The elevated CXCL5 levels in circulation are associated with lung function decline in COPD patients and cigarette smoking-induced mouse model of COPD

Abstract

Introduction

C-X-C motif chemokine 5 is primarily chemotactic for neutrophils and previously shown to increase in the bronchoalveolar lavage fluid of patients with chronic obstructive pulmonary disease. However, whether C-X-C motif chemokine 5 levels correlate with lung function decline in patients or mouse model of chronic obstructive pulmonary disease was not clear.

Methods

The mouse model was induced by cigarette smoke exposure. Plasma/serum and bronchoalveolar lavage fluid were obtained from patients and mouse model of chronic obstructive pulmonary disease; C-X-C motif chemokine 5 levels were assessed and correlated with lung functions and granulocyte-colony stimulating factor levels, respectively.

Results

The C-X-C motif chemokine 5 levels increased and correlated to granulocyte-colony stimulating factor levels in both plasma/serum and bronchoalveolar lavage fluid obtained from patients and mouse model of chronic obstructive pulmonary disease. Circulating levels of C-X-C motif chemokine 5 correlated to lung functions decline in patients and mouse model.

Conclusions

Granulocyte-colony stimulating factor might coordinate with C-X-C motif chemokine 5 in the pathogenesis of neutrophilic inflammation in chronic obstructive pulmonary disease. Circulating C-X-C motif chemokine 5 might serve as a potential blood-based biomarker to add additional modest predictive value on the preliminary screening and diagnosis of chronic obstructive pulmonary disease.

Key messages

• Circulating C-X-C motif chemokine 5 might serve as a potential blood-based biomarker to add additional modest predictive value on the preliminary screening and diagnosis of COPD.

• Granulocyte-colony stimulating factor might coordinate with C-X-C motif chemokine 5 in the pathogenesis of neutrophilic inflammation in chronic obstructive pulmonary disease.

Introduction

Chronic obstructive pulmonary disease (COPD) is a leading cause of morbidity and mortality with a global prevalence of 11.7% [1] and three million deaths annually [2], and has become the third leading cause of death worldwide between 2000 and 2016 [3]. However, the awareness of COPD in some areas were surprisingly low. For example, in China only 2.6% spirometry-defined COPD patients were aware of their condition, and only 12% of people with COPD reported a previous pulmonary function test; moreover, the proportion of individuals with a history of pulmonary function testing was significantly higher in urban residents than in rural residents [4]. Hence, the identification of an easily accessible biomarker (such as a blood-based biomarker) might assist in the preliminary screening and diagnosis of COPD.

COPD is characterized by persistent respiratory symptoms and airflow limitation, and associated with abnormal inflammatory immune response of the lung to inhalation of cigarette smoke. Cigarette smoke activates innate immune cells such as macrophages and airway epithelial cells to release multiple chemotactic factors to recruit neutrophils and inflammatory monocytes to the lungs. This innate immune response plays a key role in the pathogenesis of COPD. Substantially increased numbers of neutrophils, macrophages were observed in the lungs of COPD patients [5,6] and COPD mouse models [7,8]. In COPD patients, inhibition of neutrophil migration by using roflumilast, an oral phosphodiesterase type 4 inhibitor, benefited patients by improving their forced exiratory volume in 1 second (FEV₁) and reducing their exacerbation frequency [9–11]. In cigarette smoking treated mice, the recruitment of neutrophils by different stimuli on cell surface receptors, such as formyl-peptide receptor [12,13], adenosine triphosphate receptor [14–17], and receptors for advanced glycation end products [18] were necessary to develop pulmonary emphysema and airway remodeling in COPD murine model; however, severe reduction in number and function of peripheral T cells did not influence the development of pulmonary changes induced by cigarette smoke [19]. These studies together suggest that the innate immunity driven by the neutrophils might be a leading actor in the early development of COPD, while the adaptive immune response may play a role at later stages.

