Blockade of EGFR Activation Promotes TNF-Induced Lung Epithelial Cell Apoptosis and Pulmonary Injury

Pneumonitis is the leading cause of death associated with the use of epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (EGFR-TKIs) against non-small cell lung cancer (NSCLC). However, the risk factors and the mechanism underlying this toxicity have not been elucidated. Tumor necrosis factor (TNF) has been reported to transactivate EGFR in pulmonary epithelial cells. Hence, we aimed to test the hypothesis that EGFR tyrosine kinase activity regulates TNF-mediated bronchial epithelial cell survival, and that inhibition of EGFR activity increases TNF-induced lung epithelial cell apoptosis. We used surfactant protein C (SPC)-TNF transgenic (tg) mice which overexpress TNF in the lungs. In this model, gefitinib, an EGFR-TKI, enhanced lung epithelial cell apoptosis and lymphocytic inflammation, indicating that EGFR tyrosine kinase prevents TNF-induced lung injury. Furthermore, IL-17A was significantly upregulated by gefitinib in SPC-TNF tg mice and p38MAPK activation was observed, indicative of a pathway involved in lung epithelial cell apoptosis. Moreover, in lung epithelial cells, BEAS-2B, TNF stimulated EGFR transactivation via the TNF-α-converting enzyme in a manner that requires heparin binding (HB)-EGF and transforming growth factor (TGF)-α. These novel findings have significant implications in understanding the role of EGFR in maintaining human bronchial epithelial cell homeostasis and in NSCLC treatment.


Introduction
Epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) are given as standard therapy to non-small cell lung cancer (NSCLC) patients who carry activating mutations in the EGFR gene. The EGFR-TKI gefitinib, when administered to NSCLC patients carrying activating EGFR mutations (e.g., 15-bp deletion in exon 19, L858R in exon 21), significantly improved response and survival [1,2]. Furthermore, these EGFR mutations were predominantly found in patients of East-Asian origin [3]. However, although EGFR-TKI treatment has significant effects on response and prognosis, a study has reported that 40% of patients treated with EGFR-TKI experience severe adverse events, and treatment had to be discontinued in 7.7% of the patients because of unmanageable severe adverse events [4]. Diarrhea and rashes were the most frequent adverse effects but were Table 1. Cytokine mRNA expression profile of wildtype (WT) vs surfactant protein C-tumor necrosis factor transgenic (SPC-TNF tg) mouse lung tissue. Lung mRNA transcripts of WT and SPC-TNF tg mice were quantified by real-time PCR and normalized to GAPDH. The data are represented relative to control values, as means ± SEM (n = 8). Significant values are indicated in boldface. Gefitinib (100 or 200 mg/kg) was orally administered once daily in wild type (WT; C57BL/6) or SPC-TNF tg mice for 14 days. In wild type (WT) mice, no significant pathological change was observed ( Figure 1a). In SPC-TNF tg mice, administration of gefitinib resulted in severe lymphocytic infiltration in the alveolar septa, and the mice also developed thicker alveolar septa than non-administered mice (Figure 1b). After 14 days of administration, there was also no death related to gefitinib administration. Moreover, there was no change in body weight in either WT or SPC-TNF tg mice (Figure 1c). To determine the inflammatory cytokines that mediate lymphocytic infiltration in SPC-TNF tg mice, real-time RT PCR was performed (Tables 2 and 3). In wt mice, although mRNA levels for the pro-inflammatory cytokine IL-1β were significantly increased, no inflammation and lung injury was observed, suggesting a maintenance of balance by other anti-inflammatory cytokines, or through an unknown mechanisms. The mRNA level of IL-17α was found to be significantly increased by gefitinib administration. Taken together, these results suggest that treatment with EGFR-TKI might cause lung inflammation in SPC-TNF tg mice that overexpress TNF in the lung. Gefitinib enhances lymphocyte infiltration and causes lung tissue inflammation. Lung tissue sections of (a) wild type (WT) mice and (b) surfactant protein C-tumor necrosis factor (SPC-TNF) transgenic (tg) mice were stained with hematoxylin and eosin. The images were obtained at a magnification of 40× and 200×, in the upper and lower rows, respectively. (c) Body weight changes are shown for WT and SPC-TNF Tg mice over 14 days of daily oral treatment with 1% tween 80 as control or with gefitinb (100 or 200 mg/kg). The WT; C57BL/6 or SPC-TNF tg mice used in these studies were 6-8 weeks old. The scale bars represent 500 μm and 100 μm for 40× and 200× magnification, respectively. Data in the figure represent mean body weights ± SEMs; * p < 0.05, Student's t-test.  Figure 1. Gefitinib enhances lymphocyte infiltration and causes lung tissue inflammation. Lung tissue sections of (a) wild type (WT) mice and (b) surfactant protein C-tumor necrosis factor (SPC-TNF) transgenic (tg) mice were stained with hematoxylin and eosin. The images were obtained at a magnification of 40× and 200×, in the upper and lower rows, respectively. (c) Body weight changes are shown for WT and SPC-TNF Tg mice over 14 days of daily oral treatment with 1% tween 80 as control or with gefitinb (100 or 200 mg/kg). The WT; C57BL/6 or SPC-TNF tg mice used in these studies were 6-8 weeks old. The scale bars represent 500 µm and 100 µm for 40× and 200× magnification, respectively. Data in the figure represent mean body weights ± SEMs; * p < 0.05, Student's t-test.

