Tartrate-resistant acid phosphatase 5 serves as a viable target against pulmonary brosis by modulating β-catenin signaling

Idiopathic pulmonary brosis (IPF) is a fatal interstitial lung disease with limited therapeutic options. Tartrate-resistant acid phosphatase 5 (ACP5) performs a variety of functions. However, its role in IPF remains unclear. Here, we demonstrated that the levels of ACP5 were increased in IPF patient samples and mice with bleomycin (BLM)-induced pulmonary brosis. In particular, higher levels of ACP5 were noted in the sera of IPF patients with a diffusing capacity of the lungs for carbon monoxide (DLCO) less than 40% of the predicted value. Additionally, Acp5 deciency protected mice from BLM-induced lung injury and brosis and reduced the differentiation and proliferation of broblasts. Mechanistic studies revealed that Acp5 was upregulated by TGF-β1 in a TGFβR1/Smad3-dependent manner, after which Acp5 dephosphorylated p-β-catenin at Ser33 and Thr41, inhibiting the degradation of β-catenin and subsequently enhancing β-catenin signaling in the nucleus, which promoted the differentiation and proliferation of broblast. Notably, the treatment of mice with BLM-induced brosis with Acp5 siRNA-loaded liposomes robustly reversed lung brosis. Collectively, these data indicate that Acp5 plays an essential role in the initiation and progression of pulmonary brosis; therefore, strategies aimed at silencing Acp5 could be novel therapeutic approaches against pulmonary brosis in a clinical setting. CCA


Introduction
Idiopathic pulmonary brosis (IPF) is a fatal interstitial lung disease characterized by the deposition of excessive extracellular matrix (ECM), destruction of the lung parenchyma and a pattern of usual interstitial pneumonia (UIP) by radiology and histopathology 1 . Although the development of new anti brotic agents (pirfenidone and nintedanib) has improved patient wellbeing, the incidence and mortality rate of IPF have barely improved, resulting in a median survival time following diagnosis of only 3 to 5 years 2,3 . Therefore, exploration of the pathogenesis of IPF and the discovery of new therapeutic methods for all IPF patients are urgently needed.
According to the current paradigm, the main pathological features of IPF include epithelial injury 4 , the recruitment of in ammatory cells 5 , the aberrant differentiation and proliferation of broblasts 6 and the persistence of apoptosis-resistant myo broblasts 7 in brotic lesions. Resident lung broblast-derived myo broblasts are the major contributors to the processes of ECM deposition and tissue distortion in IPF 8,9 . Under stimulation with brotic factors, such as transforming growth factor-β (TGF-β) 10 , plateletderived growth factor (PDGF) 11 and connective tissue growth factor (CTGF) 12 , resident broblasts in the lung lesion transform into myo broblasts, which are characterized by a spindle or stellate morphology with α-smooth muscle actin (α-SMA) stress bers coupled with a hypersecretion phenotype due to which they produce copious amounts of brillary ECM proteins, such as collagen and bronectin.
Tartrate-Resistant Acid Phosphatase 5 (ACP5), also named purple acid phosphatase, which maps to chromosome 19p13.2, contains a binuclear iron center 13 . Previous studies, including ours, have demonstrated that ACP5 plays a critical role in the pathogenesis of tumors, such as lung adenocarcinoma 14 , colorectal cancer 15 , breast cancer 16 and hepatocellular carcinoma 17 . Given that IPF and cancer share several pathogenic pathways 18,19 , we hypothesize that ACP5 is involved in the pathogenesis of IPF.
To assess the feasibility of this hypothesis, we rst detected the expression of ACP5 in IPF patients and mice with bleomycin (BLM)-induced pulmonary brosis. Notably, the levels of ACP5 in the sera of IPF patients were positively correlated with disease severity. Furthermore, the loss of Acp5 signi cantly protected mice from BLM-induced lung injury and brosis, accompanied by a marked reduction in myo broblast accumulation in the lung. Mechanistically, Acp5, a kind of phosphatase, selectively bound and dephosphorylated β-catenin at Ser33 and Thr41 in the cytoplasm and then reduced the degradation of β-catenin, which enhanced the levels of β-catenin in the nucleus to promote broblast differentiation, proliferation and migration. Therefore, intratracheal administration of liposomes carrying Acp5 siRNA signi cantly protected mice from BLM-induced pulmonary brosis. Collectively, our data support the notion that ACP5 plays a critical role in the progression of pulmonary brosis; therefore, strategies aimed at silencing ACP5 could be viable therapies against pulmonary brosis in clinical settings.

