The Hsp27-Mediated IkBα-NFκB Signaling Axis Promotes Radiation-Induced Lung Fibrosis

Radiotherapy is a mainstay of lung cancer treatment. However, delivery of high radiation doses to the tumor is often hampered by the risk of radiation (IR)-induced lung injury. Lung damage due to thoracic IR is mediated via acute responses including inflammation and pneumonitis as well as chronic effects such as pulmonary fibrosis (1, 2). Fibrosis is the end stage of persistent tissue damage and chronic inflammatory reactions. It is characterized by excessive accumulation of extracellular matrix (ECM) and disruption of normal tissue architecture (3, 4).

During epithelial–mesenchymal transition (EMT), cells undergo a morphologic switch from the epithelial polarized phenotype to the mesenchymal fibroblastoid phenotype. EMT is characterized by the loss of epithelial differentiation markers, and the induction of mesenchymal markers. EMT plays a key role in embryonic development, chronic inflammation, and fibrosis (5, 6). Furthermore, EMT was observed during tumor cell invasion and metastasis in various solid tumors, (7). The exact molecular mechanisms leading to the development of IR-induced pulmonary fibrosis have yet to be fully identified.

Hsp27 (Hsp27 in humans and Hsp25 in mice) is an ATP-independent molecular chaperone that is highly induced in response to cellular stresses (8). Hsp27 is a critical mediator in cancer progression, preventing apoptosis in transformed cells (9–11). In addition, Hsp27 enhances migration and invasion (12) and mediates EMT in cancer cells (13). It is also an inducer of EMT during fibrosis including idiopathic pulmonary fibrosis (IPF; ref. 14). Overexpression of Hsp27 was reported in patients diagnosed with IPF (15). The upregulation of Hsp27 plays a pivotal role in myofibroblast differentiation and may represent a promising therapeutic target in fibrotic diseases. Hsp27 gene silencing by OGX-427, a second-generation antisense oligonucleotide, inhibited development of bleomycin-induced lung fibrosis and EMT via degradation of Snail (14). Accordingly, Hsp27 inhibition is an attractive therapeutic strategy.

Three pharmacologic treatments for IPF, namely pirfenidone, nintedanib (BIBF1120), and NAC are commercially available. Pirfenidone has been shown to prevent the accumulation of hydroxyproline, procollagen I and III, inflammatory cells, and TGFβ1 in bronchoalveolar lavage (BAL), and/or lung tissue (16–22). Pirfenidone has also been shown to diminish the fibrocyte pool and the migration of these cells in mouse models of lung fibrosis (23). Nintedanib was discovered as a byproduct in large screening assays targeting cyclin-dependent kinase (CDK4; ref. 24). Nintedanib was systematically developed as a potent angiogenesis inhibitor. However, conclusive evidence is unavailable to support its clinical role in the treatment of pulmonary disease, especially in IR-induced lung fibrosis.

Previously, we developed a mouse model simulating clinical stereotactic body radiotherapy (SBRT) and validated the induction of lung fibrosis by high-dose IR (25) and this model did not show any difference in the incidence of pulmonary fibrosis according to the radiosensitive and radioresistant mouse strains (26). In this study, we identified the molecular targets in lung fibrosis development. Hsp27 expression was increased during IR-induced lung fibrosis, and functional inhibition of Hsp27 using a small-molecule ameliorated lung fibrosis. While investigating mechanisms of Hsp27 in the development of lung fibrosis, we found that IkBα–NFkB signaling activation by direct interaction of IkBα with Hsp27, is involved in the EMT process that is tightly connected to the development of IR-induced lung fibrosis.

All procedures were approved by the Animal Care and Use Committees of Yonsei University Medical School (Seoul, Korea; 2015–0267) and were performed in accordance with the relevant guidelines. A single dose of 75 or 90 Gy was delivered using an X-RAD 320 platform (Precision X-ray) as described previously (27). Mice were administered intraperitoneally with J2 (7.5 mg/kg or 15 mg/kg), pirfenidone (100 mg/kg), and amifostine (100 mg/kg) for 4 weeks on alternate days after IR, and lung tissues (n ≥ 3 per group) were collected at 4 or 6 weeks after IR.

