Cigarette Smoke or Motor Vehicle Exhaust Exposure Induces PD-L1 Upregulation in Lung Epithelial Cells in COPD Model Rats

Abstract A high smoking-independent chronic obstructive pulmonary disease (COPD) prevalence is observed in lung cancer patients. However, the underlying connection between these two diseases still remains unclear. Cigarette smoking and ambient air pollution are common risk factors for COPD and lung cancer. In this study, we established rat COPD model through exposure to cigarette smoke (CS) or motor vehicle exhaust (MVE). The model rats developed COPD-like phenotypes, manifested as lung functions decline, lung inflammation, emphysema-like alveolar enlargement and airway remodeling. The programmed death-ligand 1 (PD-L1), a factor contributing to immune escape of tumor cells, was overexpressed in lungs from COPD model rats, though more severe COPD phenotypes did not bring with further PD-L1 overexpression in lung. The upregulations of proinflammatory cytokines and PD-L1 were also observed in cultured human bronchial epithelial cells BEAS-2B upon treatment with cigarette smoke extract (CSE) or diesel-related particulate matter 2.5 (PM2.5, SEM1650b). The inflammatory cytokines produced in BEAS-2B cells reflected the PD-L1 levels. Furthermore, ERK1/2, a kinase mediating PD-L1 upregulation in premalignant bronchial cells or NSCLC cells, and STAT1/3, which was reportedly associated with PD-L1 expression in lung tumors, were activated in COPD rats’ lungs or in BEAS-2B cells treated with CSE or PM2.5. Therefore, we proposed that inflammation associated PD-L1 overexpression in airway epithelial cells could be the underlying factor facilitating lung cancer incidence in COPD.


Introduction
Several epidemiological studies highlighted the link between chronic obstructive pulmonary disease (COPD) and lung cancer, including the association between airflow limitation and increased lung cancer incidence and mortality [1]. Further study indicated the prevalence of COPD in lung cancer cases was independent of age, sex or smoking history [2]. The main risk factors for COPD include but are not limited to cigarette smoke (CS) exposure and ambient air pollution, both of which are common risk factors for lung cancer. However, the operative mechanisms between COPD prevalence and lung cancer are remained elusive [3][4][5].
Transmembrane protein PD-L1 is overexpressed in many cancer types including lung cancer. Programmed death-ligand 1 (PD-L1) inhibited the functions of activated cytotoxic CD8+ T cells through interaction with PD-1 on cell surface of T cells [6]. The overexpression of PD-L1 was supposed contribute to lung cancer initiation by facilitating immune escape in premalignant human bronchial epithelial cells [7]. Clinical study showed that immunotherapy targeting on immune-checkpoint proteins PD-1/PD-L1 displayed promising clinical effects on non-small cell lung cancer [8]. The associations among COPD, PD-1/PD-L1 axis activation and lung cancer therapy were revealed in clinical trial that immune checkpoint inhibitors (ICIs) displayed better therapeutic outcome in lung cancer patients underlying COPD than those without COPD [9]. It was proposed that chronic lung inflammations commonly seen in COPD patients enhance the risk of tumorigenesis by weakening immune surveillance in lung. This hypothesis was partially verified by the findings that lipopolysaccharide-mediated chronic inflammation upregulated PD-L1 expression through TLR4/ERK axis in lung cancer cells, or activated PD-1/ PD-L1 axis in lung cancer model mice lung in which therapy coupled with anti-PD-1 treatment elicited better anti-tumor efficacy [10,11].
As the main risk factors for COPD, both CS and MVE exposure induced lung inflammation in COPD rodent animal models [12,13]. However, the PD-L1 overexpression in COPD rodent animals has not been fully studied. Tobacco smoke was reported induce PD-L1 increases in lung of A/J mice, a mouse strain having comparatively high lung tumor rate upon carcinogen treatments [14]. Furthermore, although the air pollutant particles activated some proteins, such as ERK1/2 and STAT, which are able to promote PD-L1 expression in bronchial epithelial cells or lung tumor cells, whether air pollutants affected lung PD-L1 expression in COPD patients or in COPD rodent animals are not fully explored [7,[15][16][17].
In this study, we established rat COPD models through exposure to CS or MVE mimicking air pollution and reported that PD-L1 was upregulated in lung tissues from the two types of COPD model rats. PD-L1 levels in cultured human bronchial epithelial cells BEAS-2B were increased by CSE or diesel-related PM 2.5 . In lung tissues and cells, PD-L1 was upregulated along with production of proinflammatory cytokines and the activation of ERK and STAT. Our findings gave a clue that inflammation associated PD-L1 upregulation induced by CS or MVE in lung epithelial cells might contribute to the increased lung cancer incidence in COPD patients.