Material And Methods
Human samples Sera from patients with IPF (n = 20) and control subjects (n = 13), lung explant material from IPF patients (n = 3) and resected para-carcinoma lung tissues from non-small-cell lung cancer (NSCLC) patients (n = 3) were collected at Tongji Hospital with informed consent. An IPF diagnosis was made according to consensus diagnostic criteria from the European Respiratory Society (ERS)/American Thoracic Society (ATS) 20 . The experiments were approved by the Human Assurance Committee of Tongji Hospital. Clinical data from IPF patients and control subjects are provided in table 1.

Measurement of Serum ACP5 Levels
To detect ACP5 in the sera of the IPF patients and control subjects, an ACP5 ELISA kit (Hycult Biotechnology, Uden, The Netherlands) was used in accordance with the manufacturer's protocol. Brie y, whole-blood samples originating from IPF patients and control subjects were centrifuged at 3000 rpm for 15 minutes. Then, the serum samples were frozen at -80°C. Each sample was assayed in duplicate. ELISA plates were scanned on a microtiter plate reader (ELx800, BioTek Instruments, Inc., Winooski, VT) at 450 nm. Levels of ACP5 were calculated according to a standard curve.

BLM-mediated induction of pulmonary brosis
Pulmonary brosis was induced in male WT and Acp5 -/mice (8-10 weeks old) as previously described with minor modi cations 22 . Brie y, the mice were anesthetized with 1% pentobarbital sodium (60 mg/kg) and then intratracheally administered 2 U/kg BLM (Huirui, Shanghai, China) in 40 μl of sterile saline. Mice administered the same volume of sterile saline served as controls. SiRNA-loaded liposomes were injected into the anesthetized animals via intratracheal injection on days 14 and 18 after BLM injection. The mice were euthanized on day 21 following BLM challenge to analyze pulmonary brosis.

Pathological staining and histopathologic assessment
The lungs were removed at 21 days after bleomycin or saline administration. Following xation, the lungs were embedded in para n and sectioned. The sections were then stained with hematoxylin and eosin (H&E), Masson's trichrome stain and Sirius red as previously described 23 . Fibrosis was scored on a scale of 0-8 using the Ashcroft scoring method 24 . The severity of brotic changes in each histological section of the lung was assessed as the mean of the severity scores from the observed microscopic elds. Fifty randomly chosen regions from each mouse lung were graded, after which their scores were averaged, and the average scores are depicted in a graph at 200-fold magni cation. Grading was performed in a blinded manner by three independent observers.

Hydroxyproline assay
Lung collagen deposition was assessed by measuring the hydroxyproline content of lung homogenates with a hydroxyproline assay kit (BioVision, CA, USA) as previously described 25 . Brie y, lung tissues were homogenized in ddH 2 O, after which an equal volume of concentrated 12 N HCl was added to the tissues in a pressure-tight, capped Te on vial and hydrolyzed at 120°C for 3 h. After centrifugation at 14000 rpm for 3 min, 10 μl of each hydrolyzed sample was transferred to a 96-well plate. In each well, 100 μl of chloramine T reagent was added to the sample, which was incubated at room temperature for 5 min, after which 100 μl of DMAB reagent was added. After incubation for 90 min at 60°C, the plates were read at 560 nm with a microtiter plate reader (ELx800, BioTek Instruments, Inc., Winooski, VT), and the hydroxyproline concentration in the sample was calculated from a standard curve and related to the amount of lung tissue used. The hydroxyproline contents in the lung tissues are given as μg of hydroxyproline per mg of lung tissue.
Cell culture and treatment Primary human pulmonary broblasts (PHLFs) were isolated from para-carcinoma lung tissues resected from NSCLC patients, and primary mouse pulmonary broblasts (PMLFs) were isolated from the lung tissues of Acp5 -/or WT mice as described previously 24,26 . Brie y, broblasts were generated by mincing lung tissue into submillimeter-sized pieces, plated evenly in 100 mm plates containing 2 ml of medium, which was changed after 24 hours. Cells were cultured in DMEM containing 10% fetal bovine serum (FBS) and penicillin/streptomycin at 37°C and tested for mycoplasma regularly. PHLFs and PMLFs between passages 3 and 5 were used.