Therefore, targeting the recruitment process of neutrophils might shed light on the identification of biomarkers in COPD. One of the most important chemokines to recruit neutrophils in COPD is C-X-C motif chemokine 5 (CXCL5, also called epithelial-cell-derived neutrophil activating peptide-78, ENA-78, with the murine homologue CXCL5/LIX).

CXCL5 has been demonstrated as a strong chemotactic agent to elicit neutrophil accumulation both in vitro [20] and in vivo [21]. It was thought that the lung resident cells accounted for the release of CXCL5 in cigarette smoking-induced lung inflammation [22], while the platelets contributed to the homeostatic CXCL5 in blood [23]. In the rodent lung exposed to LPS, the activation of primary alveolar epithelial type II cells was important for the production of CXCL5 [24]. The mRNA expression or levels of CXCL5 can be induced either in the lungs or in the bronchoalveolar lavage fluid (BALF) of mice treated with cigarette smoke exposure for 2 weeks [25], 8 weeks [26], or 6 months [27]. In COPD patients, CXCL5 levels in the BALF have been described to be increased [28]. Similarly, the amount of CXCL5 derived from the cells in BALF was higher in COPD patients compared to healthy non-smokers [29]. Furthermore, CXCL5 expression was significantly increased in epithelial cells during exacerbations of COPD [30]. The increased expression profile of CXCL5 indicates it as a possible biomarker in COPD, however, it has not been demonstrated whether the circulation or BALF level of CXCL5 in either COPD patients or COPD murine model associates with their lung function decline.

Here, we aimed to evaluate the plasma/serum and BALF levels of CXCL5 in COPD patients and cigarette smoking induced mouse model of COPD, and assess the correlation of CXCL5 levels to the changes of lung functions and the neutrophil priming.

Methods

Criteria for inclusion and exclusion

All protocols was approved by the Chinese Ethics Committee of Registering Clinical Trials (approval number: ChiCTR1900022271), and conducted in accordance with the amended Declaration of Helsinki. From January 2016 to December 2018, 30 healthy volunteers and 91 patients diagnosed as COPD according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria were recruited at the West China Hospital, and all provided their written informed consent. The basic information including gender, age, body mass index (BMI) and smoking history, as well as the medical history in details and the chest CT taken in recent 5 years were reviewed, the follows were excluded: (1), suffering from the diseases other than COPD or acute exacerbation of COPD known to affect plasma or BALF CXCL5, including asthma, lung cancer and acute respiratory distress syndrome; (2), previously receiving standard COPD-related therapies such as inhaled corticosteroids, bronchodilator etc.; (3), unable or unwilling to perform lung function tests; (4), unable or unwilling to cooperate with a doctor.

Collection of human lung function, plasma and BALF

All subjects received a standard lung function test according to the European Respiratory Society guidelines [31]. Patients showed a ratio of forced expiratory volume in the first second to forced vital capacity (FEV₁/FVC) below 70% after brochodilation and an increase in FEV₁ below 12% after inhalation of 200 μg salbutamol were diagnosed as COPD. Plasma samples were obtained from 24 healthy volunteers and 63 COPD patients the morning after enrollment following standard protocol. Briefly, venous blood was collected from the median cubital vein and plasma was separated and stored immediately at –80 °C for further measurements. BALF samples were obtained from 24 healthy volunteers (with 17 overlaps receiving plasma test) and 28 COPD patients following standard protocol. Briefly, a bronchofiberoscope was wedged in the segmental bronchus of the right middle lobe, and four portions of sterile 0.9% saline solution were instilled. Aliquots of BALF samples were collected and stored at –80 °C for further measurements.

Animals

Aged-matched male C57Bl/6J mice were obtained from GemPharmatech Co. Ltd. (Nanjing, Jiangsu, China) and bred in the specific pathogen free environment at the Animal Experimental Center of West China hospital. All animal experiments were conducted in accordance with the Animal Ethics Committee of West China hospital (approval number: 2018049A).