Hsp25 mice were generated, interbred, and maintained in pathogen-free conditions at Macrogen, Inc (see Supplementary Information).

The mouse lung carcinoma LLC1 cells, at 1 × 10⁶ in 200-μL physiologic saline, were injected in the tail vein of 7-week-old male C57BL/6N mice. Two weeks after the intravenous injection, a single dose of 90 Gy was delivered to the left whole lung using an image-guided small-animal irradiator. The mice were randomly divided into three groups (four to six mice per group) as follows: (i) LLC1 group—intravenous injection only; (ii) LLC1+ 90-Gy group—mice were exposed to a single dose of 90 Gy delivered to the left whole lung 2 weeks after intravenous injection; (iii) LLC1 + 90 Gy + J2 group—the mice were administered J2 intraperitoneally (15 mg/kg) for 2 weeks on alternate days after irradiation. On week 4, the mice were sacrificed by CO₂ asphyxiation, and lung tissues were collected for analysis.

The study of patient specimens of radiation-induced lung fibrosis (RILF) was approved by Severance Hospital, Yonsei University (Seoul, Korea). Each patient's tissue contained an irradiated fibrotic and a nonirradiated normal area. The degree of protein expression was compared between the fibrotic and normal areas in each patient's tissue.

Total RNA from the mouse lung tissues was extracted using the Easy-SpinTM Total RNA Extraction Kit according to the manufacturer's instructions (iNtRON Biotechnology; see Supplementary Information).

NCI-H460 (human non–small cell lung cancer cell line), L132 (human normal lung epithelial cell line), and LLC1 (mouse Lewis lung carcinoma) were obtained from the ATCC and cultured in RPMI or DMEM (Gibco) supplemented with 10% FBS (Gibco) in a 37°C incubator with 5% CO₂. Lentiviruses were used to create stable cell lines expressing shRNA for Hsp27 (puromycin-resistant gene). The control shRNA lentiviral particle (sc-108080), Hsp27 shRNA lentiviral particle (sc-29350), and polybrene (sc-134220) were obtained from Santa Cruz Biotechnology. To generate the sh-Control cells and sh-Hsp27 cells, cell lines were selected using puromycin (1 μg/mL) for at least 1 week. Human pulmonary fibroblasts (HPF) were obtained from PromoCell and used within nine passages. The mesenchymal-like A549 cells (A549TD) that were generated by chronic exposure with TGFβ as described previously (28), were obtained from Professor H.J. Cha (Seoul National University, Seoul, Korea). Cell lines were tested by BioMycoX Mycoplasma PCR Detection Kit (JCBIO Co., Ltd) to ensure that they were Mycoplasma-free.

Total RNA was isolated from the sample using TRIzol reagent (Qiagen; see Supplementary Information). Primer sequences for RT-PCR and qRT-PCR are listed in Supplementary Table S1.

Immunoblotting were performed as described previously (29) using antibodies: Twist (GeneTex and Abcam); N-cadherin, p65, Hsp27, LaminB, and β-actin (Santa Cruz Biotechnology); phospho-Hsp27 (Ser82), phospho-IkBα (Ser32/36), phospho-STAT3 (Tyr705), STAT3 (Cell Signaling Technology); IkBα, IL6, IL1β, and pro-SPC (Abcam); fibronectin, vimentin, and E-cadherin (Becton-Dickinson Laboratories); ZO-1 (Thermo Fisher Scientific); α-SMA (Sigma); and Alexa488-conjugated phalloidin (Invitrogen).

Cells in 60-mm and 100-mm petri dishes were exposed to radiation (5 or 10 Gy as a single dose) generated by a 137 Cs gamma-ray source (Elan 3000, Atomic Energy of Canada, Mississauga, Canada) at a dose rate of 3.81 Gy/minute. Radiation workers received radiation safety management training annually, provided by the Korea Foundation of Nuclear Safety (KoFONS).