Cell transfection
Small interfering RNAs (siRNAs) speci c for ACP5 and a corresponding scrambled siRNA were purchased from RiboBio (Guangzhou, China) and then transiently transfected into PHLFs using Lipofectamine 3000 (Invitrogen, Shanghai, China) as previously reported 27 . Brie y, PHLFs were seeded in 6-well or 96-well plates at 24 h before transfection. SiRNA transfection was performed when the cells reached 50-60% con uence. Transfection e ciency was monitored by RT-PCR or Western blotting at 48 h after transfection. The siRNA speci c for ACP5 targeted the following sequence in ACP5 mRNA: 5′-GAC ACT ATG TGG CAA CTC A-3′.
ACP5-plasmid and a vector were purchased from GeneChem (Shanghai, China), and 2 μg of puri ed DNA was mixed with transfection reagent and applied to the cells. Forty-eight hours after transfection, the cells were analyzed by Western blotting.

Cell proliferation assay
PMLFs were cultured in 96-well plates at a density of 2×10 3 cells/well. Cell proliferation was then measured using the EdU proliferation assay (Ribobio, Guangzhou, China) as previously reported 21 . Brie y, 18 h after being seeded in the plates, cells were labeled with EdU for 2 h at 37°C, treated with 100 µL of Apollo reaction cocktail and stained with 100 µL of Hoechst 33342. Finally, the cells were observed under a uorescence microscope (Olympus, Shinjuku, Japan).

Cell migration assay
Cell migration assays were performed using Transwell inserts with a membrane with a pore size of 8.0 mm (Corning, MA, USA) according to previously reported methods 30 . The cells (2.5×10 4 ) were resuspended in 2% serum-containing medium and seeded into the upper chambers. The lower chambers were lled with complete culture medium containing 10% FBS to function as a chemoattractant. After incubation at 37°C for 24 h, the cells migrated through the membrane lter and were stained with 0.1% crystal violet (Sigma-Aldrich, St. Louis, MO, USA). The experiment was performed in triplicate.

Coimmunoprecipitation assay
Immunoprecipitation was performed according to a previous protocol 14 . Anti-ACP5 and anti-β-Catenin antibodies were used to form immune complexes with the ACP5 and β-Catenin proteins in lysates that were immunoprecipitated with magnetic beads (Cell Signaling Technology, Danvers, MA, USA). Finally, equivalent protein samples were subjected to Western blot analysis for ACP5 and β-Catenin. Immunohistochemistry Immunohistochemistry (IHC) was performed on 3 μm adjacent lung sections according to the protocol in a previous report 31 . The primary antibodies used were anti-Pcna (Cell Signaling Technology, MA, USA, 1:100) and anti-Fsp1 (Proteintech, Wuhan, China, 1:100). The primary antibodies were incubated with the sections overnight at 4 °C, followed by incubation with a secondary antibody for 1 h at room temperature. Finally, the sections were stained with DAB reagent for 2 min at room temperature, and the nuclei were then counterstained with hematoxylin. Stained cells were observed under an optical photographic light microscope at ×400 magni cation.

Nanoparticles
SiRNA-loaded liposomes were prepared as reported 32 . A lipid solution in which the lipidoid, cholesterol, DSPC and mPEG-DMG at a molar ratio of 50:38.5:10:1.5 were dissolved in a solution of 90% ethanol and 10 mM sodium citrate was prepared. Then, siRNA was dissolved in 10 mM citrate buffer, and the lipid components were mixed with the dissolved siRNA by vortexing such that the nal weight ratio of lipidoid:siRNA was 5:1. The next step was ultra ltration centrifugation to exclude free siRNA. Finally, the siRNA-liposomes were diluted in PBS. The hydrodynamic diameter, polydispersity, zeta potential and stability of the liposomes were measured by dynamic light scattering (DLS) (Malvern Zetasizer Nano-ZS, UK). A RiboGreen assay was employed to calculate the siRNA entrapment e ciency. After staining with 2% phosphotungstic acid, the liposomes were characterized by transmission electron microscopy (TEM, Jeol, Japan).

Statistical analyses
All statistical analyses were performed using GraphPad Prism (San Diego, CA, USA) or SPSS 25.0 (IBM, Armonk, NY, USA). Correlations between ACP5 expression and the clinical pathological features of IPF patients and control subjects were analyzed by the χ2 test or Fisher's exact test. Other data are expressed as the mean ± SEM, and an independent Student's t-test was administered to analyze the statistical signi cance of differences between two groups. p<0.05 was used to indicate statistical signi cance. All data were tested for normalization before analysis.

