Silica Perturbs Primary Cilia and Causes Myofibroblast Differentiation during Silicosis by Reduction of the KIF3A-Repressor GLI3 Complex

The purpose of this study was to determine the effects of Kinesin family member 3A (KIF3A) on primary cilia and myofibroblast differentiation during silicosis by regulating Sonic hedgehog (SHH) signalling. Methods: Changes in primary cilia during silicosis and myofibroblast differentiation were detected in silicotic patients, experimental silicotic rats, and a myofibroblast differentiation model induced by SiO2. We also explored the mechanisms underlying KIF3A regulation of Glioma-associated oncogene homologs (GLIs) involved in myofibroblast differentiation. Results: Primary cilia (marked by ARL13B and Ac-α-Tub) and ciliary-related proteins (IFT 88 and KIF3A) were increased initially and then decreased as silicosis progressed. Loss and shedding of primary cilia were also found during silicosis. Treatment of MRC-5 fibroblasts with silica and then transfection of KIF3A-siRNA blocked activation of SHH signalling, but increased GLI2FL as a transcriptional activator of SRF, and reduced the inhibitory effect of GLI3R on ACTA2. Conclusion: Our findings indicate that primary cilia are markedly altered during silicosis and the loss of KIF3A may promote myofibroblast differentiation induced by SiO2.


Introduction
Silicosis characterised by the formation of silicotic nodules is the most serious occupational disease in China and caused by long-term inhalation of dust [1,2]. According to our previous study on downregulation of acetylated tubulin α (Ac-α-Tub) in silicotic rats [3], we speculated that primary cilia, which are microtubule-based cellular protrusions, may also exhibit some changes during the progression of silicosis.
The primary cilium is a solitary, non-motile organelle that acts as a mechanosensory and chemosensory antenna protruding from the plasma membrane of many mammalian cell types [4]. The characteristics of many acquired diseases include abnormalities in both the cilium structure and function. Excessive ciliogenesis is found in the hyperplastic alveolar epithelium and fibroblastic foci of idiopathic pulmonary fibrosis (IPF) [5] as well as cardiac fibrosis [6]. Moreover, primary cilium loss Ivyspring International Publisher contributes to a switch from reversible to irreversible kidney injury in unilateral ureteral occlusions (UUOs) [7]. Detection of the primary cilium length, primary cilium fragments, or ciliary proteins in urine can be useful to evaluate the severity of kidney diseases, at least in acute kidney injury (AKI) [8]. Disruption of cilia can induce epithelial-myofibroblast transition (EMyT) [9]. Emerging observations implicate dysfunctional primary cilia in fibrosis of many tissues and organs, and primary cilia appear to be necessary to initiate the transition and sustained activation of myofibroblasts [10].
Kinesin family member 3A (KIF3A) is a subunit of heterotrimeric kinesin-2 and acts as a microtubule plus end-directed motor protein that regulates the cilium length to maintain primary cilia for compensation upon alteration [11]. The Kif3a knockout mouse is an established non-orthologous mouse model of cystic kidney disease characterised by loss of primary cilia, rapid cyst growth, and fibrosis [12]. Moreover, tissue-specific loss of Kif3a in pancreatic cells results in severe pancreatic fibrosis [13]. However, the mechanisms underlying the influence of Kif3a knockout on fibrosis are still unclear.
Recently, we reported a major role of the Sonic hedgehog (SHH) pathway in silicosis [14]. SHH relieves the inhibitory effects of patched-1 (PTC1) on smoothened proteins (SMO), both of which are ciliary membrane-bound receptors, and initiates the signalling cascade via the Glioma-associated oncogene homolog (GLI) family (GLI1-3) of transcription factors [15]. GLI1 is a transcriptional activator and GLI1 is a target gene of SHH signalling. Therefore, its function is reinforced by a positive feedback loop through SHH pathway activation. GLI2 and GLI3 have both full-length forms (GLI FL ) as transcriptional activators and proteolytically processed forms as transcriptional repressors (GLI R ). GLI3 R performs its function exclusively as a strong repressor of SHH pathway target genes [16,17]. GLI proteins interact selectively and synergistically with KIF3A [18]. In addition, kif3a -/mouse embryonic fibroblasts (MEFs) express a significantly lower level of GLI3 R [19].
In the present study, we found loss of primary cilia and shedding of ciliary proteins accompanied by silicosis progression, indicating that silica disrupts the structure and function of primary cilia in fibroblasts. We also found that silica activated SHH signalling to promote myofibroblast differentiation, and that initial primary cilium enlargement was followed by complete deciliation. Knockdown of KIF3A inhibited SHH signalling activation, but also increased CLI2 FL and decreased GLI3 R transcription to promote myofibroblast differentiation. Our findings indicate that primary cilia are markedly altered during silicosis and loss of KIF3A may promote myofibroblast differentiation.

