbow joints of mice from each group were sectioned and stained with H&E. Scale bars, 100 μm. (L and M) Protein concentration of IL-6 and MCP-1 in serum (L) and mRNA expression of IL-6 and MCP-1 in spleen tissues (M) of CIA-treated mice were measured by ELISA and RT-PCR, respectively. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To further determine whether the blocking abilities of FSTL1 nAbs were effective in the pathogenesis of RA, we immunized DBA/1 mice with type II collagen (CII) on day 0 and day 21 to induce arthritis. The 2K6 and 4D22 nAbs and their isotype-matched control mAbs were administered every third day after day 21, and the mice were sacrificed on day 60 (Figure 6G). A considerable reduction in the arthritis index and paw swelling was observed in mice treated with the FSTL1 nAbs (Figures 6H–6J). Histology examination confirmed the remarkable alleviation of inflammatory infiltration, synovial hyperplasia, and the erosion of bone and cartilage after the treatment of FSTL1 nAbs (Figure 6K). In addition, the administration of 2K6 and 4D22 nAbs decreased the expression of proinflammatory cytokines IL-6 and MCP-1 in serum and spleen tissues in CIA-treated mice (Figures 6L and 6M). Collectively, our data demonstrate that FSTL1 plays a crucial role in fibrogenesis and the inflammatory response. As a proof of principle, neutralizing FSTL1 with antibodies substantially attenuated fibrosis in bleomycin or TGF-β1 models of lung and skin fibrosis and the inflammatory response in a CIA mouse model. These data are encouraging because blockade of FSTL1 using nAbs would be an attractive strategy for patients with progressive organ fibrosis and systemic autoimmune diseases.

Discussion

FSTL1 is a highly conserved matricellular protein that does not contribute structurally to the ECM, but it mediates interactions at the cell-matrix interface. This link governs fundamental cellular functions, including cell adhesion, proliferation, and differentiation. Although mounting evidence has implicated FSTL1 in tissue remodeling and inflammatory processes, pharmacological inhibition of FSTL1 is not currently available. The findings reported in the present study add new FSTL1 nAbs with distinct targeting epitopes as tools for the efficient blockade of FSTL1 in the extracellular space. We provide data at the cellular and animal levels and proof-of-principle intervention data to support the blockade of FSTL1 using nAbs as an attractive strategy for patients with progressive organ fibrosis and systemic autoimmune diseases.

We offer the following lines of evidence, together with other studies,^24,32 to demonstrate FSTL1 nAbs (2K6 and 4D22) as new mechanistically targeted anti-fibrotic and anti-inflammatory treatments. Elevated levels of tissue and/or serum FSTL1 are seen in fibrotic disorders, such as IPF and keloid, and systemic autoimmune diseases, such as RA.²⁴ Genetic mutation of Fstl1 in mice significantly attenuated bleomycin-induced pulmonary and dermal fibrosis and prevented a type II collagen-induced inflammatory response.³² Fstl1 promoted the activation/differentiation of myofibroblasts, ECM production, the inflammatory response of stromal cells, and the production of chemokines. Blockade of FSTL1 using nAbs reduced bleomycin- and/or TGF-β1-induced pulmonary and dermal fibrosis in vivo and ex vivo and prevented the type II collagen-induced inflammatory response in mice in vivo.

We found that FSTL1 was predominantly secreted from the lung mesenchymal fraction rather than from epithelial cells and macrophages, and its expression closely correlated with myofibroblast activation and differentiation. FSTL1 was also elevated in fibrotic disease states in various mouse models of lung fibrosis.^7,35 Maruyama et al.³⁶ reported a decrease in cardiac myofibroblast number and ECM production following endogenous Fstl1 ablation in S100A4-Cre/Fstl1^f/f mice using a LAD (left anterior descending artery)-induced mouse model of myocardial infarction (MI). Shang et al.³⁷ reported that Fstl1 played an important role in liver fibrosis and that the deletion of Fstl1 attenuated hepatic stellate cell activation via suppression of the TGF-β1/Smad3 signaling pathway. Higher serum levels of FSTL1 were reported in many systemic autoimmune diseases, such as RA, osteoarthritis (OA), SSc, and Sjögren’s syndrome.^24,38 Studies have also shown that FSTL1 was primarily produced by cells of mesenchymal lineage, including osteocytes, chondrocytes, and synovial fibroblasts, during the progression of these systemic autoimmune diseases and exerted proinflammatory effects.^6,33,39 Notably, the Fstl1-regulated activation of mesenchymal cells appears to be a common mechanism underlying the pathological fibrogenesis of multiple organs and the inflammatory response in multiple systemic autoimmune diseases. Therefore, the use of FSTL1 nAbs (2K6 and 4D22) as a therapeutic strategy is attractive for the treatment of patients with liver and heart fibrotic diseases and other systemic autoimmune diseases, such as OA and SSc, but its use needs further verification.