Establishment of COPD mouse model

Restrain of mouse

Mouse model of COPD was established using a nose-only cigarette smoking exposure method as reported previously [32]. Mouse was restrained in a custom-designed nose-only exposure tubes (China Pattern number: 201810963763.4) with a restraint stopper to close the tube at the back (Supplementary Figure 1(A,B)). The nose of mouse was exposed to the smoke or air through a one-way flow opening in the front of nosepiece (Supplementary Figure 1(C,D)). The restrained mice were then mounted to a smoking chamber constructed with a flow-in layer in the middle and the flow-out layers at two sides (Supplementary Figure 1(E,F)).

Cigarette smoking

Cigarette smoke was generated by burning of commercially available cigarettes (Marlboro, Philips Morris, USA; 1.0 mg nicotine and 11 mg tar per cigarette) on a Baumgartner-Jaeger CSM2082i automated cigarette smoking machine (CH Technologies, West-Wood, NJ, USA). The smoke was diluted with fresh air using a fixed-rate pump (CH Technologies, West-Wood, NJ, USA). The mice were subjected to 75 min exposure of smoke generated from ∼25 cigarettes for each session, two sessions a day (morning and afternoon, separated by a recovery period), 5 days a week, for total 12 weeks. The control mice were exposed to filtered air by using the same protocol.

Determination of smoke dose

The concentration of smoke was determined by the ratio of total particulate matter weight in smoke to the total flow volume (mg/m³). The dose of smoke was chosen by performing 12 weeks dose–response experiments using different levels of smoke exposure. The selected dose induced significant increases in the total lung capacity (TLC) and functional residual capacity (FRC), and significant decrease in the ratio of forced expiratory volume in the first 100 ms to forced vital capacity (FEV₁₀₀/FVC) without inducing a majority of death in the mice.

Mouse lung functions

Mice were anesthetized with 50 mg/kg sodium pentobarbital (Sigama-Aldrich) via intraperitoneal injection, then tracheostomized and placed in a forced pulmonary maneuver system (Buxco Research Systems). To acquire the lung function of mice, the Boyle’s Low FRC, Quasistatic PV and Fast Flow Volume maneuver were performed with the Buxco system [33]. The FRC was determined with the Boyle’s Law FRC maneuver; the TLC, residual volume (RV), vital capacity (VC), inspiratory capacity (IC), chord compliance between 0 and 10 cmH₂O (Cchord) and chord compliance at 50% of VC (Cfvc50) were acquired by the Quasistatic PV maneuver; the forced vital capacity (FVC), forced expiratory volume (FEV) at 100 ms (FEV₁₀₀), FEV₂₀₀, peak expiratory flow (PEF) and flow at 75% of FVC (FEF_75%) were recorded with the Fast Flow Volume maneuver. Each maneuver was repeated at least three times.

Collection of mouse serum and BALF

Mice were sacrificed by exsanguination from the right ventricle to allow the collection of blood in coagulation-promoting tubes. After centrifuge at 3000 g for 10 min, the serum was collected and immediately frozen at −80 °C for further assessment. To collect BALF, the right lung was lavaged with 0.5 mL sterile ice-cold phosphate buffered saline (PBS) supplemented with protease inhibitors (MCE) for three times. The BALF was then centrifuged at 1000 g at 4 °C for 5 min. The supernatant was immediately frozen at −80 °C for further assessment. The cell pellet was treated with lysis buffer, and resuspended with 500 μL ice-cold PBS. The total cell counts in BALF were performed with a hemocytometer (ORFLO). The differential cell counts were determined by cell smears stained by Wright–Giemsa method. At least 200 cells from five different random views in each slice were recorded by a microscope (Nikon). The percentage of neutrophils, macrophages, and lymphocytes were analyzed, respectively, by two independent experienced investigators who were blinded to the details of experiments.