Animals
Male SD rats (8-12 weeks old) were purchased from the Guangdong Medical Laborator y Animal Center (Guangzhou, China) and housed in the specific pathogen-free facilities accessible to food and water freely. The experimental methods were approved by the Ethics Committee for Animal Experiments of the General Hospital of Southern Theater Command of PLA. The rats were exposed to CS or MVE according to previous methods [12,13] and detailed steps described in Supplemental Materials.

Lung function measurement
The rats exposed to CS or MVE were subjected to lung function measurement by using a Forced Pulmonary Maneuver System (Buxco Research Systems, Wilmington, NC) following a protocol described previously [13]. The values of parameters airway resistance, chord compliance, total lung capacity and the ratio of FEV50/FVC were determined.

Cell culture
The human bronchial epithelial cells BEAS-2B were subjected to treatments with CS extract (CSE, 0-20%) or diesel-related PM 2.5 Standard Reference DEP material -SRM 1650 b (0-300 µg/ml), respectively. Cell viability was determined by using a CCK-8 test kit and presented as percentage of control.

Quantitative RT-PCR assay
The relative mRNA levels of rat or human PD-L1 were quantified, normalized to GAPDH and expressed as percentage of control.

Western blotting
The protein levels of ERK1/2, p-ERK1/2, STATs and p-STATs in total proteins extracted from cultured cells or lung tissues were determined by western blotting and semi-quantitated by using the software Image J.

Immunofluorescence microscopy
The distribution of PD-L1 in lung tissue sections was indicated by immunofluorescence staining with a PD-L1 antibody.

ELISA
The ELISA kits from eBioscience (San Diego, CA) were used to measure the levels of rat IL6 (88-50,625) and rat TNF-α (88-7,340) in BALF and the levels of human IL6 (88-7,066) and human IL8 (88-8,086) in cell culture medium according to the manufacturer's instructions. The levels of Muc5ac in BALF were measured by using a capture antibody for Muc5ac (sc-21701, Santa Cruz Biotechnology, Santa Cruz, CA) and corresponding detection antibody obtained from KPL, Inc. (Gaithersburg, MD) according to protocol described previously [18].

Statistical analysis
All in vitro experiments were carried out in triplicate and repeated three times. Comparison between two groups was determined by Students t-test, one-way ANOVA further followed by Bonferroni post-hoc test was adopted for comparison in experiments with ≥ 3 groups. SPSS version 16.0 (SPSS Inc., Chicago, IL) and Graphpad prism version 5.0 (Graphpad Inc., La Jolla, CA) were used for statistics and graphing, respectively. Data were presented as mean ± SEM. N denotes the number of repeats in cell experiments or the number of animals. p Values less than .05 or .01 were considered as statistically significant or extremely statistically significant respectively in all cases.
A detailed description of the study methods can be found in the Supplementary Information section.

COPD-like symptoms developed in rats exposed to CS or MVE
The COPD-like symptoms were observed in rats exposed to CS or MVE. The lung functions were declined in rats exposed to CS or MVE indicated by the significantly increased parameters of resistance, chord compliance and total lung capacity and the decreased ratios of FEV50/FVC when compared with those rats in respective control group (Figure 1(A)). The mean linear intercept values, a histological measure of emphysema-like alveolar enlargement, were also increased in rats exposed to CS or MVE (Figure 1(B)). Furthermore, significant differences were observed in parameters resistance, total lung capacity, FEV50/FVC and mean linear intercept value between CS and MVE exposed rats, revealed more severe lung injury in CS exposed rats.
The inflammatory responses developed in lungs from rats exposed to CS or MVE, manifested by increased leukocytes and pro-inflammatory cytokine levels in BALF. When compared CS-with MVE-exposed rats, we found although CS-exposed rats had much more inflammatory cells in BAL fluid than MVE-exposed rats (p < .01), the levels of proinflammatory cytokines IL-6 (p < .05) and TNF-α (p > .1) in BAL fluid did not show similarly obvious difference between Figure 1. COPD like lung function loss was found in rats exposed to cigarette smoke or motor vehicle exhaust. male sD rats were exposed to cigarette smoke (Cs) for 60 d in a whole body exposure chamber or motor vehicle exhaust (mve) in an exposure system for 24 weeks. (a) the values of lung function parameters, including airway resistance, chord compliance, total lung capacity and the ratio of fev50/fvC were determined. (b) representative h&e staining images of rat tissues sections with or without Cs or mve exposure treatment. each calculation for mean linear intercept (lm) in lung was based on six rats and five representing slides across different regions of the lung from each rat. Data are represented as mean ± sem (n = 6). **p < .01 for Cs or mve group versus respective Ctl group; # p < .05 and ## p < .01 for Cs group versus mve group. the two COPD models (Figure 2(A,B)). Mucus hypersecretion in model rats was characterized by goblet cell hyperplasia indicated by positive PAS staining and upregulated Muc5ac levels in BALF. In the two COPD models, CS exposure induced more obvious mucus hypersecretion than MVE exposure ( Figure 2(C,D)). Furthermore, the increased ratios of wall area/total bronchial area were seen in both COPD models, though no difference existed between them ( Figure  2(E)). It meant that the two COPD models presented the same extent in airway remodeling.