For immunoprecipitation, cells were lysed in lysis buffer (500 mmol/L NaCl, 50 mmol/L Tris-HCl pH 7.5, 0.5% Triton X-100, 1 mmol/L EDTA, and 1 mmol/L DTT), clarified by centrifugation, incubated with IkBα antibody, and immunoprecipitated with protein A (Sigma-Aldrich). The precipitates were washed three times and analyzed by Western blotting.

For the histologic study, 4-μm tissue sections were stained with hematoxylin and eosin (H&E) and Masson trichrome (MT). IHC staining was carried out using anti-Twist (1:100 dilution; GTX127310, GeneTex), anti-IL6, and anti-IL1β (1:100 dilution; ab6672 and ab9722, Abcam, respectively), and anti-Hsp27 (1:200 dilution; sc-13132, Santa Cruz Biotechnology) at 4°C overnight. Slides were then incubated with Avidin–Biotin peroxidase complex (ABC kit, Vector Laboratories) and developed using 3, 3′-diaminobenzidine tetrachloride (DAB; Zymed Laboratories).

Cells were fixed in 2% paraformaldehyde for 1 hour, followed by blocking and incubation with primary antibodies at 4°C overnight. Anti-α-SMA (A5228, Sigma, 1:200), anti-p65 (sc-8008, Santa Cruz Biotechnology, 1:100), anti-Twist (sc-15393, Santa Cruz Biotechnology, 1:100), and anti-Hsp27 (sc-13132, Santa Cruz Biotechnology, 1:200) were used to detect expression. The morphologic change was investigated by Alexa488-conjugated phalloidin staining (A12379, Invitrogen, 1:200). For immunofluorescence staining, tissue sections stained with pro-SPC, α-SMA, and IkBα (1:100 dilution; ab90716, ab7817 and ab32518, Abcam, respectively) were incubated with appropriate fluorescent secondary antibodies and counterstained with 4,6-diamidino-2-phenylindole dihydrochloride (DAPI). Images were viewed under a confocal microscope (LSM700, Zeiss).

Micro-CT analysis was performed as described previously (27).

Lung function in irradiated mice was evaluated with the Flexivent system (Flexivent; SCIREQ), which measures flow–volume relationships in the respiratory system, including forced oscillation, to distinguish airway and lung tissue variables (ref. 30; see Supplementary Information and Supplementary Table S2).

Comparisons of all results were performed by one- or two-way ANOVA and Newman–Keuls test where indicated. The difference was considered statistically significant at P ≤ 0.05; P ≤ 0.01; and P ≤ 0.005. All statistical analyses were performed using GraphPad Prism.

To confirm the fibrosis, lung sections were stained with Masson Trichrome to visualize the deposition of blue-colored collagen. At 4 weeks, extensive collagen was observed, correlating with late-stage fibrosis (Fig. 1A). Antibody protein arrays showed that irradiation induced the secretion of Hsp27 alone into the blood without increasing the levels of other Hsps such as Hsp90 and Hsp70 (Supplementary Fig. S1A). Therefore, we examined the Hsp27 expression in lung tissues by IHC analysis and found that Hsp27 protein expression was increased during lung fibrosis (Fig. 1B). No increase in hspb1 mRNA expression was detected during fibrosis (Supplementary Fig. S1B). To elucidate the role of Hsp27 in lung fibrosis directly in vivo, Hsp25 TG mice were used (Supplementary Fig. S1C and S1D). IR-induced lung fibrosis was exacerbated after focal exposure to high-dose IR (75 Gy) in Hsp25 TG mice, 6 weeks after IR. After IR exposure, the Hsp25 TG mice showed an abundance of neutrophils and mononuclear cells in the alveoli, greater destruction of alveolar septa, intra-alveolar hyaline membrane formation, and a marked increase in collagen deposition compared with control C57BL/6N (BL6) mice. CT images may be used to predict fibrosis, and micro-CT is comparable with clinical CT in humans (31). Six weeks after IR, the typical micro-CT manifestations of SBRT-induced lung injury, such as ground-glass opacities and consolidation (32), were observed in the irradiated left lung. These effects were strongly induced in Hsp25 TG mice. Normal lung volume after IR in Hsp25 TG mice was lower compared with that in control BL6 mice (Fig. 1C). Functional lung parameters evaluated in this study are listed in Supplementary Table S2. There were significant differences in inspiration capacity (IC), quasi-static compliance (Cst), and tissue damping (G) of the lungs following exposure to IR between BL6 and Hsp25 TG mice. These results reflect the respiratory distress induced by irradiation. The respiratory distress in Hsp25 TG mice appears to have been significantly potentiated (Fig. 1D).