Mouse lung histology

The left lung without lavage was fixed in 4% phosphate buffered paraformaldehyde (pH 7.4), embedded in paraffin, sliced into 4-µm thick sections The paraffin sections were then stained with hematoxylin and eosin solution for morphological examination, or Alcian blue-periodic acid Schiff stain to evaluate mucus secretion, or Masson’s trichrome stain to evaluate collagen deposition. Alveolar enlargement were determined by the average linear intercept, the ratio of total length of alveoli to the number of alveoli per field. Histologic inflammatory scores were obtained by an experienced histologist blinded to experimental details, who took into accounts of perivascular infiltration, peribronchial infiltration, parenchymal infiltration, and epithelial damage and gave an evaluation for each index. For each evaluation index, 0 point indicates no sign of disease whereas 5 point represents profound inflammation, thus the inflammation degree was presented as a total score point between 0 and 20 as described before [34].

Measurement of CXCL5 and G-CSF in supernatants

The levels of CXCL5 and granulocyte-colony stimulating factor (G-CSF) in the plasma/serum and BALF obtained from human and mouse were measured using the Magnetic Luminex Screening Assay (R&D systems) on a Bio-Plex 200 system (Bio-Rad), following the manufacture’s instruction. Briefly, beads coated with capture antibodies against known cytokines were used simultaneously in the multiplex assay.

Statistical analysis

The data were analyzed by SPSS Statistics 19 (IBM) or GraphPad Prism 7.00 (GraphPad Software Inc.), the figures were created by GraphPad Prism 7.00. For comparison analysis, Shapiro-Wilk normality test or KS normality test were used to determine whether the data obey the normal distribution, then the data were tested by unpaired t-test or Mann-Whitney test according to their distributions in GraphPad Prism 7.00. The correlations analysis were performed by using Pearson’s partial correlation test to correct for age, gender, BMI and smoking history, then multiple linear regression analysis was conducted to confirm the aforementioned relationship in SPSS Statistics 19. All data were presented as mean with 95% confidence interval (CI), and p value <.05 was considered statistically significant.

Results

Clinical characteristics of subjects

Thirty healthy controls and 91 COPD patients were enrolled. Among 91 COPD patients, there were 17 GOLD-1, 37 GOLD-2, 24 GOLD-3 and 13 GOLD-4 as classified by the GOLD criteria [1]. The demographic and clinical characteristics including age, gender, BMI, smoking status, lung function were listed in Table 1. No significant differences in gender were observed. However, COPD groups were generally older than healthy group, the BMI and pack-year for ever-smoker in COPD patients receiving plasma test were different from healthy control. In addition, the majority of smoker were males. As compared to healthy controls, FVC, FEV₁/FVC and the percentage of FEV₁ to predicted FEV₁ (FEV₁%Pred) were significantly decreased in COPD patients.

Table 1.

Characteristics of healthy subjects and COPD patients.