Rat model
Animal studies were conducted with the protocol approved by the Institutional Animal Care and Use Committee of the North China University of Science and Technology, Tangshan, China (2013-038). Three-week-old male Sprague-Dawley (SD) rats with 180±10 g of body weight were purchased from Vital River Laboratory Animal Technology Co. Ltd. (SCXY 2009-0004, Beijing, China). They were fed and housed in the SPF-class laboratory at North China University of Science and Technology in accordance with the National Institutes of Health (NIH) guidelines.
The silicotic model was made using a HOPE MED 8050 exposure control apparatus (HOPE Industry and Trade Co. Ltd, Tianjin, China) as previous report [20]. Briefly, the rats were exposed to SiO2 (99% particle size of 0.5-10 μm, S5631, Sigma-Aldrich, St. Louis, MO, USA) for 3 h daily for 4, 12, and 24 weeks (n=10). The SiO 2 concentration is 50±10 µg/m 3 . The rats in the corresponding control group were exposed to the same condition of pure air without SiO 2 for 3 h daily for 4, 12, and 24 weeks (n=10).

Human lung tissue specimens and Bronchoalveolar lavage fluid (BALF)
The human study was approved by the Medical Ethics Committee of North China University of Science and Technology (2015-046). Written informed consent was obtained from each subject to confirm their voluntary participation in this study.
Slides of formalin-fixed, paraffin-embedded lung tissue sections of coal worker's pneumoconiosis (CWP) were obtained from the Department of Pathology of North China University of Science and Technology, Tangshan, China. The autopsies of silica-exposed workers were from 32 patients with an average of 19 years of occupational exposure history who were diagnosed with silicosis using the diagnostic criteria for occupational pneumoconiosis of China (GBZ 70-2015) and evaluated by occupational pulmonary pathologists. The pericarcinomatous tissue served as the negative control. The typical pathological changes in silicotic patients, including macrophage alveolitis, cellular silicotic nodules, and fibrous silicotic nodules, are shown in Figure S1. BALF samples were collected from patients who underwent massive whole lung lavage (instillation and recovery of 0.9% sterile saline in the bronchopulmonary segment) without clinical contraindications from China's Coal Miners Beidaihe Sanatorium. The study included 48 patients who were diagnosed with silicosis by the Occupational Diseases Committee, and those patients were divided into three grades, i.e. silicosis phase I (16 cases), II (16) and III (16), by diagnostic criteria for occupational pneumoconiosis of China (GBZ 70-2015). Stage 0+ workers (8 cases) were chosen as the control group. All of the subjects had no serious diseases in the heart, brain, liver, or kidneys, and those over 65 years of age were excluded. All of the participants were male. The recovered fluid was condensed via centrifuge at 600×g at room temperature (RT) for 15 min, and the supernatants were collected for detection. The demographic characteristics, occupational exposure, and pulmonary function tests of the enrolled subjects who underwent massive whole lung lavage are summarised in Table 1.

Cell culture and treatments
A human embryo lung fibroblast MRC-5 cell line was purchased from the Chinese Academy of Sciences Cell Library (TCHu150, Shanghai, China). The MRC-5 fibroblasts were treated with SiO 2 at a concentration of 50 μg/cm 2 . Culture medium was used as a vehicle control.

RNA interference
The MRC-5 fibroblasts were cultured in antibiotic-free medium and treated with lipofectamine 2000 reagent (11668, Invitrogen, Carlsbad, CA, USA) and a siRNA mixture for 8 h, followed by exchange medium, and maintained for 48 h.

Histopathological examination
Lung tissues were fixed, embedded, and transversed. Slices (4 μm) were stained with haematoxylin and eosin (Sigma-Aldrich, St. Louis, MO, USA), Sirius Red (Maxin Company, Fuzhou, China), and Masson's trichrome staining (Maxin Company, Fuzhou, China). Images were obtained using an Olympus DP80 light microscope (Olympus Corporation, Tokyo, Japan). The histopathological images were only for pathological observation, and the pathological figures were reviewed by two clinical pathologists.