PD-L1 upregulations in lungs exposed to CS or MVE
The lysates of lung tissues were subjected to quantitative RT-PCR or western blotting for measurements of mRNA or protein levels of PD-L1. As shown in Figure 3(A,B), both mRNA and protein levels of PD-L1 were upregulated in rats' lungs exposed to CS or MVE. However, concerning the changes of PD-L1 mRNA or protein levels, no difference existed between the two COPD models. Immunofluorescence staining with antibody against PD-L1 indicated PD-L1 predominantly expressed in bronchial epithelial cells. When compared with the normal rats, enhanced immunofluorescence staining was observed in sections from COPD model rats' lungs ( Figure 3(C)).

Both CSE and PM 2.5 induced inflammatory responses in lung epithelial cells
In order to investigate the effects of CS or MVE exposure on bronchial epithelial cells, we treated the cultured immortal normal human bronchial epithelial cells BEAS-2B with cigarette smoke extracts (CSEs) or diesel-related PM 2.5 which mimicked the effects of MVE. Both CSE and PM 2.5 in high levels, namely more than 10% for CSE and more than 200 µg/ml for PM 2.5 , reduced cell viability for more than 50% (Figure 4(A,B)). Therefore, we used 2% CSE or 100 µg/ml PM 2.5 in cell treatments hereafter. In these dosages, both CSE and PM 2.5 induced inflammatory responses in BEAS-2B cells manifested by upregulated cytokine levels of IL-6 and IL-8 in cell culture medium. Moreover, CSE induced more robust productions of IL-6 and IL-8 than PM 2.5 (Figure 4(C)).

PD-L1 upregulations in bronchial epithelial cells treated with CSE or PM 2.5
Both CSE and PM 2.5 enhanced PD-L1 expression in mRNA and protein levels in BEAS-2B cells. The mRNA and protein levels of PD-L1 were higher in cells exposed to CSE than those cells exposed to PM 2.5 ( Figure 5(A,B)). The   . PD-l1 was increased in lung exposed to Cs or mve. the relative mrna levels of PD-L1 in lung tissues were presented as percentage of the control group. (b) Protein levels of PD-l1 in lung tissues were determined by using western blotting. the blotted bands were relatively semi-quantitated to internal control (GaPDh) by densitometry and presented as relative amount to control groups. Data are presented as mean ± sem. **p < .01 for Cs or mve group versus respective Ctl group. (C) representative images from lung tissue sections with immunofluorescence staining of PD-l1. arrows indicated the cells with relatively brighter fluorescence staining than average. Protein levels of PD-l1 were determined by western blotting. the blotted bands were relatively semi-quantitated to internal control (GaPDh) by densitometry and presented as percentage amount to control cells. Data are presented as mean ± sem (n = 6). **p < .01 for Cse or Pm 2.5 group versus respective Ctl group; # p < .05 and ## p < .01 for Cse group versus Pm 2.5 group. (C) representative images from immunofluorescence staining of PD-l1 in cultured beas-2b cells.
upregulations of PD-L1 induced by CSE or PM 2.5 were confirmed by the immunofluorescence staining ( Figure 5(C)).

In vitro and in vivo activation of ERK and STAT upon CS/CSE or MVE/PM 2.5 exposure
The signaling mediated by ERK or STAT was activated in lung from COPD model rats. Here we found that both CS and MVE induced phosphorylations of ERK1/2, STAT1 and STAT3 in lung tissues from the two rat COPD models in which more obvious ERK1/2 phosphorylation found in CS-induced model (Figure 6(A)). In cultured cells, both CSE and PM 2.5 induced very obvious ERK1/2 phosphorylation. However, CSE preferred to induce STAT3 but not STAT1 phosphorylation, whereas PM 2.5 inclined to induce STAT1 phosphorylation in BEAS-2B cells (Figure 6(B)).