Characteristics	Healthy controls receiving plasma and BALF test		Healthy controls, N = 30	COPD patients receiving plasma test		COPD plasma test, N = 63	p Value and differences between means/medians with 95% CI	COPD patients receiving BALF test		COPDBALF test, N = 28	p Value and differences between means/medians with 95% CI
	Nonsmokers, N = 19	Smokers, N = 11		Nonsmokers, N = 14	Smokers, N = 49			Nonsmokers, N = 13	Smokers, N = 15
Age, years	53.42 (50.66 – 56.19)	50.55 (45.76 – 55.33)	52.37 (50.01 – 54.73)	62.43 (56.01 – 68.84)	64.63 (61.85 – 67.42)	63.97 a (61.79 – 66.16)	p < .0001 13.5 (9 – 16)	67.54 (61.69 – 73.38)	62.47 (57.76 – 67.17)	64.84 a (61.55 – 68.13)	p < .0001 12.47 (8.487 – 16.46)
Gender, male/female	7/12	10/1	17/13	4/10	48/1	52/11	NA	7/6	15/0	22/6	NA
BMI, kg/m2	25.09 (23.26 – 26.92)	25.78 (24.31 – 27.25)	25.34 (24.12 – 26.56)	25.06 (20.71 – 29.41)	23.28 (22.29 – 24.27)	23.68 a (22.51 – 24.85)	p = .0293 1.168 (0.1651 – 3.325)	24.93 (22.79 – 27.07)	23.63 (21.79 – 25.47)	24.23 (22.91 – 25.56)	p = .2136 1.106 (−0.6553 – 2.868)
Ratio of current/former smokers	NA	6/5	NA	NA	17/32	NA	NA	NA	5/10	NA	NA
Pack-years for ever-smoker	NA	20.45 (7.366 – 33.52)	NA	NA	37.9 a (31.56 – 44.25)	NA	p = .0189 17.46 (2.988 – 31.92)	NA	36.42 (24.22 – 48.61)	NA	p = .0673 15.97 (1.231 – 33.17)
FVC, L	3.363 (2.969 – 3.757)	3.97 (3.447 – 4.493)	3.586 (3.27 – 3.901)	2.278 (1.881 – 2.675)	2.99 (2.761 – 3.22)	2.832 a (2.624 – 3.04)	p < .0001 0.91 (0.44 – 1.16)	2.595 (2.084 – 3.106)	3.393 (3.035 – 3.752)	3.023 a (2.698 – 3.347)	p = .017 0.275 (0.08 – 0.95)
FEV1/FVC, %	87.51 (82.64 – 92.37)	84.6 (77.58 – 91.62)	86.44 (82.65 – 90.23)	54.16 (49.15 – 59.17)	50.45 (46.67 – 54.22)	51.27 a (48.17 – 54.38)	p < .0001 34.39 (29.69 – 40.2)	57.98 (51.17 – 64.8)	53.05 (45.11 – 60.99)	55.34 a (50.3 – 60.39)	p < .0001 29.73 (25.01 – 36.23)
FEV1%Pred, %	99.29 (89.91 – 108.7)	96.53 (84.79 – 108.3)	98.28 (91.37 – 105.2)	58.84 (50.58 – 67.09)	57.51 (51.04 – 63.98)	57.8 a (52.55 – 63.06)	p < .0001 40.47 (31.59 – 49.35)	53.58 (39.93 – 67.23)	48.82 (37.17 – 60.47)	51.03 a (42.69 – 59.36)	p < .0001 47.25 (36.72 – 57.77)

Data are presented as mean (lower 95% CI of mean to upper 95% CI of mean); unpaired t-test or Mann-Whitney test were used to test the significance of the difference between two groups after normal distribution examination by Shapiro-Wilk normality test.

^a Significantly different from healthy control, p < .05 was considered to be statistically significant, the p value and the differences between means/medians with 95% CI were given in the column on the right.

CI: confidence interval; BALF: bronchoalveolar lavage fluid; COPD: chronic obstructive pulmonary disease; BMI: body mass index; FVC: forced vital capacity; FEV₁: forced expiratory volume in the first second; FEV₁%Pred: and the percentage of FEV₁ to predicted FEV₁; NA: not applicable.

CXCL5 and G-CSF levels increased in plasma and BALF of COPD patients

The plasma CXCL5 levels of COPD patients were significantly increased as compared to healthy control (Figure 1(A)). Both COPD nonsmokers and smokers showed increased plasma CXCL5 levels compared to healthy nonsmokers and smokers (Figure 1(B)). Both current and former COPD smokers showed increased plasma CXCL5 levels compared to current healthy smokers (Figure 1(C)). However, there is no difference of plasma CXCL5 between current and former smokers in either healthy controls or COPD patients. The BALF CXCL5 levels of COPD patients were significantly increased as compared to healthy control (Figure 1(D)). In addition, COPD smokers showed increased BALF CXCL5 levels compared to healthy nonsmokers (Figure 1(E)). However, no difference of BALF CXCL5 between current and former smokers in either healthy controls or COPD patients was observed (Figure 1(F)). On the other hand, the levels of G-CSF, a well-known cytokine that promotes the production and priming of neutrophils, were found to be increased in both plasma (Figure 1(G)) and BALF (Figure 1(H)) of COPD patients as compared to the healthy subjects.