Alternative strategies to target the TGF-β1 signaling pathway were investigated for IPF. Integrin αvβ6 binds and activates latent TGF-β1 regulation of pulmonary inflammation and fibrosis.⁴⁰ Early human studies targeting integrin αvβ6 in IPF are underway.41, 42, 43 A phase I study designed to examine the safety and tolerability of a single nebulized dose of GSK3008348 (an integrin αv antagonist) was recently started in healthy volunteers, and the treatment will be subsequently tested in patients with IPF (ClinicalTrials.gov: NCT02612051). A mAb targeting integrin αvβ6 (BG00011, formerly STX-100) was tested recently in a phase II clinical trial (ClinicalTrials.gov: NCT01371305). Connective tissue growth factor (CTGF) is a downstream mediator of TGF-β that confers susceptibility to bleomycin-induced fibrosis in a fibrosis-resistant BALB/c mouse strain.⁴⁴ Recent phase II trials of anti-CTGF mAbs in humans showed no particular side effects, and there was no evidence of disease progression using pulmonary function testing in patients who received the drug.⁴⁵ Phase III clinical trials of the anti-CTGF mAb pamrevlumab (formerly F-3019) in IPF are currently underway. Our findings suggest that blockade of FSTL1, which is a key activator of TGF-β1 signaling in lung, is an alternative approach, and it may be possible to interfere in the TGF-β1 signaling pathway at sites of FSTL1 upregulation without affecting other homeostatic roles of TGF-β1 signaling. Our proof-of-principle neutralizing antibody experiments are encouraging. Therefore, the inhibition of FSTL1 with neutralization antibodies for IPF treatment may provide an advantage compared to blockade of TGF-β1 or its receptors directly, similar to the antibodies to integrin αvβ6.⁴¹ This approach of blocking FSTL1-TGF-β1 signaling may also be a promising therapeutic agent for broad fibrotic disorders. Because FSTL1 nAbs (2K6 and 4D22) prevented and reversed pulmonary and skin fibrosis, these agents may be useful in other organs, such as fibrosis in MI and liver cirrhosis.

The ECM provides structural support and regulates cellular responses. Central to the ECM structural support and cell-matrix interactions are nonstructural matricellular proteins, such as the SPARC family. There is evidence of the clinical importance of SPARC, FS, and other matricellular proteins as potential biomarkers and therapeutic targets.⁴⁶ FSTL1 shows functional similarity to SPARC and SMOC2 in promoting bleomycin-induced lung fibrosis.^47,48 An increased production of SPARC was also found in patients with RA and OA.⁴⁹ SPARC expression has been correlated with increased migration and poor prognosis in human malignancies, such as glioblastoma, pancreatic adenocarcinoma, and gastric cancer.50, 51, 52 SPARC also acted as a key player in the pathologies associated with obesity and diabetes.⁵³ These data identify a core pathway that regulates tissue injury/repair and the immune response and suggest that the pharmacological targeting of different matricellular proteins has clinical utility in the treatment of patients with a broad range of fibrotic and systemic autoimmune diseases.

There are currently no effective therapies for organ fibrosis and systemic autoimmune diseases. We demonstrated that FSTL1 nAbs (2K6 and 4D22) were promising therapeutic agents for fibrotic disorders, such as IPF and keloid, and systemic autoimmune diseases, such as RA. Because these two nAbs blocked Fstl1-mediated mesenchymal cell activation and reversed dermal and pulmonary fibrosis and inflammatory responses, they may inhibit fibrosis in other tissues, such as liver cirrhosis, and attenuate inflammatory responses in other systemic autoimmune diseases, such as SSc, OA, and Sjögren’s syndrome. Further efforts to examine the therapeutic potential of FSTL1 nAbs are actively being pursued in our laboratories.