To elucidate the cellular role of Hsp27 during the development of IR-induced lung fibrosis, we initially examined the morphologic changes in L132 cells following IR. Control L132 cells were round or polygonal and exhibited very close cell–cell proximity reminiscent of cellular tight junctions. IR transformed the cells into a spindle shape. These changes became clearer and more abundant with increased dose. However, cells with Hsp27 shRNA (shHsp27) showed inhibition of these IR-induced morphologic features (Supplementary Fig. S2A). Western blot results confirmed that IR decreased the expression of the epithelial markers such as ZO-1. However, the expression of mesenchymal markers including Twist, fibronectin, and α-SMA increased in L132 cells and shHsp27 attenuated these phenomena. Quantitative RT-PCR for twist1 and fn1 also showed similar patterns of protein expression (Fig. 2A).

Previously, we identified the functional inhibition of Hsp27 following altered dimerization at the cysteine residue (Cys 137) of Hsp27 using small molecules to ameliorate Hsp27-mediated chemo- or radioresistance in lung cancer cells (29, 33). In this study, we investigated the inhibition of IR-mediated EMT using J2, which is a chromone derivative and a small-molecule Hsp27 inhibitor. J2 strongly altered the cross-linking of Hsp27 compared with SW15 (xanthone structure derivative). J2 cross-linked Hsp27 even at low concentrations of 0.1 or 0.5 μmol/L and high concentrations of 10 μmol/L, which overcame the chemoresistance, via strongly altered crosslinking of Hsp27 (Supplementary Fig. S2B). J2 treatment of recombinant Hsp27 protein strongly inhibited the formation of large oligomers of Hsp27 in nonreducing gel systems (Supplementary Fig. S2C). Pretreatment with J2 mitigated the altered expression of EMT-related proteins such as Twist, fibronectin, α-SMA, vimentin, and ZO-1 induced by IR in L132 cells assayed at 24 and 48 hours after IR (Fig. 2B; Supplementary Fig. S2D).

Increased adhesion is a characteristic of cells with a mesenchymal phenotype. Immunofluorescent hair-like fibers stained with phalloidin (green) protruding from cell surfaces into the collagen matrix assembled at the leading edge of irradiated cells, whereas treatment with shHsp27 or J2 reduced these protrusions (Fig. 2C). J2 treatment also reduced EMT-related proteins such as fibronectin, N-cadherin, α-SMA, and Twist resulting from IR in HPF cells on Western blotting and immunofluorescence images of α-SMA (Fig. 2D). Continuous exposure to TGFβ by A549 lung carcinoma cells [A549 TGFβ-differentiated cell lines (A549TD)] yielded a lower expression of E-cadherin and a higher expression of Hsp27 and vimentin than occurred in the parent A549 cells (Supplementary Fig. S3A), without affecting other HSPs such as Hsp70 and Hsp90 (Supplementary Fig. S3B), indicating Hsp27-mediated EMT. Moreover, J2 treatment dramatically restored the morphologic changes of A549TD cells and decreased the expression of E-cadherin. The expression of vimentin in A549TD cells was also restored by 0.1 and 0.5 μmol/L J2 treatment (Fig. 2E; Supplementary Fig. S3C), suggesting that Hsp27 inhibition modulated EMT. J2 concentrations (0.05, 0.1, and 0.5 μmol/L) neither induced cellular cytotoxicity in a colony-forming assay nor demonstrated cytotoxicity on flow cytometry using L132 lung epithelial cells, NCI-H460, and A549 lung cancer epithelial cells. Cell-protective effects by J2 after IR were observed and the protective effect was predominantly seen in normal cells of L132 rather than in NCI-H460 and A549 cancer cells (Supplementary Fig. S3D and S3E).