Figure 1.

Increased CXCL5 and G-CSF levels in plasma and bronchoalveolar lavage fluid (BALF) of COPD patients compared to healthy controls. (A) CXCL5 levels in plasma of healthy control and COPD patients (n = 24 and 63, respectively); (B) CXCL5 levels in plasma of healthy nonsmokers, healthy smokers, COPD nonsmokers and COPD smokers (n = 16, 8, 14 and 49, respectively); (C) CXCL5 levels in plasma of healthy current smokers, healthy former smokers, COPD current smokers and COPD former smokers (n = 5, 3, 17 and 32, respectively); (D) CXCL5 levels in BALF of healthy control and COPD patients (n = 24 and 28, respectively); (E) CXCL5 levels in BALF of healthy nonsmokers, healthy smokers, COPD nonsmokers and COPD smokers (n = 16, 8, 13 and 15, respectively); (F) CXCL5 levels in BALF of healthy current smokers, healthy former smokers, COPD current smokers, and COPD former smokers (n = 4, 4, 5 and 10, respectively); (G) G-CSF levels in plasma of healthy control and COPD patients (n = 24 and 63, respectively); (H) G-CSF levels in BALF of healthy control and COPD patients (n = 24 and 28, respectively). Data was presented as mean with 95% confidence interval (CI), the differences with p value were tested by unpaired t-test or Mann–Whitney test, p < .05 was considered statistically significant, the difference between medians/means (with 95% CI of difference) were given following p value.

Circulated CXCL5 levels were correlated with lung function decline and G-CSF levels in COPD patients

The plasma CXCL5 levels in COPD patients were negatively correlated with their FEV₁/FVC, FEV₁%Pred, FVC, respectively (Figure 2(A–C)), and positively correlated with the plasma G-CSF levels and platelet count, (Figure 2(D–E)) after adjusting for age, gender, BMI, and smoking history (pack-years). The BALF CXCL5 levels in COPD patients were positively correlated with the BALF G-CSF levels (Figure 2(I)) after adjustment. However, the correlation between BALF CXCL5 levels and FEV₁/FVC, FEV₁%Pred, FVC, respectively, were not significant (Figure 2(F–H)).

Figure 2.

The correlation of CXCL5 levels to the lung function or G-CSF levels in COPD patients after adjusting for age, gender, body mass index (BMI) and smoking history. Plasma CXCL5 levels was negatively correlated with (A) the ratio of forced expiratory volume at 1 s to forced vital capacity (FEV₁/FVC, adjusted r = –0.263, p = .044), (B) the ratio of tested FEV₁ to predicted FEV₁ (FEV₁%Pred, adjusted r = –0.267, p = .041), (C) forced vital capacity (FVC, adjusted r = –0.307, p = .018), and positively correlated with (D) plasma G-CSF levels (adjusted r = 0.272, p = .039), and (E) platelet count (adjusted r = 0.331, p = .010); bronchoalveolar lavage fluid (BALF) CXCL5 levels was positively correlated with (I) BALF G-CSF levels (adjusted r = 0.595, p = .002), but not correlated with (F) FEV₁/FVC (adjusted r = –0.345, p = .099), (G) FEV₁%Pred (adjusted r = –0.244, p = .250) or (H) FVC (adjusted r = –0.322, p = .126). The correlations analysis with p value were performed by Pearson’s partial correlation test to correct for the age, gender, BMI and smoking history, and followed by multiple linear regression analysis, p < .05 was considered statistically significant.