Co-immunoprecipitation (CoIP)
The interaction of KIF3A with GLI2 and GLI3 was evaluated by CoIP. The cells were lysed in RIPA (R0020, Solarbio Life Science, Beijing, China) buffer containing 1% protease inhibitors. Then 30 µl of sepharose beads (FO115, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and cell lysates (2 g/L) were mixed to a volume of 400 µl and incubated for 2 h at 4 °C on a shaker for preclearing. The clear supernatant was incubated overnight with anti-KIF3A, anti-IgG antibody, and Protein A sepharose overnight at 4 °C. The beads were collected and washed 3 times with PBS before being boiled in 2× loading buffer at 95 °C for 5 min. Western blotting was used to analyse the CoIP results.

Chromatin immunoprecipitation (ChIP) assay
ChIP assays were performed following the indicated protocol of a Chromatin Immunoprecipitation Kit (17-371, EMD Millipore, Temecula, CA, USA). 1×10 7 cells were used for each ChIP enrichment. Briefly, MRC-5 fibroblasts were stimulated by SiO 2 for 12 h and treated with or without KIF3A-siRNA for another 24 h. The cells were then cross-linked with formaldehyde (with a final concentration of 1%) for 10 min at RT and then stopped by glycine (with a final concentration of 1.25 mol/L) for 5 min. Shearing cross-linked DNA to 200-1000 base pairs in length was performed by sonication. The sonicated chromatin was incubated with 2 μg of anti-GLI2 (DF7541, Affinity Biosciences, Cincinnati, OH, USA), positive control, and normal mouse IgG overnight with rotation at 4 °C, followed by elution of the protein/DNA complexes, reverse cross-linking, and DNA purification. ChIP-qPCR was performed and the results were calculated as a percentage of the input DNA. The qPCR reagent assembly for 1 reaction was as follows: ddH2O (9.5 ul, Corning, USA), Primer mix (1ul, Invitrogen), SYBR Green MasterMix (12.5 ul, Takara, Shiga, Japan). After initial denaturation for 10 min at 4 °C, they underwent denaturing 50 times at 94 °C for 20 s and annealing and extension at 60 °C for 1 min. The relative gene expression was calculated using the 2 -ΔΔCT method.

Statistical analyses
SPSS 22.0 software was used to perform the statistical analysis. All of the results were presented as median±standard deviation (SD). Analysis of variance (ANOVA) was performed to compare the outcomes of the averages of more than two groups, constant variance using the LSD test, and heterogeneity of variance using Tamhane's test. The independentsamples t-test was used to compare the means of the two samples. The chi-squared test was used to compare the qualitative data obtained in the analysed groups, P-values <0.05 were considered statistically significant. The correlations were analysed between the changes in KIF3A, IFT88, and Ac-α-Tub protein expression and lung function indices of silicosis.

SiO 2 induces collagen deposition and myofibroblast differentiation in vivo and in vitro
As described previously [20,22], silicotic lesions, interstitial fibrosis, and collagen deposition depend on the time of exposure to silica in rats ( Figure 1A). Accordingly, the levels of COL I and α-SMA were upregulated in silicotic rats in a time-dependent manner ( Figure 1B, C). In addition, SiO2-stimulated MRC-5 fibroblasts in vitro exhibited increases in the levels of COL I and α-SMA in a time-dependent manner ( Figure 1D, E).

Primary cilia are lost during SiO 2 -induced myofibroblast differentiation
In the present study, we used Ac-α-Tub as a cilium marker [6] to reveal changes of primary cilia during myofibroblast differentiation induced by SiO 2 . The length of primary cilia was increased gradually until 12 h of SiO 2 treatment compared with 0 h. However, after 48 h, most primary cilia were lost (Figure 2A-C). In addition, coexpression of Ac-α-Tub with other cilium markers (ARL13B, and γ-Tubulin [5]) indicated a decrease of primary cilia in MRC-5 fibroblasts treated with silica for 48 h ( Figure S2). The levels of Ac-α-Tub and ARL13B in MRC-5 fibroblasts also showed similar trends in western blots ( Figure  2D, E).
Primary cilia are lost in silicotic lesions of silica-exposed rats and silicotic patients IF staining showed a decrease of primary cilia (marked by anti-Ac-α-Tub green fluorescence) in silicotic lesions accompanied by an increase of α-SMA-positive cells ( Figure 3A). The frequency and length of primary cilia were decreased in silicotic nodules compared with alveoli walls (Figure 3B, C). This result was confirmed by Ac-α-Tub IHC staining ( Figure S3), costaining of γ-Tubulin and Ac-α-Tub ( Figure 3D), and western blot analysis of Ac-α-Tub and ARL13B in rats ( Figure 3E, F). We also found that primary cilia were consistently long in pericarcinomatous and perinodule lung tissues, and lost in cellular and fibrous silicotic nodules with rich collagen deposition and α-SMA IHC staining ( Figure  3G-J).
Next, we quantified cilium-related proteins including kinesin-2 motor KIF3A and anterograde axoplasmic cargo transporter IFT88 [23]. The levels of KIF3A and IFT88 were increased in rats exposed to silica for 4 weeks and decreased in the 24-week silicosis group compared with their control groups. Accordingly, expression of KIF3A and IFT88 in SiO2-induced myofibroblasts was significantly upregulated at 12 h and then decreased until 72 h ( Figure 4A-D). Taken together, these results showed that the emergence of a myofibroblast phenotype during silicosis was accompanied by a loss of primary cilia.