To elucidate whether a small-molecule Hsp27 cross-linker J2 inhibits IR-mediated lung fibrosis in mice, we compared the changes in left lung surface morphology in the control and IR group. In contrast to the brown colored lungs in control mice, the lungs of irradiated mice exhibited a definite white, ring-like appearance. Intraperitoneal injection of J2 resulted in less injury than was apparent in the IR-only mice. Alveolar infiltration of inflammatory cells and the formation of intra-alveolar hyaline membranes in the IR group were significantly greater than in the control group. J2-treated mice exhibited reduced tissue damage. Masson trichrome staining revealed a marked increase in collagen deposition in the IR group than in the control group, which was significantly reversed by J2 treatment. Six weeks after IR, ground-glass opacities and consolidation were observed in the irradiated left lung; in contrast, these effects were decreased in J2-treated mice. Normal lung volume in the IR group was lower than in the control mice. However, normal lung volume appeared to significantly recover in J2-treated mice (Fig. 3B; Supplementary Fig. S4A). Also, there were significant differences in IC, Cst, G, and tissue elastance (H) of the lungs between IR group and control mice. The IC and Cst of the IR group were significantly lower compared with those of the control group. The values of G and H in the IR group were higher than in the control group. Mice treated with 15 mg/kg J2 exhibited significant differences in IC, Cst, and G parameters, indicating the protective effect of J2 on IR-induced lung injury (Supplementary Fig. S4B). The effects of J2 were more prominent than those of pirfenidone or amifostine, even though administration dosage of J2 was less than pirfenidone and amifostine.

To develop evidence that epithelial cells express a mesenchymal phenotype during IR-induced lung fibrosis, we performed immunofluorostaining for both alveolar epithelial cell–specific protein SPC and myofibroblast-specific marker α-SMA. The luminal layer of alveolar cells was immunoreactive for SPC and the SPC-positive cell number was not substantially different among the groups. Moreover, α-SMA expression was increased by IR and SPC/α-SMA costaining cells were also increased by IR, indicating that an EMT process occurred during IR-induced lung fibrosis. However, J2 treatment with IR decreased the SPC/α-SMA costaining cells (Fig. 3C).

We also investigated the effects of J2 on IR-induced lung fibrosis in Hsp25 TG mice. Similar to the findings of normal BL6 mice, irradiated areas of the left lung clearly exhibited a local injury in BL6 mice. In the Hsp25 TG mice, we observed aggravated lung injury grossly and histologically, which was attenuated by J2 treatment. Masson trichrome, Sirius red, and IHC hydroxyproline staining revealed a marked increase in collagen deposition in Hsp25 TG mice after IR compared with IR-treated control BL6 mice and J2 also inhibited collagen deposition in Hsp25 TG mice. Six weeks after IR, the normal lung volume was lower in irradiated Hsp25 TG mice than in irradiated BL6 mice. However, it appears to have been significantly restored in J2-treated Hsp25 TG mice (Fig. 3D; Supplementary Fig. S5A and S5B). Increased costaining of SPC/α-SMA after IR was restored by combined treatment with J2, indicating that the IR-induced EMT process was blocked by J2 treatment (Fig. 3E). There were significant differences in IC, Cst, G, and H of the lungs between irradiated Hsp25 TG and irradiated control BL6 mice. However, IR-induced respiratory distress in J2-treated mice appeared to be significantly reversed even in Hsp25 TG mice (Supplementary Fig. S5C).

To identify the underlying mechanisms of inhibition of IR-mediated lung fibrosis by J2, we performed a cDNA microarray of lung tissues after focal exposure to 75 Gy with or without 4 weeks of intraperitoneal administration of J2. Temporal changes in gene expression were hierarchically clustered (Supplementary Fig. S6A). Analysis of fibrosis-related genes in the microarray data revealed upregulation of only twist1, il6, and il1β genes by focal irradiation, which was reversed by J2 treatment. We also determined the mRNA levels of twist1, il-6, and il-1β using qRT-PCR (Fig. 4A) and protein levels using IHC (Fig. 4B); the levels were similar to those obtained with cDNA microarray. J2 treatment restored the significant increase induced by focal IR. We also found that J2 did not alter the expression of Hsp25 protein in lung tissues (Supplementary Fig. S6B). We next examined whether Hsp27 knockdown or J2 treatment modulated IR-induced twist1, il6, and il1β genes in a cell system. The qRT-PCR analysis of L132 lung epithelial cells revealed that the increased expression of twist1, il6, and il1β genes by IR was suppressed by shRNA of Hsp27 or J2 pretreatment, based on the results detected at 12 hours after IR (Fig. 4C). Immunofluorescence data in L132 cells revealed that the basal Twist level was inhibited by shHsp27 and IR-induced Twist activation was also ameliorated by knockdown of Hsp27 (Fig. 4D).