Materials and Methods

Study Design

The objective of our study was to screen and identify specific monoclonal nAbs to functionally block FSTL1 and evaluate the application of FSTL1 nAbs in fibrotic and systemic autoimmune diseases. We used luciferase and quantitative RT-PCR (qRT-PCR) assays to select nAbs that specifically block FSTL1-facilitated TGF-β1 responsiveness. MST and pull-down assays were used to verify the affinity of selected FSTL1 nAbs. Biological assays of FSTL1 nAbs were utilized in several in vitro models that were established based on reported biological functions of FSTL1. The effects of FSTL1 nAbs in fibrotic diseases were evaluated in vitro using primary human lung or skin fibroblasts, ex vivo using human skin, and in vivo using bleomycin-induced pulmonary and skin fibrosis in mice. The effects of FSTL1 nAbs in systemic autoimmune diseases were evaluated in vitro using human synovial fibroblasts SW982 and primary mouse MSCs, and in vivo using a CIA mouse model.

Human Samples

The peripheral blood samples from IPF patients (n = 19) and normal control individuals (n = 20) were obtained from the China-Japan Friendship Hospital (Beijing, China). The IPF patients were diagnosed by physical examination, pulmonary function studies, chest high-resolution computed tomography (HRCT), and bronchoalveolar lavage (BAL) findings according to the respective diagnostic criteria for this disease⁵⁴. Human primary pulmonary fibroblasts⁵⁵ were provided by C.A. Feghali-Bostwick (University of Pittsburgh School of Medicine, Pittsburgh, PA, USA) and cultured in DMEM (Gibco) supplemented with 10% FBS (HyClone) and antibiotics in 5% CO₂ at 37°C in a humidified atmosphere. Cells were used from five to seven generations. The human skin tissues from keloid patients (n = 7) and normal control individuals (n = 3) were obtained from the Third Hospital of the Beijing Armed Police Corps. The keloid patients were diagnosed according to the respective diagnostic criteria for this disease⁵⁶. Human primary fibroblasts from keloid patients and normal control individuals were isolated and cultured as previously described.⁵⁷ Cells were used between passages 2 and 5. Human thigh skin with bulk area used for ex vivo human skin assays was obtained from cosmetic plastic surgery in the Third Hospital of the Beijing Armed Police Corps. The clinical characteristics of patients and normal control individuals are shown in Table 2.

Table 2.

Clinical Characteristics of the Study Groups

	Normal (n = 20)	IPF patients (n = 18)	Normal (n = 3)	Keloid Patients (n = 7)
Age, years	59.7 ± 2.8	63.7 ± 2.2	25.0 ± 3.8	24.0 ± 1.3
Sex
Male, n (%)	7 (35)	15 (83.3)	1 (33.4)	2 (28.6)
Female, n (%)	13 (65)	3 (16.7)	2 (66.6)	5 (71.4)
Pathology	–	IPF/UIP	–	keloid
Serum FSTL1 (ng/mL)	20.6 ± 3.5	36.9 ± 4.5∗∗∗	–	–

Values are mean ± SEM. ∗∗∗p < 0.001, for difference compared with normal.

Mice

Fstl1^+/− mice were generated as previously described,⁶ and mice were backcrossed onto the C57BL/6J background for at least 12 generations before use. 6- to 8-week-old male C57BL/6J mice and 6- to 8-week-old male DBA/1J mice were purchased from Beijing HFK Bioscience. All mice were housed and cared for in a pathogen-free facility in Nankai University.

Cell Line Culture

Human pulmonary epithelial cell line A549 and monkey kidney fibroblast cell line COS-7 were obtained from ATCC. Human fetal lung fibroblast cell line HFL1 was purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). NIH 3T3 cells that were stably transfected with CAGA-luciferase reporter (also named CAGA-NIH 3T3 cells) were kindly provided by Prof. Zhinan Yin (Nankai University). Human synovial fibroblast cell line SW982 was purchased from Procell Life Science & Technology (Wuhan, China). A549, COS-7, CAGA-NIH 3T3, and SW982 cells were cultured in DMEM (Gibco), and HFL1 cells were cultured in F12-K (Gibco), which were all supplemented with 10% FBS (HyClone) and antibiotics in 5% CO₂ at 37°C in a humidified atmosphere.