IPF is characterized by altered ACP5 expression
We rst sought to examine the levels of ACP5 in the sera of IPF patients and control subjects.
Interestingly, ACP5 levels in the serum derived from control subjects were low, and higher levels of ACP5 were detected in the serum samples from IPF patients ( Figure 1A). Then, we assessed the correlation between the levels of ACP5 and disease severity, which were estimated by the lung's gas diffusing capacity 33 (as measured by the diffusing capacity of the lungs for carbon monoxide, DLCO). Notably, the levels of ACP5 were higher in severe IPF patients whose DLCO was less than 40% of the predicted value than in patients with less severe IPF ( Figure 1B). To further address the role of ACP5 in pulmonary brosis, we then detected the expression of ACP5 in the lungs of IPF patients and mice with the onset of BLM-induced pulmonary brosis. Interestingly, the lung homogenates of IPF patients exhibited 6-fold higher ACP5 expression than those of control subjects, and markedly higher expression levels of COL1A1 and α-SMA, markers of brosis, were also noted in IPF patients ( Figure 1C). Similarly, signi cantly higher Acp5 expression was detected in the lungs of mice following BLM injection than in those of saline-treated mice ( Figure 1D). Consistent results were also obtained by RT-PCR analysis of Acp5 expression ( Figure  1E). Next, we sought to examine cells showing altered ACP5 expression in the lung sections from IPF patients and control subjects. Coimmunostaining showed that ACP5 was almost undetectable in the lung sections from control subjects, while IPF patient-derived lung sections were characterized by a large amount of lung broblast aggregation, manifesting as high levels of ACP5 revealed by costaining for ACP5 with broblast speci c protein 1 (FSP1), a broblast marker ( Figure 1F). We further detected lung sections from mice after BLM injection and found that the progression of pulmonary brosis was highly correlated with the severity of broblast accumulation and Acp5 overexpression in the lungs ( Figure 1G).
Acp5 is upregulated in a canonical TGF-β signaling-dependent manner To further con rm the above results, we cultured primary lung broblasts from WT mice and control individuals and then stimulated them with TGF-β1. Indeed, low levels of ACP5 were observed in broblasts before stimulation. Upon TGF-β1 stimulation, a substantial increase in ACP5 levels in primary mouse pulmonary broblasts (PMLFs) (Figure 2A) and primary human pulmonary broblasts (PHLFs) ( Figure 2B) was noted. Given that TGF-β/Smad3 signaling plays a pivotal role in broblast differentiation induced by TGF-β1 34 , we conducted coimmunostaining for Acp5 and p-Smad3 in PMLFs after 1 h of TGF-β1 treatment. Interestingly, Acp5 colocalized with p-smad3 in cultured broblasts ( Figure 2C). These ndings suggested to us that Acp5 may be a downstream target of canonical TGF-β signaling. To assess this hypothesis, we treated PMLFs with TGF-β1 and SB431542, a speci c inhibitor of TGF-β type I receptor (TGFβRI). As expected, the expression of Fibronectin, Col1a1 and α-SMA was signi cantly enhanced following TGF-β1 treatment ( Figure 2D). However, inhibition of TGFβRI activation by SB431542 absolutely abolished the upregulation of these brotic proteins. Surprisingly, the overexpression of Acp5 induced by TGF-β1 was reversed after SB431542 treatment ( Figure 2D). Interestingly, we observed similar results when PMLFs were stimulated with TGF-β1 and SIS3-HCl (a speci c inhibitor of smad3 phosphorylation) ( Figure 2E). Collectively, these data support the notion that Acp5 expression in broblasts is controlled by canonical TGF-β signaling, of which TGFβR1/Smad3 are essential mediators.