Because NFκB is a regulator of Twist, IL1β, and IL6 (34–36), we investigated the association between p65, one of the components of NFκB, and IR-mediated EMT markers using siRNA and BAY11–7082, an NFκB inhibitor. BAY11-7082 or p65 knockdown dramatically inhibited the Twist protein level (Supplementary Fig. S7A and S7B), suggesting Twist as a downstream effector of NFκB. To elucidate whether the binding between Hsp27 and IkBα was affected by IR, immunoprecipitation (IP) was performed. The results indicated an increase in the binding activity between Hsp27 and IkBα by IR, and a decreased binding activity between IkBα and p65 (Fig. 5A). Moreover, siRNA of twist1 did not inhibit il6 and il1β genes in L132 cells, suggesting that NFκB activation by Hsp27 may be a master regulator of twist1, il6, and il1β genes (Supplementary Fig. S7C). We also investigated whether Hsp27 level regulates NFκB-mediated Twist expression by IR and found that Hsp27 knockdown or Hsp27 cross-linker J2 inhibited IR-mediated Twist expression, accompanied by inhibition of IkBα phosphorylation, at 3 hours of IR before Twist activation (24 hours of IR). In the case of STAT3 phosphorylation, another transcription factor of Twist, IR slightly induced STAT3 phosphorylation. However, Hsp27 inhibition did not affect the phosphorylation (Fig. 5B and C). IR-mediated nuclear translocation of p65 was inhibited by shHsp27 or J2 treatment (Fig. 5D; Supplementary Fig. S7D). The siRNA of p65 inhibited IR-mediated mRNA expression of twist1, il-6, and il-1β as shown in qRT-PCR, suggesting that NFκB activation by IR acted upstream of twist1, il6, and il1β (Supplementary Fig. S7E). Activation of NFκB was reflected by IkBα degradation and our immunofluorescence results indicated that IkBα expression was lower in irradiated lungs of Hsp25 TG mice compared with the BL6-IR and J2 treatment to Hsp25 TG increased the intensity of fluorescence of IkBα. To investigate IkBα–NFκB signaling activation by Hsp27 is connected to the development of IR-induced EMT process in irradiated lungs, the level of IkBα and α-SMA were assessed via coimmunofluorescence staining. Expression of lower IkBα and higher α-SMA was shown in irradiated lungs of Hsp25 TG mice compared with the BL6-IR. J2 treatment to Hsp25 TG reversed the intensity of fluorescence of IkBα and α-SMA. These results indicate that IkBα-NFκB activation by Hsp27 induces IR-induced lung fibrosis through the EMT process (Fig. 5E).