Anti-Fstl1 mAbs

A panel of mouse mAbs (derived from 38 independent hybridoma clones) that specifically targets FSTL1 protein was generated by Abmart using SEAL technology.²⁵ A total of 29 different peptides that uniquely represented FSTL1 protein surface sequences were selected as antigens to immunized mice. After hybridomas were acquired, cells went through several rounds of limiting dilution cloning and were screened by peptide antigens. A total of 38 mAbs recognizing 16 different peptide antigens were identified and characterized for functional blocking.

Luciferase Assay

Serum-starved CAGA-NIH 3T3 cells were pretreated with 1 μg/mL mAbs for 1 h and then treated with 5 ng/mL TGF-β1 for 18 h. After washing with PBS, cells were harvested and the luciferase activity of cell lysates was determined by using a luciferase assay system (Promega) as described by the manufacturer. The total light emission during the initial 20 s of the reaction was measured with a luminometer (Lumat LB 9501, Berthold Technologies). All assays were repeated in triplicate.

Isolation and Culture of Primary MEFs

MEFs were prepared as described previously.⁵⁸ Newly isolated MEFs were cultured in DMEM supplemented with 10% FBS (HyClone) and antibiotics in 5% CO₂ at 37°C in a humidified atmosphere. For TGF-β1 treatment, MEFs were serum starved for 24 h and then pretreated with 1 μg/mL mAbs for 1 h followed by 5 ng/mL TGF-β1 (R&D Systems) treatment for another 24 h. Total RNA was isolated with TRIzol (Invitrogen), and mRNA expression of Co1a1, Col1a2, and Fn were measured by qRT-PCR.

MST

The fluorescence labeling of ligands (mouse FSTL1, human FS, and human FSTL1) was performed as previously described.⁵⁹ In brief, 100 μL of a 20 μM solution of ligand proteins in labeling buffer was mixed with 100 μL of 60 μM NT647-NHS fluorophore (NanoTemper Technologies, Germany) in labeling buffer and was incubated for 30 min at room temperature (RT). Interactions observed between 2K6/4D22 antibodies and NT647 ligands were performed with a NanoTemper Monolith NT.115 (NanoTemper Technologies, Germany). Serial dilutions of 2K6 and 4D22 antibodies were incubated with 50 nM NT647 ligands in assay buffer (PBS including 0.05% Tween 20) to yield a final volume of 20 μL per dilution. The reaction mixtures were loaded into standard-treated capillaries and then analyzed by MST at 20% excitation power and at a light-emitting diode (LED) intensity of 30%.

Pull-Down Assay

2 μg of 2K6 or 4D22 was incubated with 200 ng of FSTL1, TGF-β1, or FS protein in 0.3 mL of TNE buffer containing 25 mM Tris-HCl (pH 7.6), 150 mM NaCl, 5 mM EDTA, and 0.1% Triton X-100 overnight at 4°C. 20 μL of resuspended protein G agarose (QIAGEN) was added to the samples and incubated for 4 h at 4°C. The supernatant was discarded after washing with TNE buffer three times, and agarose was boiled and analyzed by using a western blot assay. His-tagged FSTL1 was concentrated by nickel-nitrilotriacetic acid (Ni-NTA)-agarose beads (QIAGEN), which could capture His-tagged proteins. After washing, the beads were incubated with TGF-β1 and FSTL1 nAbs (or immunoglobulin [Ig]G2b) for 4 h at 4°C, then analyzed by immunoblotting with anti-TGF-β1 antibodies. HEK293 cells were transfected with Myc-His-tagged Fstl1 and hemagglutinin (HA)-tagged TβRII with Lipofectamine (Invitrogen) and treated with FSTL1 nAbs or IgG2b for 24 h. In vitro-translated Myc-His-tagged FSTL1 was pulled down using Ni-NTA agarose beads and then analyzed by immunoblotting with anti-HA antibody.

Isolation and Culture of Primary Neonatal Rat Ventricular Myocytes

Primary neonatal rat ventricular myocytes (NRVMs) were prepared as described previously.²⁶ Newly isolated NRVMs were incubated in DMEM supplemented with 10% FBS for 48 h and serum starved for 24 h. After pretreatment with 1 μg/mL 2K6/4D22 or control IgG2b for 1 h, NRVMs were treated with 100 ng/mL PE (Sigma-Aldrich) with/without 100 ng/mL FSTL1 for another 24 h. Morphological changes were observed with a phage-contrast light microscopy (DFC420C; Leica), and cell surface areas were examined by Image-Pro Plus version 6.0.