Indocyanine Green (ICG)-Loaded Liposomes Accumulated in Whole Lungs of Gefitinib-Treated SPC-TNF tg Mice
To evaluate the extent of gefitinib-induced inflammation in mouse lungs, we used indocyanine green (ICG)-loaded liposomes, labeled with sialyl Lewis X (SLX) on the surface (SLX-Lipo-ICG). These SLX-Lipo-ICG could accumulate in the regions of inflammation or tumor owing to the interaction of SLX and E-selectin, which enhanced E-selectin expression by inflammatory cytokines, such as TNF, in vascular endothelial cells [22]. In WT and SPC-TNF tg mice, we measured whole-lung fluorescence ex vivo using the Clairvivo OPT fluorescence imaging system. We observed slight accumulation of ICG-loaded liposomes in whole lungs of SPC-TNF tg mice that were not treated with gefitinib compared with whole lungs of WT mice. Furthermore, after 14 days of gefitinib administration, accumulation of liposomes was enhanced in whole-lung samples taken from SPC-TNF tg mice. However, there was no enhancement in liposome accumulation after gefitinib treatment in WT mice (Figure 2a,b). These results suggest that EGFR tyrosine kinase activity protects mice from TNF-induced lung inflammation in mice model. interaction of SLX and E-selectin, which enhanced E-selectin expression by inflammatory cytokines, such as TNF, in vascular endothelial cells [22]. In WT and SPC-TNF tg mice, we measured wholelung fluorescence ex vivo using the Clairvivo OPT fluorescence imaging system. We observed slight accumulation of ICG-loaded liposomes in whole lungs of SPC-TNF tg mice that were not treated with gefitinib compared with whole lungs of WT mice. Furthermore, after 14 days of gefitinib administration, accumulation of liposomes was enhanced in whole-lung samples taken from SPC-TNF tg mice. However, there was no enhancement in liposome accumulation after gefitinib treatment in WT mice (Figure 2a,b). These results suggest that EGFR tyrosine kinase activity protects mice from TNF-induced lung inflammation in mice model. Figure 2. Gefitinib enhances lung inflammation in surfactant protein C-tumor necrosis factor (SPC-TNF) transgenic (tg) mice. Indocyanine green (ICG)-loaded liposomes carrying sialyl Lewis X were used to detect regions of inflammation in whole lungs of (a) wildtype (WT) and (b) SPC-TNF tg mice after daily gefitinib (100 mg/kg) treatment for 14 days. Images were captured using the Clairvivo OPT in vivo imaging system (Shimadzu, Kyoto, Japan).

EGFR Tyrosine Kinase Activity Protected Lung Epithelial Cells from TNF-Induced Apoptosis in the SPC-TNF tg Mice Model.
TNF is a pleiotropic cytokine that activates both anti-and pro-apoptotic signaling pathways. Cell survival is determined by the balance of these two pathways [20]. Therefore, we detected the cells undergoing apoptosis by in situ oligo-ligation (ISOL) in WT and SPC-TNF tg mice under gefitinib administration or not (Figure 3a,c). Gefitinib administration significantly enhanced apoptosis of lung epithelial cells in SPC-TNF tg mice but not in WT mice (Figure 3b,d). EGFR Figure 2. Gefitinib enhances lung inflammation in surfactant protein C-tumor necrosis factor (SPC-TNF) transgenic (tg) mice. Indocyanine green (ICG)-loaded liposomes carrying sialyl Lewis X were used to detect regions of inflammation in whole lungs of (a) wildtype (WT) and (b) SPC-TNF tg mice after daily gefitinib (100 mg/kg) treatment for 14 days. Images were captured using the Clairvivo OPT in vivo imaging system (Shimadzu, Kyoto, Japan).