To elucidate the expression of Hsp27 in tumor model and human lung tissues, first, orthotropic lung tumors using LLC1 cells were established and 90-Gy IR was irradiated to the left whole lung of mice for 2 weeks. Most of the tumors regressed in mice treated with IR and no detectable residual tumor was observed in mice treated with both IR and J2. Increased collagen deposition in irradiated normal lung lesions of orthotopic mice model was decreased by J2 (Fig. 6A). Hsp25 expression was also increased in irradiated normal lungs, while in the tumor lesions, no increase of Hsp25 by IR was observed. Without IR, Hsp25 protein was more abundantly expressed in tumor lesions than nontumor lesions (Supplementary Fig. S8A). Moreover, J2 showed a more positive effect on tumor regression than was seen in the IR-alone group. J2 concentrations (0.05, 0.1, and 0.5 μmol/L) induced cellular cytotoxicity in demonstrated cytotoxicity on flow cytometry using LLC1 mouse Lewis lung adenocarcinoma cells. Microscopic analysis revealed that NCI-H460 lung adenocarcinoma cells are in the form of a polygonal cobblestone and very close together. After exposure to IR, the cells transformed into a spindle-like shape that was more definite as dose increase. However, cells with Hsp27 shRNA (shHsp27) showed inhibition of IR-induced morphologic change (Supplementary Fig. S8C). Western blot results confirmed that IR decreased the expression of the epithelial markers such as ZO-1. Also, the expression of mesenchymal markers including Fibronectin, Twist, and α-SMA increased in lung cancer cells. However, shHsp27 attenuated these phenomena. Quantitative RT-PCR for twist1 and fn1 also showed similar patterns of protein expression (Supplementary Fig. S8D), suggesting that Hsp27 inhibition regulated EMT in cancer cells as well as in normal epithelial cells.

Next, we examined whether expression of Hsp27 is increased in RILF tissues of patients. Fibrotic tissues of patients with lung cancer who had surgery following radiotherapy for lung adenocarcinoma were selected on the basis of H&E staining. IHC staining for Hsp27 and Twist were performed on 14 patient tissue samples of RILF; the fibrotic areas of RILF patient tissues exhibited upregulated Hsp27 expression compared with the normal areas. The increased expression of Twist in the irradiated fibrotic areas of patient tissues was well correlated with Hsp27 expression. Three representative results are shown in Fig. 6B and the clinicopathologic characteristics of the patients are summarized in Supplementary Table S3.

In this study, we demonstrate the novel mechanisms of Hsp27 in IR-induced lung fibrosis development and propose Hsp27 as a possible therapeutic target for IR-induced lung fibrosis. In the analysis of irradiated lungs, we identified Hsp27 upregulation during IR-induced lung fibrosis. Previous proteomics studies also revealed an upregulation of Hsp27 in lung fibroblast cell lines upon treatment with TGFβ1 and in IPF lung tissues (15, 37). Moreover, effective attenuation of BLM-induced lung fibrosis in mice via airway delivery of siRNAs of Hsp27 with downregulation of myofibroblast-associated proteins such as fibronectin, type 1 collagen, and OPN has been reported (38). Another report suggested that Hsp27 antisense oligonucleotide effectively suppressed adenovirus-expressing TGFβ1-induced subpleural fibrosis in rats, which suggested that Hsp27 prevented Snail degradation by the proteasomal system (14). However, no other molecular mechanisms of Hsp27 in lung fibrosis, especially IR-induced lung fibrosis, were reported.

IR did not transcriptionally induce Hsp27 or its upstream regulator, HSF1 (data not shown). However, the protein expression of Hsp27 was dramatically increased, suggesting that IR may regulate Hsp27 protein stability rather than its transcriptional activation. Hsp27 protein accumulation in IR-induced fibrotic tissues without affecting Hsp27 transcription suggests that proteasomal inhibition was induced by IR, which may affect the increase in Hsp27 protein levels. The proteasomal inhibitor MG-132 also promotes Hsp27 phosphorylation (39). Hsp27 phosphorylation is catalyzed by MAPK-activated protein kinase 2 (MAPKAPK-2), which regulates its expression levels (40, 41). IR activates MAPKAPK-2 and phosphorylates Hsp27 (42). Our data also suggest Hsp27 phosphorylation by IR, which may affect Hsp27 protein stability.

We used generated Hsp25 TG mice to elucidate whether the increased Hsp27 expression initiates the development of lung fibrosis. Hsp25 TG mice showed increased collagen deposition and defective lung function by IR compared with control mice. When EMT-related genes were evaluated in human lung cell lines, shHsp27 inhibited IR-induced EMT and the small-molecule functional inhibitor of Hsp27 (J2), inhibited IR-induced EMT in normal epithelial cells, and lung fibrosis in both normal C57BL/6N and Hsp25 TG mice with better lung functions. Moreover, Hsp27 was overexpressed in irradiated fibrotic lung tissues in an orthotopic lung tumor model and nontumor lesions of human lung tissues after radiotherapy. J2 also inhibited IR-induced lung fibrosis in an orthotopic lung tumor model without affecting therapeutic effects of IR on the tumor, and even more dramatic tumor regression effects than in the IR-alone group. Moreover, Hsp27 inhibition also diminished IR-induced EMT in cancer cells. Therefore, Hsp27 inhibition is an effective strategy for the inhibition of IR-induced lung fibrosis during radiotherapy.