Isolation and Culture of Primary Lung Fibroblasts from Adult Mice

Primary pulmonary fibroblasts isolated from 6- to 8-week-old C57BL/6J mice were cultured in DMEM supplemented with 10% FBS and antibiotics in 5% CO₂ at 37°C in a humidified atmosphere as described previously.⁷ Cells of passages 3–4 were used for myofibroblast activation and ECM production assays. After the pretreatment with 1 μg/mL 2K6, 4D22, or control IgG2b and the following 5 ng/mL TGF-β1 treatment, fibroblasts and medium were collected, respectively. The Smad2/3 phosphorylation level in cell extracts was measured 30 min after TGF-β1 treatment, and α-SMA expression levels in cell extracts or the type I collagen production levels in supernatants of medium 24 h after TGF-β1 treatment were measured by using western blot analysis.

Bleomycin-Induced Pulmonary Fibrosis Model and nAb Administration

For bleomycin model building, 2 U/kg bleomycin (Nippon Kayaku, Japan) in 20–25 μL of PBS was administered intratracheally by using a 1-mL syringe with a 25G needle with insertions between the cartilaginous rings of the trachea. To investigate whether neutralizing antibody could prevent fibrotic progression in the bleomycin mouse model, FSTL1-neutralizing antibody (clone 2K6/4D22) or its control isotype antibody (IgG2b) was i.p. injected (50 μg/mouse/each time) on days −3, −2, and −1, separately, after bleomycin treatment. The lungs were harvested 14 d after injury. To prove the therapeutic role of neutralizing antibody during the early stage of fibrogenesis, FSTL1-neutralizing antibody (clone 2K6/4D22) or its control isotype antibody (IgG2b) was i.p. injected (50 μg/mouse/each time) on days 5, 8, and 11, respectively, after bleomycin treatment. Samples were harvested 14 days after bleomycin injury. During the late stage, FSTL1-neutralizing antibody (clone 2K6/4D22) or its control isotype antibody (IgG2b) was i.p. injected (50 μg/mouse/each time) on days 8, 11, 14, and 17 after bleomycin treatment. Samples were harvested 21 days after bleomycin injury. Masson’s trichrome was taken for the evaluation of the fibrosis degree, and collagen contents were measured with a hydroxyproline assay.

Hydroxyproline Assay

Collagen contents in right lungs of 10 mice in each group were measured with a conventional hydroxyproline method.⁶⁰ The ability of the assay was confirmed by measuring samples containing known amounts of purified collagen.

Bleomycin-Induced Skin Fibrosis Model and nAb Administration

Bleomycin was used to induce dermal fibrosis in mice as previously described.³⁰ In brief, male 6- to 8-week-old Fstl1^+/− mice and their WT littermates or C57BL/6J mice were subcutaneously injected with 20 mg of bleomycin into the upper back daily for 14 consecutive days in a total volume of 100 μL. For nAb treatment, 2K6/4D22 or control IgG2b (10 μg/mL) was administered in the same total volume with bleomycin from day 1 to day 14 or day 7 to day 14 of the bleomycin treatment. Mice were sacrificed at day 14, and skin surrounding the injection site was harvested for a biomarker assay or histological analysis.

Isolation and Culture of Primary Dermal Fibroblasts

Primary dermal fibroblasts from newborn C57BL/6J mice were isolated and cultured as previously described.⁶¹ Cells were used between passages 2 and 5.

Ex Vivo Human Skin Assays

The model of TGF-β1-induced fibrosis in human skin explants was established as previously described.⁶² In brief, skin samples were cut into 1-cm² size and cultured in an air-liquid interface with the epidermal side up. The following were injected intradermally in a total volume of 100 μL of 1× PBS: PBS alone, TGF-β1 alone (1 ng/mL, 10 ng/mL, 50 ng/mL), 2K6/4D22, or control IgG2b (10 μg/mL) in combination with TGF-β1 (10 ng/mL), 2K6/4D22, or control IgG2b (10 μg/mL) alone on day 3 of TGF-β1 treatment. After 1 week, skin tissue corresponding to an area with an 8-mm diameter centered around the injection site was harvested and fixed in 10% formalin and embedded in paraffin for histological analysis.