Discussion
The prevalence of COPD in lung cancer patients is greater than that in both smokers and randomly recruited control group [2]. In order to elucidate the association between COPD independent of smoking and lung carcinogenesis, we established COPD model in rats by CS or MVE exposure and treated airway epithelial cells with CSE or diesel-related PM 2.5 . PD-L1 upregulation was observed in COPD model rats' lung, accompanied with lung function decline and lung inflammation. ERK1/2 and STAT1/3, both of which reportedly modulated PD-L1 expression in lung inflammation and lung cancer cells, were activated in COPD rats' lung. Inflammatory responses, PD-L1 upregulation and activations of ERK1/2 and STAT1/3 were reproduced in airway epithelial cells upon CSE or PM 2.5 treatment. Our results suggested a possible interpretation for the comorbid of COPD and lung cancer that overexpression of PD-L1 under inflammatory condition might contribute to the lung cancer initiation in COPD patients.
The increase of PD-1/PD-L1 axis in lung cancer generally exhibited the upregulated expressions of PD-1 in T cells and of PD-L1 in tumor cells which lead to immune escape of tumor cells [19,20]. The main immunological dysfunction in COPD patients' lung is T-cell exhaustion due to the binding of immune checkpoints on T-cells. The interaction of PD-1/PD-L1 established an immunosuppressive microenvironment facilitating the tumor initiation, development and progression [21]. However, despite the increased CD8 + PD1 levels in COPD patients' lung [22], at least one study mentioned no difference found in CD4/8 + T cells PD-1 levels in resected tumor tissues between COPD and non-COPD NSCLC patients [9]. PD-L1 overexpression alone is unable to induce cancer genesis, unless with the presence of carcinogen [14]. Therefore, PD-Ll overexpression in lung epithelial cells is critical for establishment of immune escape contributing to increased lung cancer incidence in COPD patients.
By comparing the status of lung functions decline and lung PD-L1 upregulations between the two COPD models, we noticed that more severe lung functions decline in CS model did not bring higher PD-L1 expressions in lung (Figures 1 and 3). Although the PD-L1 upregulation occurred in COPD, the extent of lung function decline was not correlated with PD-L1 level. Concerning the PD-L1 overexpression in lung cancer, one point of view supposed the PD-L1 overexpression was modulated by crosstalk between NSCLC and activated macrophages [17]. In a study focusing on resected lung cancer tissues, tumor PD-L1 expression was associated with the tumor infiltrated macrophages which accumulated around PD-L1-positive carcinoma cell areas in early lung adenocarcinoma patients. The cytokines secreted by tumor-associated macrophages, including TGF-β1, TNF-α, IL-10, IL-1α, IL-1β and IL-27, are supposed to be extrinsic regulators of tumor PD-L1 expression and contribute to tumor development in early stage [23]. However, in our study, the PD-L1 levels showed no difference between the two COPD model, though CS exposed rats had much more airway macrophages. These results reflected that macrophages infiltration in lung might correlate with PD-L1 upregulation but more macrophages infiltration did not lead to higher lung PD-L1 expression in COPD (Figures 1-3).
A prospective study supposed the elevated serum IL-6 level mean poor response to immune checkpoint therapy [24]. In epithelial inflammation, PD-L1 upregulated the production of proinflammatory cytokines [25]. We would like to evaluate the association between BAL fluid cytokine increases and PD-L1 upregulation. Both Inflammatory cytokines increase and PD-L1 upregulation were found in COPD lungs. When compared the two COPD models, Figure 6. Phosphorylation of erK and stat in lung or beas-2b cells in respond to Cs/Cse or mve/Pm 2.5 treatments. Western blotting was used to determine the levels of total and phosphorylated erK, stat1 and stat3 proteins in (a) lung tissues exposed to Cs or mve, or in (b) beas-2b cells treated with 2% Cse or 100 µg/ml Pm 2.5 . for lung tissues, samples were randomly selected from six rats per group. the results for cell lysates were from one representative experiment performed in triplicates and repeated for three times. relative semi-quantitation of the bands to internal control (GaPDh) was performed by densitometry. the ratios of phosphorylated protein to total protein were calculated and presented as percentage of respective control. Data are presented as mean ± sem. *p < .05 or **p < .01 for Cs/Cse or mve/Pm 2.5 group versus respective Ctl group; ## p < .01 for Cs/Cse group versus mve/Pm 2.5 group.