ACP5 is essential for broblast differentiation, proliferation and migration
To assess the functional role of ACP5 in lung broblasts, PMLFs were generated from WT and Acp5 -/mice and then subjected to TGF-β1 stimulation. Compared with broblast differentiation in WT PMLFs, the loss of Acp5 signi cantly attenuated the differentiation of broblasts into myo broblasts, as evidenced by the decreased expression of myo broblast markers ( bronectin, Col1a1 and α-SMA) following TGF-β1 treatment ( Figure 3A). Consistently, RT-PCR analysis of the expression of these genes demonstrated similar results ( Figure 3B). To further con rm this result, we then performed ACP5 gainand loss-of-function assays in PHLFs following TGF-β1 treatment. As expected, the expression of ACP5 was e ciently silenced or enhanced following ACP5 siRNA or ACP5 plasmid transfection ( Supplementary  gure 2A and B). Notably, both Western blot and RT-PCR analyses demonstrated that broblast differentiation to myo broblast was abrogated in ACP5 siRNA-transfected PHLFs after TGF-β1 induction ( Figure 3B), while a signi cant increase in myo broblast markers was observed in ACP5-overexpressing PHLFs (Supplementary gure 3). Additionally, we estimated the impact of ACP5 on the proliferation of PMLFs and PHLFs by EdU staining and CFSE staining, respectively. Fewer EdU-positive cells was noted among Acp5 -/-PMLFs than among WT PMLFs ( Figure 3E). Consistently, the silencing of ACP5 expression remarkably reduced the proliferation of broblasts induced by TGF-β1 ( Figure 3F). Similar data were also detected in PMLFs and PHLFs by CFSE staining (Supplementary gure 4A and B). Furthermore, we next examined the migration of broblasts by Transwell assay. Notably, the loss of ACP5 signi cantly suppressed the migration of PMLFs and PHLFs across the Transwell membrane ( Figure 3G, H).
ACP5 interacts with β-catenin and regulates the degradation of β-catenin β-Catenin signaling is known to be critical in the process of pulmonary brosis 35 . Therefore, we investigated the impact of Acp5 on β-catenin signaling in TGF-β1-stimulated broblasts. As expected, high levels of β-catenin were detected in WT PMLFs after 24 h of TGF-β1 treatment. However, the protein but not mRNA levels of β-catenin were much lower in Acp5 -/-PMLFs, indicating that of the decrease in βcatenin levels was not due to decreased gene transcription ( Figure 4A and Supplementary gure 5A).
Consistently, the knockdown of ACP5 expression in PHLFs also revealed lower levels of β-catenin following TGF-β1 induction compared to those in the group transfected with scrambled siRNA. To con rm this result, the expression of ACP5 was enhanced in Acp5 -/-PMLFs and PHLFs by plasmid transfection. In contrast, a remarkable increase in β-catenin was observed in both PMLFs and PHLFs after ACP5 plasmid transduction followed by TGF-β1 treatment ( Figure 4C and D). Nevertheless, in this experiment, the mRNA levels of β-catenin were also not affected by ACP5 overexpression.
Given that β-catenin translocation into the nucleus is critical for its function 36 , we next estimated the localization of Acp5 and β-catenin in PHLFs. Immuno uorescence assays revealed that both Acp5 and βcatenin were mainly localized to the cytoplasm before treatment, while β-catenin translocation into the nucleus upon TGF-β1 stimulation was noted, and high expression of Acp5 was noted in the cytoplasm.
However, much lower levels of β-catenin were detected in Acp5 -/-PMLFs than in WT PMLFs ( Figure 4E). To further quantify the subcellular transfer of β-catenin, we used a Western blot assay. Indeed, signi cantly higher levels of β-catenin were detected in cytoplasmic and nuclear proteins from WT PMLFs than in those from Acp5 -/-PMLFs ( Figure 4F).
As compelling evidence suggests that Acp5 is a phosphatase, we hypothesized that Acp5 dephosphorylates some sites on phosphorylated β-catenin, such as the sites Ser33, Ser37 and Thr41, which are involved in the process of β-catenin degradation 37 . Interestingly, Acp5 could interact with βcatenin in PMLFs ( Figure 5A and B) and PHLFs ( Figure 5C and D), as shown by co-IP assay. Furthermore, we discovered higher levels of β-catenin phosphorylated at Ser33, Ser37 and Thr41 in Acp5 -/-PMLFs ( Figure 5E) and ACP5-silenced PHLFs ( Figure 5F) than in WT PMLFs and scrambled RNA-transfected PHLFs, respectively. In contrast, the opposite was observed in ACP5-overexpressing broblasts ( Figure 5G and H), indicating that Acp5 appears to dephosphorylate the Ser33, Ser37 and Thr41 sites in p-β-catenin, by which it inhibits the process of degradation. To further con rm this result, we constructed 4 plasmids containing mutated β-catenin (MU1, MU2, MU3 and MU4) ( Figure 5I) and transfected the mutant βcatenin plasmids and Acp5 plasmid into PMLFs, followed by TGF-β1 stimulation. As expected, low levels of p-β-catenin (33/37/41) and high levels of β-catenin were noted in PMLFs transfected with the MU4-βcatenin plasmid. Furthermore, these changes were also observed in PMLFs transfected with the other MU plasmids except for MU2 ( Figure 5J), indicating that Acp5 dephosphorylated the Ser33 and Thr41 sites on β-catenin and inhibited the degradation of β-catenin.