Shedding of ciliary proteins KIF3A and Ac-α-Tub induced by SiO 2
It has been reported that primary cilia shed into urine during AKI, and detection of primary cilia fragments or ciliary proteins in urine can be useful to evaluate the severity of kidney injury [8]. Therefore, we hypothesised that primary cilia may be shed into the alveolar cavity during silicosis. The levels of IFT88, KIF3A, and Ac-α-Tub were measured in BALF of silicotic patients. We found that Ac-α-Tub and KIF3A in BALF were significantly increased in silicotic patients in phase II and stage III, respectively. However, the level of IFT88 was not changed significantly ( Figure 4E-H). Furthermore, we found that the levels of Ac-α-Tub and KIF3A were increased in BALF of silicotic rats and culture supernatants of SiO2-stimulated MRC-5 fibroblasts ( Figure S4). The relationships between primary ciliary markers and lung function indices are listed in Table 1. The level of KIF3A was significantly negatively correlated with VC, FVC, and DLco (r=-0.302, P=0.031; r=-0.300, P=0.033; r=-0.311, P=0.026, respectively). These results indicated that the shedding of primary cilia may be induced by silica.

KIF3A knockdown accelerates myofibroblast differentiation induced by silica
Some recent studies have demonstrated that the role of primary cilia may be associated with a biphasic mechanism where the first phase of myofibroblast differentiation requires a fibrotic signalling pathway regulated by primary cilia and the second phase appears to be specific to cell types, which may involve loss or reduction in the length of primary cilia. These mechanisms result in sustained myofibroblast activation and subsequent fibrosis [5,10]. In line with a previous report [6], silencing of KIF3A before silica treatment inhibited myofibroblast differentiation ( Figure S5). Therefore, we first treated MRC-5 fibroblasts with SiO2 for 12 h and then silenced KIF3A for another 24 h. KIF3A silenced by specific siRNA resulted in loss of cilia and reduced ciliary proteins ( Figure 5A-D). Compared with NC-siRNA transfection, KIF3A-siRNA transfection did not change the levels of α-SMA or COL I expression in unstimulated MRC-5 fibroblasts. However, KIF3A-siRNA transfection after SiO2 stimulation significantly increased the expression of α-SMA, SRF, and MRTF-A ( Figure  5E-G). Similarly, knockdown of IFT88 in SiO 2 -stimulated MRC-5 fibroblasts significantly exacerbated myofibroblast differentiation induced by SiO 2 ( Figure S6). These data suggested that cilium loss promotes or facilitates myofibroblast differentiation.

KIF3A silencing disrupts SHH signalling
Activation of SHH signalling, which is necessary for myofibroblast differentiation and proliferation, is influenced by primary cilia [24,25]. In the present study, SiO 2 treatment increased the levels of SHH, SMO, and GLI1 accompanied by a reduced level of PTC1. KIF3A silencing decreased the levels of SHH, PTC1, SMO, and GLI1 after SiO 2 treatment for 24 h (Figure 6A-C). These results indicated that the ligands and receptors of SHH signalling, including SHH, PTC1, and SMO, were disrupted because of the loss of primary cilia induced by KIF3A silencing. In addition, the level of GLI1, the major transcription factor of SHH signalling, was decreased by silencing KIF3A.

KIF3A silencing promotes GLI2 FL to bind and activate SRF transcription
It has been reported that KIF3A interacts with GLI proteins [18]. We also found that GLI2 FL protein formed a complex with KIF3A, and the binding between GLI2 FL and KIF3A was increased by SiO 2 stimulation ( Figure 7A, B). GLI2 FL was increased with silicosis progression in vivo and in vitro ( Figure S7). Next, we used an online tool (http://rna.sysu.edu.cn) to identify putative GLI-binding sites (GBSs) in the genomic sequence adjacent to the transcription start site of the SRF gene to determine whether GLI2 binds to the SRF promoter and directly activates its transcription. We found a putative GBS (Figures 7C, D  and S8) within the -253 genomic regions relative to the 5′ initiation site of SRF. A ChIP assay showed that GLI2 bound to SRF ( Figure 7E). Taken together, these results identified SRF, a transcription factor of α-SMA [26], as a novel direct target gene of GLI2 FL . Furthermore, treatment with GANT61, an inhibitor of GLI2, markedly blocked the SiO 2 +KIF3A-siRNAinduced upregulation of GLI2 FL along with downregulation of SRF and α-SMA ( Figure 7F-H).