To elucidate the role of Hsp27 in the regulation of lung fibrosis, we initially analyzed EMT genes based on the microarray data of lung tissues. The results showed that increased mRNA levels of twist1, il1β, and il6 were diminished by treatment with an Hsp27 inhibitor and these genes were downstream molecules in NFκB pathways. Hsp27 is known to directly interact with IkBα and facilitates its proteasome degradation, which is the main mechanism of NFκB activation by Hsp27 (43). IR is also known as an NFκB activator (44). Therefore, Hsp27 may represent a more potent activator of NFκB signaling pathways including twist1, il1β, and il6 genes, which are some of the major regulators of Hsp27-mediated EMT progression. Indeed, Hsp27 knockdown or pharmacologic inhibition of Hsp27 inhibited NFκB pathway and the expression of twist1, il1β, and il6 genes in vitro and in vivo.

TGFβ1 is a well-known factor in lung fibrosis and several studies suggest that TGFβ1 is produced during IR-mediated lung fibrosis (45). Our previous studies also showed that TGFβ1 expression was overexpressed in our SBRT mimicking fibrotic lungs and serum (46, 47). Moreover, J2 treatment inhibited TGFβ1 protein expression in earlier time point, 14 days after IR (47), and our preliminary data also suggested the high expression of TGFβ1 in 75-Gy–irradiated lungs was blocked by cotreatment of J2. Therefore, involvement of TGFβ1 in IR-induced lung fibrosis was not ruled out and Hsp27 inhibitor also affect TGFβ-mediated fibrosis mechanisms.

It was recently shown that EMT occurring in peritoneal, kidney, and lung fibrosis, as well as in breast cancer stem cells, was associated with increased Hsp27 expression (48–50), suggesting a rationale for the development of Hsp27 inhibitors in fibrosis treatment. Even though Hsp27 is an attractive therapeutic target in fibrosis, unlike Hsp90 or Hsp70, it lacks an active site or ATP-binding pocket. Hence, only two Hsp27 inhibitors are in the clinical trials. However, the limitations associated with intracellular delivery of OGX427 relate to the small size of the inhibitor and lack of mechanism underlying Hsp27 in the case of RP101 (51–53). Aside from RP101, no small molecules have been developed as Hsp27 inhibitors and clinical trial data of RP101 so far are not good.

Even though currently approved therapies for lung fibrosis such as pirfenidone and nintedanib are clinically available, an alternate strategy to address the unmet therapeutic need lung fibrosis is needed. Our results for the first time demonstrate the increased expression of Hsp27 in IR-induced lung fibrosis and increased Hsp27 aggravated IkBα–NFκB signaling pathways to increase EMT. Therefore, pharmacologic Hsp27 inhibitors may be effectively used as inhibitors of IR-induced lung fibrosis, especially after radiotherapy (Fig. 6C).

No potential conflicts of interest were disclosed.

Conception and design: Y.-J. Lee, J. Cho, Y.-S. Lee

Development of methodology: J. Cho

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Jeon, Y.J. Yoo, H. Jin, H.Y. Won, K. Yoon, Y. Na, J. Cho

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.-Y. Kim, E.S. Hwang, Y. Na, J. Cho, Y.-S. Lee

Writing, review, and/or revision of the manuscript: J.-Y. Kim, S. Jeon, E.S. Hwang, J. Cho, Y.-S. Lee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-Y. Kim, S. Jeon, J. Cho

Study supervision: J. Cho, Y.-S. Lee

This work was supported by grants from the National Research Foundation of Korea, (NRF-2017R1A2B2002327, NRF-2017M2A2A702019560, and NRF-2018R1A5A2025286), funded by the Korean government (Ministry of Science and ICT).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.