Global deletion of Acp5 protected mice from BLM-induced lung injury and brosis
To assess the requirement of Acp5 in the development of lung brosis, Acp5 -/and WT mice were subjected to BLM treatment (intratracheal injection), and lung injury and brosis were assessed after 21 days of induction. Hematoxylin and eosin (H&E), Masson's trichrome and Sirius red staining of the lung sections from mice in the BLM group demonstrated a remarkable increase in lung parenchymal brotic lesions compared to those in the saline group. Notably, compared with Acp5 -/mice, BLM-treated WT mice were more susceptible to bleomycin toxicity, as evidenced by higher Ashcroft scores ( Figure 6A), increased levels of hydroxyproline ( Figure 6B) and lower survival rates ( Figure 6C). To further quantitatively determine whether the loss of Acp5 could inhibit brotic marker expression and ECM production in the lungs of BLM-treated mice, we evaluated the protein and mRNA expression levels of brotic genes by Western blot and RT-PCR, respectively. As illustrated in Figure 6D and E, the protein and mRNA levels of bronectin, collagen, and α-SMA after BLM treatment were signi cantly lower in the Acp5 -/mice than in the WT mice. Additionally, Acp5 de ciency remarkably attenuated the transition of broblasts to myo broblasts, as evidenced by decreased uorescence intensity for Fsp1 + /α-SMA + cells in the brotic lesion ( Figure 6F). Consistent with the effect of ACP5 on broblast proliferation in vitro, staining for PCNA and Fsp1 in the adjacent lung sections showed the decreased ability of broblasts in Acp5 -/mice to proliferate (Supplementary gure 6). Furthermore, the levels of β-catenin and p-β-catenin were also detected in WT and Acp5 -/mice following BLM induction. Consistently, Western blot analysis demonstrated that the levels of β-catenin were increased and the levels of p-β-catenin were decreased in the lung homogenates derived from the BLM-induced mice compared to uninduced mice ( Figure 6G), and loss of Acp5 largely reversed changes in the levels of β-catenin and p-β-catenin induced by BLM ( Figure  6G). Collectively, our data demonstrate that the loss of Acp5 protected mice against BLM-induced lung injury and brosis.
In vivo treatment with nanoparticles carrying Acp5 siRNA evoked an anti brotic response in BLM-induced pulmonary brosis Finally, we sought to transform the above discoveries into a therapeutic approach to remedy pulmonary brosis. First, we veri ed the knockdown e ciency of three siRNAs in PMLFs ( Figure 7A and B) and chose the most e cient siRNA (#3) to generate lipid-based nanoparticles loaded with Acp5 siRNA ( Figure  7C). The prepared liposomes demonstrated a >90% entrapment e ciency for loading siRNA with a zeta potential of 4.1 mV (Supplementary gure 7A). Additionally, as observed in the transmission electron microscopy (TEM) images (Supplementary gure 7B), the prepared liposomes showed an average diameter of ~100 nm and a uniform sphere morphology. Furthermore, those liposomes illustrated a normal hydrodynamic diameter distribution (Supplementary gure 7C) and were continuously stable over 24 hours (Supplementary gure 7D). To demonstrate the distribution of liposomes in the lung following BLM induction, DiR-labeled liposomes were used. Interestingly, most of the broblasts overlapped the liposomes, which were mainly located in the brotic area, revealing the highly e cient absorption of broblasts ( Figure 7D). Then, the temporal expression of Acp5 was assessed after delivery of the liposomes. Notably, a signi cant decline in Acp5 was noted, and the lowest Acp5 expression was detected at day 3 after the intratracheal delivery of Acp5 siRNA-loaded liposomes; however, Acp5 expression was restored to normal levels at day 5 ( Figure 7E).
Ultimately, the WT mice were administered scrambled siRNA-or Acp5 siRNA-loaded liposomes on day 14 and day 18, respectively, by intratracheal instillation (100 nmol/kg). Indeed, the therapeutic effects of siRNA-loaded liposomes were validated in BLM-induced mice, as manifested by the results of histopathological analysis ( Figure 7F), brotic scores ( Figure 7F) and decreased levels of hydroxyproline in the lung ( Figure 7G). Consistently, mice administered Acp5 siRNA-loaded liposomes exhibited marked attenuation of brotic markers ( bronectin, Col1a1, α-SMA) at the protein level ( Figure 7H) and mRNA level ( Figure 7I). Collectively, our experiments reveal the clinical potential of intratracheal Acp5 siRNAloaded liposome administration in the treatment of pulmonary brosis.