KIF3A knockdown reduces GLI3 R to decrease the inhibitive effect of GLI3 R on ACTA2
As described above, GLI3 exists as both a fulllength form (GLI FL ) as a transcriptional activator and a proteolytically processed form as a transcriptional repressor (GLI R ) that serves exclusively as a strong repressor of target genes [16,17]. First, we matched the transcriptional repressor GLI3 R to ACTA2 (α-SMA-encoding gene), and ChIP results showed that GLI3 bound to ACTA2, suggesting that GLI3 R directly inhibited α-SMA transcription ( Figures 8A-C  and S8). KIF3A motor protein is required for GLI3 FL proteolysis to GLI3 R [18]. The results of CoIP demonstrated that GLI3 FL and GLI3 R proteins formed a complex with KIF3A, and the binding was even high at 48 h of SiO 2 stimulation when the levels of GLI3 FL and GLI3 R were decreased significantly ( Figure 8D, E). KIF3A silencing decreased the levels of GLI3 R ( Figure 8J, K), especially in the nuclear fraction ( Figure 8F, G), indicating that the inhibitive effect of GLI3 R on ACTA2 could be reversed by reducing KIF3A. Furthermore, we found that KIF3A overexpression significantly inhibited the upregulation of α-SMA induced by SiO 2 , and the level of GLI3 R protein expression was upregulated correspondingly (Figure 8 J, K). However, the levels of GLI2 FL and SRF were not changed significantly in SiO 2 -stimulated KIF3A-cpDNA cells compared the SiO 2 -stimulated NC-cpDNA cells.

KIF3A stabilizes GLI3 R
Because SHH signalling was blocked by KIF3A silencing under SiO 2 treatment, we used SAG (a SMO agonist) to explore the role of KIF3A in GLI3 R under activated SHH signalling. As shown in Figure 9, KIF3A cpDNA increased the level of GLI3 R , whereas KIF3A-siRNA decreased the level of GLI3 R . CHX chase experiments showed that KIF3A-siRNA accelerated the degradation of GLI3 R induced by CHX ( Figure 9G, H). We also found that destabilisation of GLI3 R was not influenced by IFT88 silencing ( Figure  S5E, F).

Discussion
In the present study, we found a decrease of primary cilia during silicosis, which is different to previous studies [5,6]. In human IPF tissue samples, the distribution and increase of primary cilia are found in alveolar epithelial cells, fibroblasts, and endothelial cells compared with normal human lung tissue [5]. Ciliated fibroblasts are enriched in areas of myocardial injury [6]. In addition, we found that primary cilia and ciliary proteins increased at first and then decreased in rats and MRC-5 fibroblasts exposed to silica. However, the elongation of primary cilia is almost located in the margin of fibrotic lesions in IPF [5] or in ischemia/reperfusion models [6]. These morphological observations support our results in silicotic rats. Recently, a review of primary cilia in fibrosis may explain the difference, which indicated that the role of the primary cilium may be associated with a biphasic mechanism and has cell-, tissue-, and disease-specific mechanisms [10].   Most studies have recognised changes in the primary length, including elongation and shortening caused by the reabsorption or disruption of cilia, as adaptive responses to environmental stimuli [8]. In IPF [5], injured myocardium [6], kidneys [27,28], and pancreatic ducts [13,29], the cilium length increases significantly in response to cellular injury. However, primary cilium loss contributes to a switch from reversible to irreversible kidney injury in the unilateral ureteral occlusion (UUO) model [7]. Additionally, detection of the primary cilium length and fragments in urine can be useful to evaluate the severity of kidney diseases [8]. We also found that the levels of Ac-α-Tub and KIF3A were increased in BAFL and culture supernatants of SiO2-stimulated MRC-5 fibroblasts, indicating that organ fibrosis may be accompanied by primary cilium shedding.

Ethical approval
The animal experiment was reviewed and approved by the Institutional Animal Care and Use Committee at North China University of Science and Technology. The human experiment was approved by the Medical Ethics Committee of North China University of Science and Technology (2013-038, 2015-046). The study protocol received institutional review board approval, and written informed consent was obtained in all cases.

Transparency document
The transparency document associated with this article can be found in the online version.