EGFR Tyrosine Kinase Activity Protected Lung Epithelial Cells from TNF-Induced Apoptosis in the SPC-TNF tg Mice Model
TNF is a pleiotropic cytokine that activates both anti-and pro-apoptotic signaling pathways. Cell survival is determined by the balance of these two pathways [20]. Therefore, we detected the cells undergoing apoptosis by in situ oligo-ligation (ISOL) in WT and SPC-TNF tg mice under gefitinib administration or not (Figure 3a,c). Gefitinib administration significantly enhanced apoptosis of lung epithelial cells in SPC-TNF tg mice but not in WT mice (Figure 3b,d). EGFR phosphorylation (P-EGFR) was significantly increased in SPC-TNF tg mice compared to WT mice (Figure 3e,f). Moreover, EGFR tyrosine kinase activity, as indicated by EGFR phosphorylation, was significantly inhibited in SPC-TNF tg mice after gefitinib administration (Figure 3e,g). Activation by phosphorylation of downstream signals AKT and ERK1/2 also decreased in SPC-TNF tg mice ( Figure 3h). Anti-apoptotic responses can be mediated by AKT, ERK1/2, or Raf; else, p38MAPK is activated through MKK3/6 phosphorylation, leading to apoptosis [23,24]. Therefore, apoptosis in SPC-TNF tg mice may be induced by p38 MAPK pathways via MAP kinase kinase (MKK) 3/6, in addition to the inhibition of AKT and ERK1/2 activation with gefitinib exposure (Figure 3i).

TNF Stimulated EGFR Phosphorylation and Prevented Apoptosis in Lung Epithelial Cells
To determine the mechanisms of EGFR transactivation by TNF exposure in lung epithelial cells, BEAS-2B cells were employed. At first, BEAS-2B cells were exposed to TNF (100 ng/mL) from 1 to 240 min; stimulation of EGFR phosphorylation was initiated from 15 min after TNF exposure ( Figure 4a). Moreover, EGFR phosphorylation was enhanced by TNF in a dose-dependent matter ( Figure 4b). TNF-induced EGFR phosphorylation was inhibited by gefitinib, suggesting the necessity of EGFR tyrosine kinase activity in this regard. (Figure 4c). Transfection of small interfering RNA (siRNA) against EGFR into BEAS-2B cells led to decreased EGFR expression and attenuated TNF-induced EGFR phosphorylation ( Figure 4d). Therefore, TNF-induced EGFR phosphorylation required the expression and activation of EGFR. Interestingly, exposing BEAS-2B cells simultaneously with TNF and gefitinib, enhanced TNF-induced apoptosis (Figure 4e,f). These results suggest that EGFR phosphorylation by TNF inhibited TNF-induced apoptosis.

TNF Stimulated EGFR Phosphorylation and Prevented Apoptosis in Lung Epithelial Cells.
To determine the mechanisms of EGFR transactivation by TNF exposure in lung epithelial cells, BEAS-2B cells were employed. At first, BEAS-2B cells were exposed to TNF (100 ng/mL) from 1 to 240 min; stimulation of EGFR phosphorylation was initiated from 15 min after TNF exposure ( Figure  4a). Moreover, EGFR phosphorylation was enhanced by TNF in a dose-dependent matter (Figure 4b). TNF-induced EGFR phosphorylation was inhibited by gefitinib, suggesting the necessity of EGFR tyrosine kinase activity in this regard. (Figure 4c). Transfection of small interfering RNA (siRNA) against EGFR into BEAS-2B cells led to decreased EGFR expression and attenuated TNF-induced EGFR phosphorylation (Figure 4d). Therefore, TNF-induced EGFR phosphorylation required the expression and activation of EGFR. Interestingly, exposing BEAS-2B cells simultaneously with TNF and gefitinib, enhanced TNF-induced apoptosis (Figure 4e,f). These results suggest that EGFR phosphorylation by TNF inhibited TNF-induced apoptosis.

TNF Transactivates EGFR through TACE and Regulates Cell-Survival via the PI3-Kinase/AKT Pathway
TACE facilitates the shedding of EGFR ligands, which are then available to bind and stimulate EGFR again in a process called the autocrine loop [6]. In BEAS-2B cells, the TACE inhibitor TAPI-1 suppressed TNF-induced EGFR phosphorylation, but not EGF-stimulation (Figure 5a). Furthermore, TAPI-1 enhanced TNF-induced apoptosis significantly (Figure 5b and c), suggesting that TACE mediates EGFR phosphorylation, leading to cell survival upon TNF exposure. The major downstream signals from EGFR include the PI-3K/AKT and RAS/ERK pathways. TNF transactivated EGFR by shedding mainly two EGFR ligands, and then we determined the signaling molecules downstream of EGFR that regulate cell survival in the presence of TNF. We used wortmannin and U0126 to inhibit AKT and ERK1/2, respectively. Wortmannin or U0126 alone did not induce apoptosis of BEAS-2B cells. Wortmannin significantly enhanced TNF-induced apoptosis, while U0126 did not affect TNF-induced apoptosis (Figure 5d,e). Therefore, the PI3K/AKT pathway may contribute to cell survival via TNF-induced EGFR transactivation. Although it would be difficult to recapitulate the SPC-TNF tg mouse model using BEAS-2B cells, we speculate that P38MAPK signal contributes to TNF-induced apoptosis [25].