Discussion
As outcomes in pulmonary brosis, especially IPF, remain poor despite the emergence of some new anti brotic agents, an improved understanding of factors that in uence gene expression and broblast differentiation is required. Although ACP5 has extensive pathophysiological functions depending on the cell type and disease 38,39 , to the best of our knowledge, our study provides for the rst direct experimental proof that ACP5 plays a crucial role in the pathogenesis of brosis. We detected a marked increase in the levels of ACP5 within human IPF samples compared to samples from control subjects. Remarkably, higher levels of ACP5 were noted in the sera of IPF patients with lower DLCO values. Furthermore, Acp5 de ciency protected mice from BLM-induced lung injury and brosis, with decreases in the differentiation of broblasts to myo broblasts and proliferation. Mechanistic experiments revealed that Acp5 was upregulated by TGF-β1 in a TGFβR1/smad3-dependent manner, after which Acp5, a phosphatase, speci cally bound p-β-catenin and dephosphorylated the sites Ser33 and Thr41, by which it inhibited the degradation of β-catenin and enhanced β-catenin signaling in the nucleus to promote the differentiation and proliferation of broblasts ( Figure 8). Notably, the treatment of brotic mice with Acp5 siRNA-loaded liposomes reversed pulmonary brosis in the mice. Taken together, our results not only provide novel insights into the understanding of IPF pathogenesis but also strongly suggest that strategies aimed at silencing ACP5 could be viable approaches for the treatment of pulmonary brosis in clinical settings.
Previous studies, including ours, have demonstrated that ACP5 is upregulated in the progression of carcinoma and regulates tumor cell proliferation, migration and apoptosis [14][15][16][17] . As pathways and risk factors are often shared by cancer and IPF, we detected the levels of ACP5 in sera and lung samples from IPF patients. Interestingly, high levels of ACP5 were noted in sera and lung sections from IPF patients. The most exciting discovery in this report is that the serum concentration of ACP5 was associated with decreased lung function in IPF patients. Speci cally, higher levels of ACP5 were observed in patients with severe IPF (DLCO<40% predicted), indicating that ACP5 may be a surrogate biochemical marker for the severity of IPF. Similar to previous data 40 , serum ACP5 is a useful marker of bone resorption and bears clinical applicability in the diagnosis and management of metabolic and pathologic bone diseases, such as bone metastasis.
We next focused on assessing cells that show altered ACP5 expression in lung sections from IPF patients. Interestingly, we illustrated that ACP5 was predominantly localized within FSP1 + broblasts in IPF; similarly, Acp5 was noted to be overexpressed in broblasts derived from mice with BLM-induced brosis. Additionally, the expression of Acp5 following TGF-β1 treatment was detected in PMLFs and PHLFs. Indeed, Acp5 was upregulated in myo broblasts induced with TGF-β1. Furthermore, we showed that Acp5 was expressed in myo broblasts in a TGFβR1-and Smad3-dependent manner, as manifested by the detection of decreased levels of Acp5 in myo broblasts treated with SB431542 or SIS3-HCL followed by TGF-β1 induction. Myo broblasts are critical in the process of pulmonary brosis by their secretion of ECM proteins, leading to tissue stiffness and respiratory failure 41 . Recent study has demonstrated that myo broblasts in brotic lungs are mainly derived from resident lung broblasts 9 . This discovery prompted us to assess the impact of Acp5 on broblasts following TGF-β1 treatment.
Indeed, Acp5 de ciency and silencing signi cantly abolished the differentiation, proliferation and migration of PMLFs and PHLFs, respectively. As a result, mice de cient in Acp5 exhibited a signi cant reduction in myo broblast differentiation from broblasts and broblast proliferation in the lung following BLM induction.
The next important issue addressed was how Acp5 expression promotes the differentiation, proliferation and migration of broblasts. Previous studies have illustrated that aberrant β-catenin activation is involved in the pathogenesis of IPF 42,43 . In addition, β-catenin signaling is required for TGF-β-induced broblast differentiation, proliferation and migration 44,45 . We therefore examined the effects of Acp5 on β-catenin signaling in broblasts. Indeed, the loss or knockdown of Acp5 robustly decreased the levels of β-catenin in the cytoplasm and nucleus. Consistently, more β-catenin was observed in Acp5 plasmidtransfected broblasts than in control broblasts. It is likely that the impact of Acp5 on β-catenin is not regulated at the transcriptional level, as β-catenin mRNA was not altered in Acp5-knockdown or Acp5overexpression broblasts. A previous study showed that cytoplasmic β-catenin in the destruction complex is phosphorylated at residues Ser33, Ser37 and Thr41 in the absence of Wnt ligands; phosphorylated β-catenin is recognized by the E3 ubiquitin ligase β-transducin repeat-containing protein (β-TrCP) and subsequently degraded by the ubiquitin-dependent proteasome pathway 46 . Indeed, high levels of p-β-catenin were observed in Acp5-de cient or Acp5-silenced broblasts. In particular, further data showed that Acp5, a phosphatase, bound p-β-catenin and then dephosphorylated the Ser33 and Thr41 sites, subsequently inhibiting the degradation of β-catenin.
Given that until now, no effective therapies to halt the progression of IPF have been available, we assessed the e cacy of targeting Acp5 in mice with brosis induced by BLM. Previous studies have established liposomes as drug carriers for inhalation owing to their safety and ability to provide controlled drug release in the lung 32 . Similarly, in our study, Acp5 siRNA-loaded liposomes were e ciently taken up by broblasts in lung lesions after intratracheal injection. Notably, liposomes carrying Acp5 siRNA signi cantly attenuated lung brosis during the " brotic" phase of the model, which is more applicable to the clinical management of IPF patients and re ective of IPF.
Our study has some limitations. First, although we demonstrated that Acp5 speci cally bound p-β-catenin and dephosphorylated the sites Ser33 and Thr41, until now, no commercial antibodies for p-β-catenin (Ser33), p-β-catenin (Thr41) or p-β-catenin (Ser33 and Thr41), except for commercial antibodies for p-βcatenin (Ser33, Ser37 and Thr41), have been available. Second, the use of a speci c Acp5-knockout mouse model would make our data more compelling. Third, in our study, we illustrated that Acp5 was upregulated by TGF-β1 in a TGFβR1/Smad3-dependent manner, but the detailed mechanism needs to be further explored.
In conclusion, this report demonstrates that ACP5 plays a crucial role in the progression of pulmonary brosis. High serum levels of ACP5 were signi cantly associated with the severity of IPF in patients. In addition, mice with Acp5 de ciency were protected from BLM-induced lung injury and brosis. Mechanistic experiments showed that Acp5 selectively dephosphorylates p-β-catenin at Ser33 and Thr41 in the cytoplasm, reducing the degradation of β-catenin, by which the levels of β-catenin in the nucleus are enhanced, promoting broblast differentiation, proliferation and migration. Together, our data indicate that targeting Acp5 may represent a promising therapeutic approach for the treatment of pulmonary brosis in clinical settings.   Representative results for EdU staining in WT or Acp5-/-PMLFs (E) and ACP5 siRNA or Scrambled siRNA treated PHLFs (F). G-H: Representative results for Transwell assay in WT or Acp5-/-mice derived PMLFs (G) and ACP5 siRNA or Scrambled siRNA treated PHLFs (H). The data are represented as the mean ± SEM of three independent experiments. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