Isolation and Culture of Primary MSCs

Mouse MSCs were isolated and cultured as previously described.⁶³ All of the cells were used between passages 3 and 4. MSCs were treated with 10 ng/mL IL-1β (PeproTech) and cultured for 3 days, then supernatants and cells were collected for subsequent assays.

Induction of CIA, nAb Treatment, and Clinical Scoring of Arthritis

For arthritis induction, male 6- to 8-week-old DBA/1J mice were injected intradermally at the base of the tail with 150 μg of bovine type II collagen (Chondrex) emulsified in CFA (Sigma) on day 0, and 21 days later, a second immunization (booster) of 100 μg of type II collagen in IFA (Sigma) was given in the same way as previously described.²² The immunized mice were divided randomly and received either FSTL1 nAbs (2K6/4D22) or control IgG2b (50 μg/body) i.p. twice a week from day 21. The swelling of four paws was evaluated every 3 days after the second immunization by using a macroscopic scoring system raging from 0 to 4 as follows: 0, no swelling; 1, slight edema or redness of paw or joints; 2, mild swelling of ankle or wrist joints; 3, severe swelling of paw and joints; 4, maximum edema and erythema from ankle to entire leg. Each paw was graded and the maximal score per mouse was 16. The mean arthritis score was used to compare the difference between control and experimental groups. Mice were sacrificed on day 60, and joints were used for histopathology, whereas synovial tissues were homogenized for an ELISA assay and spleens were used for RT-PCR analysis.

ELISA

Levels of IL-6 in COS-7 cell supernatant, and levels of IL-6, and MCP-1 in MSCs, SW982 cell supernatant, and synovial tissue lysates were measured with commercial ELISA kits (eBioscience) according to the manufacturer’s instructions. Levels of FSTL1 in serum of IPF patients were measured with commercial ELISA kits (Cloud-Clone) according to the manufacturer’s instructions. For synovial tissue lysate preparation, synovial tissues were isolated from the knee joints and homogenized in PBS containing a mixture of protease inhibitors (Roche). After centrifugation twice for 15 min at 10,000 × g at 4°C, the supernatants were recovered and protein levels were determined by the bicinchoninic acid (BCA) system (Pierce Biotechnology, Rockford, IL, USA). Lysates with equal amounts of protein were then subjected to an ELISA assay.

Western blotting Analysis

The proteins were extracted from cells and tissues following standard protocols as described previously.⁷ For western blotting analysis, the following antibodies were used to recognize proteins: α-SMA, β-actin, and β-tubulin (Santa Cruz Biotechnology); phosphorylated (p-)Smad2, Smad2, p-Smad3, and Smad3 (Cell Signaling Technology); FSTL1, TGF-β1, and FS (R&D Systems); and type I collagen (Abcam).

qRT-PCR

RNA isolation, reverse transcription, and qRT-PCR analysis were performed as previously described.⁶ The expression levels of Col1a1, Col1a2, Fn, N-cad, E-cad, FSTL1, Fstl1, α-SMA, IL-6, and MCP-1 were determined by using a SYBR Green master mix kit (Roche). Gene expression was measured relative to the endogenous reference gene, mouse β-actin, or human GUSB. The sequences of the specific primer sets are described in Table 3.

Table 3.

The Sequences of Primer Sets Used in qRT-PCR

Gene	Accession No.: GenBank		Sequences (5′→3′)
Col1a1	NM_007742.3	for	CCAAGAAGACATCCCTGAAGTCA
		rev	TGCACGTCATCGCACACA
Col1a2	NM_007743.3	for	GGTGAGCCTGGTCAAACGG
		rev	ACTGTGTCCTTTCACGCCTTT
Fn	NM_010233.1	for	GTGTAGCACAACTTCCAATTACGAA
		rev	GGAATTTCCGCCTCGAGTCT
N-cadherin	NM_001792.5	for	ATAGCCCGGTTTCACTTGAGA
		rev	CAGGCTTTGATCCCTCTGGA
E-cadherin	NM_009864.2	for	CAGCCTTCTTTTCGGAAGACT
		rev	GGTAGACAGCTCCCTATGACTG
FSTL1	NM_007085.4	for	TCTGTGCCAATGTGTTTTGTGG
		rev	TGAGGTAGGTCTTGCCATTACTG
Fstl1	NM_008047.5	for	TTATGATGGGCACTGCAAAGAA
		rev	ACTGCCTTTAGAGAACCAGCC
α-SMA	NM_007392.2	for	GCTGGTGATGATGCTCCCA
		rev	GCCCATTCCAACCATTACTCC
IL-6	NM_031168.2	for	TAGTCCTTCCTACCCCAATTTCC
		rev	TTGGTCCTTAGCCACTCCTTC
MCP-1	NM_011333.3	for	ATTCTGTGACCATCCCCTCAT
		rev	TGTATGTGCCTCTGAACCCAC
β-actin	NM_007393.3	for	AGGCCAACCGTGAAAAGATG
		rev	AGAGCATAGCCCTCGTAGATGG
GUSB	NM_000181.3	for	CCAAACCAGCCTGACAACTT
		rev	TCTAGCATGCTCCACCACTG