EGFR Ligands, HB-EGF, and TGF-α Dominantly Regulate TNF-Induced EGFR Transactivation via TACE in BEAS-2B Cells
Cetuximab is an EGFR-neutralizing antibody. It binds EGFR and downregulates ligand-induced EGFR phosphorylation [26]. To verify the involvement of EGFR ligands in TNF-induced EGFR transactivation, we used neutralizing antibodies for EGFR. Therefore, TNF-induced EGFR phosphorylation was inhibited by cetuximab, indicating that EGFR ligands contributed to EGFR transactivation by TNF (Figure 6a). To identify the EGFR ligands involved in TNF-induced EGFR phosphorylation, neutralizing antibodies were employed. As shown in Figure 6b, EGFR-transactivation by TNF was attenuated by the treatment of neutralizing antibodies against HB-EGF and TGF-α. Furthermore, the combination of neutralizing antibodies against HB-EGF and TGF-α exhibited remarkable inhibition of EGFR phosphorylation in the presence of TNF (Figure 6c). To measure the expression of HB-EGF and TGF-α in the culture supernatant, ELISA was performed. HB-EGF and TGF-α expression in the culture supernatant increased upon TNF exposure; however, the presence of TAPI-1 reduced the expression of these EGFR-ligands (Figure 6d,e). Moreover, knockdown of TACE using siRNA (Figure 6f), reduced the amount of HB-EGF and TGF-α in the culture supernatant (Figure 6g,h). These results indicate that the expression of HB-EGF and TGF-α in the culture supernatant was regulated by TACE and that these ligands were involved in TNF-induced transactivation in BEAS-2B cells. Taken together, lung epithelial cells BEAS-2B should be protected from TNF-induced apoptosis through EGFR transactivation via the autocrine loop of the EGFR ligands, HB-EGF and TGF-α, which are cleaved by TACE.