higher IL-6 levels or relative higher TNF-α and PD-L1 levels presented in CS model lung, though no statistically significant difference found in TNF-α and PD-L1 between the two models (Figures 2(B) and 3). More consistent changes between proinflammatory cytokines increases and PD-L1 upregulation were found in culture cells, in which more cytokines production meant higher PD-L1 expression (Figures 4(C) and 5). Therefore, inflammatory cytokines production in COPD lung meant PD-L1 upregultion. But the levels of proinflammatory cytokines in BAL fluid were not correlated with the PD-L1 expression level. The proinflammatory cytokine levels in airway epithelial cells might be more suitable to predict the lung PD-L1 level than other parameters in COPD.
The phosphorylation of ERK was triggered by CS/CSE or diesel-related PM 2.5 in in vivo and in vitro study [26][27][28]. A speculate on investigating the relationship between inflammation and tumor development suggested that the activation of TLR4/ERK/PD-L1 axis contribute to the NSCLC initiation [10]. The ratio of p-ERK/ERK, a hallmark of lung inflammation, was increased in COPD rats' lungs and in cultured bronchial epithelial cells upon CSE or diesel-related PM 2.5 treatments ( Figure 6). Unlike previously report in which diesel related PM 2.5 induced phosphorylation of ERK in bronchial epithelial cells in dosage bringing cell cycle arrest [29], our results showed that both CSE and diesel related PM 2.5 activated ERK and upregulated PD-L1 in a dosage without inducing obvious cell viability loss but enough to induce inflammatory responses (Figure 4(A) and 5). It was supposed that the carcinogens like polycyclic aromatic hydrocarbon (PAHs) in CS or PM 2.5 -bound PAHs induced ERK activation and cell transformation in bronchial epithelial cells [15,30]. However, it should be noted that higher p-ERK/ERK ratio in lung did not bring further PD-L1 overexpression when compared the two COPD models ( Figure 6). In culture cells received CSE or PM 2.5 treatment, more robust PD-L1 increase even emerged in CSE-treated cells in spite of less p-ERK augment. Thus, we proposed p-ERK is correlated with PD-L1 upregulation in COPD lung epithelial cells. But PD-L1 expression is probably modulated by elements unlimited to p-ERK.
We found both CS and MVE induced similar phosphorylation of STAT1 and 3 in lungs. However, in cultured bronchial epithelial cells BEAS-2B, CSE or diesel-related PM 2.5 activated STAT family members preferentially, namely STAT3 for CSE and STAT1 for diesel-related PM 2.5 . The bias between in vivo and in vitro effects could be attributed to the inability of CSE or diesel-related PM 2.5 to contain all components in CS or MVE. Concerning our results, it is not known whether the difference in PD-L1 levels between CSE and diesel-related PM 2.5 treatment was due to alternative activation of STAT family members. Conflicting results have been obtained regarding the involvement of p-STAT in modulating PD-L1 expression in lung epithelial cells. Both IFNγ/STAT1/STAT3 and JAK/STAT1/STAT3 signaling axis potentiated PD-L1 overexpression upon different stimuli [17,31,32]. STAT3 activation also participated in eliminating PD-L1 expression [33]. That means the regulation of STATs on PD-L1 could be complex. Further research is needed to explore the effects of STAT phosphorylation on PD-L1 upregulation in COPD developed from various risk factors.
In conclusion, our study presented the effects of CS and air pollutant on establishing COPD animal models with simultaneous PD-L1 upregulation in lung. However, more severe COPD phenotypes did not bring higher PD-L1 expression in lung. The inflammatory cytokine levels in airway epithelial cells were likely to reflect the PD-L1 level. Our hypothesis illustrated in Figure 7 suggested that PD-L1 upregulation was accompanied by activation of ERK1/2 and STAT family members and might potentially facilitate the high prevalence of lung cancer in COPD patients.

Ethics approval
The experimental methods were approved by the Ethics Committee for Animal Experiments of the General Hospital of Southern Theater Command of PLA.

Consent for publication
All authors have read the manuscript and approved the submission.

What is the Central question of this study?
Our study presented the effects of cigarette smoke and air pollutant on establishing COPD animal models with simultaneous PD-L1 upregulation in lung and suggested the premalignant PD-L1 upregulation in lung epithelial cells could be the link between COPD prevalence and lung cancer.

What is the main finding and its importance?
PD-L1 upregulation in airway epithelial cells exposed to various COPD risk factors might be associated with activation of ERK1/2 and alternative STAT family members.