for, forward; rev, reverse.

Histology and Immunohistochemistry

For toxicity evaluation of FSTL1 nAbs, the hearts, livers, spleens, left lungs, and kidneys were free from blood and fixed for 24 h and then embedded in paraffin. The sections were stained with hematoxylin and eosin (H&E) to evaluate the morphological changes.

For the bleomycin-induced pulmonary fibrosis model, the left lungs free from blood in mice of each group were dilated with 0.4 mL of 10% formalin after the trachea was cannulated. The lungs were fixed for 24 h and embedded in paraffin. The sections were stained with Masson’s trichrome to evaluate the extent of fibrosis. For pulmonary fibrosis quantization, Masson’s trichrome-stained slides were imaged with an Olympus BX53 microscopy at a magnification of ×100 and scored for a total of 10 random fields per sample. Images were analyzed by Image Pro-Plus 6.0 software (Media Cybernetics).

For the bleomycin-induced skin fibrosis model and TGF-β1-induced fibrosis in human skin explants, the skin tissues from each group were fixed in 10% formalin for 24 h and embedded in paraffin. 6-mm sections of paraffin-embedded mouse and human skin tissues were stained with H&E or Masson’s trichrome to evaluate the extent of fibrosis. Images were taken on a Olympus BX53 microscopy at a magnification of ×100, and the thickness of the dermis was measured from the epidermal-dermal junction to the dermal-fat junction with ImageJ software. Measurements from 10 random fields were averaged. Immunohistochemistry of 6-mm sections of paraffin-embedded mouse lung tissues was performed as described.⁷ The antibody specific for α-SMA (Santa Cruz Biotechnology), horseradish peroxidase (HRP)-polymer secondary antibodies, and diaminobenzidine (DAB) solution (Maxin Bio) were used for staining.

For the CIA model, mice knee joints and elbow joints from each group were fixed in 10% formalin, decalcified, dehydrated in a gradient of alcohols, paraffin embedded, sectioned, and stained with H&E as previously described.²² Images were taken on an Olympus BX53 microscopy at a magnification of ×100.

Statistical Analysis

In this study, data were expressed as the mean ± SEM and analyzed using Prism version 6.0 (GraphPad). Differences in measured variables between the experimental and control groups were assessed by using Student’s t tests. Multiple group comparisons were performed using a one-way ANOVA with Bonferroni’s multiple comparison test. Nonparametric data were analyzed by a Wilcoxon rank-sum test. Results were considered statistically significant at p < 0.05.

Study Approval

All animal experiments were approved by the Animal Facility Committee of the Model Animal Research Center (approval no. SYXK 2014-0003). All human participants in this study signed informed consent before enrolling in the study, and authorization from the Ethics Committees of the China-Japan Friendship Hospital and the Third Hospital of the Beijing Armed Police Corps was obtained for this study. The investigations were approved by the Institutional Review Boards and conducted in accordance with the ethical standards of the China-Japan Friendship Hospital and the Third Hospital of the Beijing Armed Police Corps. All experimental methods involving human subjects were completed in accordance with the relevant guidelines and regulations.

Author Contributions

W.N., D.J., and H.Z. conceived and designed the experiments. X.L., Y.F., Y.D., F.J., and Y.J. performed the experiments. S.Z., Y.L., J.G., C.Q., and C.Z. analyzed the data. J.L., C.Y., and H.Z. provided critical reagents, cell lines, and animal models for the study. W.N., D.J., and X.L. wrote the manuscript. C.W., H.D., B.W., D.J., and J.G. performed or supported the experiments involving analysis of samples from humans. All authors contributed to the discussion of the results and manuscript corrections.

Conflicts of Interest

The authors declare no competing interests.