Discussion
This study was performed to verify the hypothesis that EGFR-TKIs induce pneumonitis by blocking EGFR transactivation by TNF. The mechanisms of EGFR-TKI-induced pneumonitis have not been completely elucidated. However, this study provides evidence that EGFR tyrosine kinase activity prevents lung injury in a TNF-enriched environment. Furthermore, we have found that in lung epithelial cells BEAS-2B, TACE is required for TNF-induced EGFR activation and that PI3K/AKT is a target of EGFR transactivation. These findings show that transactivation of EGFR prevents TNF-induced lung epithelial cell apoptosis.
TNF-induced EGFR phosphorylation regulates critical physiological responses in acute lung injury. Airway inflammation, characterized by the release of the pro-inflammatory cytokine TNF, induces excessive EGFR signaling [27,28]. High amounts of EGFR ligands, amphiregulin and TGF-α secreted by bronchial epithelial cells result in the activation of EGFR in an autocrine loop, leading to increased mucus production by goblet cells [29,30]. Therefore, inhibiting EGFR tyrosine kinase activity may be beneficial to the treatment of hypersecretory diseases, such as asthma, chronic obstructive pulmonary disease (COPD), and bronchiectasis. Conversely, EGFR-ligands EGF or TGF-α can regulate lung epithelial cell repair in vitro and in vivo [31,32]. EGFR is overexpressed and activated in response to lung epithelial injury and play an important role in lung epithelial repair [33]. Increased amounts of both TGF-α and EGFR have been observed in bleomycin-injured lungs [34]. In addition, TGF-α has been identified in bronchoalveolar lavage fluid from patients with acute lung injury [34,35]. Taken together, the expression of EGFR or EGFR-ligands, and EGFR tyrosine kinase activity protect lung epithelial cells from apoptosis and enhance their recovery from lung inflammation and injury [36]. However, further studies are required to discover the role of EGFR and its ligands in inflammatory lung diseases. In this study, EGFR tyrosine kinase was transactivated by TNF in SPC-TNF tg mice. Inhibition of EGFR tyrosine kinase by gefitinib increased TNF-induced apoptosis in epithelial cells and significantly enhanced lung injury. The results of this study are consistent with those of previous reports, which showed that activation of EGFR and other ErbB family members protects colon epithelial cells from TNF-induced apoptosis [20,37].
For understanding the development of pneumonitis in whole lung, it is essential to evaluate lung inflammation by EGFR-TKI gefitinib administration ex vivo. In regions of inflammation in arthritis and tumors, vascular endothelial cells are activated by inflammatory cytokines, such as TGF-α and TNF. Leukocytes roll on the vascular endothelial cells by binding E-selectin on the vascular endothelial cell surface. Leukocytes infiltrate the vascular endothelium, and then migrate and accumulate in the inflammatory region. The sugar chain sialyl Lewis X (SLX) is well characterized as an E-selectin ligand, expressed on the surface of leukocytes. Therefore, SLX-liposomes primarily targeting E-selectin in activated endothelial cells deliver and accumulate their ICG content in regions of inflammation and within tumors [38,39]. Therefore, we employed an SLX-Lipo-ICG to evaluate lung tissue inflammation after administration of gefitinib in WT and SPC-TNF tg mice. The basal accumulation of SLX-Lipo-ICG in SPC-TNF tg mice was higher than in WT mice. Furthermore, after gefitinib administration, the accumulation of SLX-Lipo-ICG was enhanced in SPC-TNF tg mice but not in WT mice. These results indicate that gefitinib remarkably enhances TNF-induced lung inflammation in SPC-TNF tg mice.
Based on our results, IL-17A was the most significantly increased cytokine after gefitinib treatment in SPC-TNF tg mice. IL-17A appears to act primarily on non-hematopoietic cells, such as endothelial cells, epithelial cells, and fibroblasts [40]. Increased expression of IL-17A has been linked to inflammatory diseases such as asthma, cystic fibrosis, and COPD in the airway, as well as inflammatory bowel disease, and it appears to play an important protective role against infection [41][42][43]. Iyoda et al. reported that IL-17A and IL-17F can stimulate the phosphorylation of p38 MAPK and then enhance the expression of cytokines such as monocyte chemoattractant protein (MCP)-1 and macrophage inflammatory protein (MIP)-2 in cultured mouse mesangial cells [44]. In the SPC-TNF tg mice, gefitinib treatment led to increased phosphorylation of p38 MAPK, suggesting that IL-17A may stimulate p38 MAPK activity and induce lung epithelial cell apoptosis.
A meta-analysis of 16 trials on EGFR TKI treatment in lung cancer patients revealed that pneumonitis was the most common cause of death related to EGFR TKI toxicity [4]. A previous clinical study has identified risk factors of EGFR TKI-induced pneumonitis including old age, smoking history, pre-existing interstitial lung disease, and poor performance status [45]. For example, higher plasma concentration of TNF was measured in the older group than in the younger group [46]. Smoking may influence TNF-mediated systemic inflammation, such as in COPD or interstitial pneumonitis [47]. Furthermore, poor-performance status and cachexia, associated with cytokines of IL-6 and TNF, are well known [48]. Taken together, these risk factors are associated with TNF-enriched lung conditions. Therefore, for cases of EGFR TKI-induced pneumonitis, EGFR tyrosine kinase activity may be required to maintain homeostasis in TNF-rich lung environments where chronic inflammation is present.
In summary, we were able to elucidate the molecular mechanisms related to EGFR-TKI-induced lung injury using SPC-TNF tg mice. EGFR activation prevents TNF-induced apoptosis and severe inflammation. IL-17A was the most upregulated cytokine and may be involved. Furthermore, in lung epithelial cells BEAS-2B, we found that TNF activates EGFR phosphorylation by stimulating TGF-α and HB-EGF via TACE activity. This EGFR transactivation by TNF regulates cell survival. Taken together, our findings provide the biological mechanisms underlying EGFR-TKI-induced lung injury. Therefore, our findings suggest that caution will have to be taken prior to administration of EGFR-TKI in NSCLC cases that are associated with high levels of TNF or inflammation. This study has potential limitations. The effects reported are based on SPC-TNF tg mice, which highly express TNF in whole lung. We cannot, however, exclude the effect of other cytokines and signal transductions associated with lung fibrosis; these would be addressed in future research.
TAPI-1 (1 µM) was pretreated for 60 min, and then were treated with TNF (100 ng/mL) in various durations. Supernatants were analyzed by ELISA according to manufacturer's recommendations.

Statistical Analysis
Significant difference between two groups was determined by paired Student's t-test, with statistical significance set to p < 0.05. Each experiment was repeated at least three times on separate occasions.