Figure 4
Altered Acp5 expression affects the levels of β-catenin. A-D: Western blot analysis of the levels of βcatenin in WT or Acp5-/-PMLFs (A), ACP5 siRNA or Scrambled siRNA treated PHLFs (B), Acp5 plasmid or Vector treated Acp5-/-PMLFs (C) and ACP5 plasmid or Vector treated PHLFs (D). E: Representative results for co-immunostaining of Acp5 and β-catenin in PMLFs from WT and Acp5-/-PMLFs following TGF-β1 induction. The nuclei were stained blue by DAPI, and the images were taken under original magni cation ×400. F Western blot analysis of the levels of β-catenin in cytoplasm and nuclear. The data are represented as the mean ± SEM of three independent experiments. *, p < 0.05; **, p < 0.01.  Comparison of the severity of lung brosis between WT and Acp5-/-mice after BLM induction. A: Histological analysis of the severity of lung brosis in mice after BLM induction. Left panel: representative images for H&E (top), Masson staining (middle) and Sirius red (bottom). Right panel: A bar graph showed the quantitative mean score of the severity of brosis. Images were captured at ×200 magni cation. B: Quanti cation of hydroxyproline contents in WT and Acp5-/-mice. C: the survival ratio in WT and Acp5-/-mice after BLM induction. D-E: Western blot (D) and RT-PCR (E) analysis of bronectin, Col1a1 and α-SMA expression. F: Co-immunostaining of Fsp1 and α-SMA in the lung sections. The nuclei were stained blue by DAPI, and the images were taken under original magni cation ×400. G: Western blot analysis of the levels of β-catenin and p-β-catenin (S33, S37 and T41) in WT and Acp5-/-mice after BLM challenge. Nine mice were included in each study group. The data are represented as the mean ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001. The nuclei were stained blue by DAPI, and the images were taken under original magni cation ×400. E: Figure 8 Schematic illustration of the mechanisms of Acp5 in broblasts. Acp5 is upregulated by TGF-β1 in a TGFβR1/smad3 depend manner, and then Acp5 specially binds to p-β-catenin and dephosphorylate the sites of Ser33 and Thr41, by which it resists the degradation of β-catenin and enhanced β-catenin signaling in the nuclear to promote the differentiation, proliferation and migration of broblasts