A receiver operating characteristic analysis was performed to determine the best cut-off values of CXCL5 in plasma and BALF, respectively, for distinguish COPD patients (Figure 3). The corresponding area under curve (AUC) value with 95% CI for the plasma CXCL5 was 0.882 (95% CI: 0.803–0.961, Figure 3(A)), and for the BALF CXCL5 was 0.712 (95% CI: 0.568–0.856, Figure 3(B)). The optimal cut-off value for plasma CXCL5 was 153.34 pg/mL with sensitivity of 88.9% and specificity of 70.8% (Table 2), while the optimal cut-off value for BALF CXCL5 was 60.235 pg/mL with sensitivity of 90.3% and specificity of 45.8%. Overall, plasma CXCL5 correlated with the airflow limitation in COPD patients, and showed more potent as a biomarker for COPD with relatively higher sensitivity and specificity.

Figure 3.

Receiver operating characteristic analysis revealed that plasma CXCL5 levels was better to distinguish COPD than bronchoalveolar lavage fluid (BALF) CXCL5 levels. The corresponding area under curve (AUC) value with 95% confidence interval (CI) was (A) 0.882 for plasma CXCL5 (95% CI: 0.803–0.961), and (B) 0.712 for the BALF CXCL5 (95% CI: 0.568–0.856), respectively.

Table 2.

Coordinates of the receiver-operating characteristic curve for COPD patients and healthy controls.

COPD positive if ≥	Sensitivity	1-Specificity
17.080	1.000	1.000
21.060	1.000	0.958
31.415	1.000	0.917
40.960	1.000	0.833
45.210	1.000	0.792
48.300	1.000	0.750
52.245	1.000	0.708
58.905	1.000	0.667
69.600	1.000	0.583
81.495	1.000	0.542
91.145	1.000	0.500
101.930	1.000	0.458
110.585	0.984	0.417
115.735	0.968	0.417
125.720	0.952	0.417
133.435	0.937	0.417
138.225	0.937	0.375
144.360	0.921	0.375
146.710	0.921	0.333
148.595	0.905	0.333
149.885	0.889	0.333
153.340	0.889	0.292
157.810	0.873	0.292
159.105	0.841	0.292
164.715	0.841	0.250
171.275	0.825	0.250
175.120	0.810	0.250
178.405	0.794	0.250
180.035	0.778	0.250
183.280	0.762	0.250
185.780	0.746	0.250
192.145	0.746	0.208
198.345	0.730	0.208
201.470	0.730	0.167
205.455	0.683	0.167
208.545	0.667	0.167
214.675	0.651	0.167
220.765	0.635	0.167
223.780	0.619	0.167
224.940	0.587	0.167
226.935	0.587	0.125
232.725	0.571	0.125
240.585	0.556	0.125
244.785	0.540	0.125
245.755	0.540	0.083
247.400	0.524	0.083
251.930	0.492	0.083
260.075	0.492	0.042
273.475	0.492	0.000
285.960	0.476	0.000
294.165	0.460	0.000
302.295	0.444	0.000
308.125	0.429	0.000
310.805	0.413	0.000
315.670	0.397	0.000
322.725	0.365	0.000
340.000	0.333	0.000
359.285	0.317	0.000
369.780	0.302	0.000
380.180	0.286	0.000
397.750	0.270	0.000
412.380	0.254	0.000
416.390	0.238	0.000
430.660	0.222	0.000
449.540	0.206	0.000
458.100	0.190	0.000
463.505	0.175	0.000
468.125	0.159	0.000
508.810	0.143	0.000
549.095	0.127	0.000
551.635	0.111	0.000
555.255	0.095	0.000
577.410	0.079	0.000
611.420	0.063	0.000
626.110	0.048	0.000
649.115	0.032	0.000
691.895	0.016	0.000
714.060	0.000	0.000

The value in bold represents the optimum cut-off point.

COPD: chronic obstructive pulmonary disease.

Cigarette smoke exposure induced emphysema-like phenotype in C57Bl/6 mice

To assure that the cigarette smoking treatment of C57Bl/6 mice by using our custom-made apparatus can reproduce the pathologic features of COPD, we assessed their lung functions by using the forced pulmonary maneuver system. A significant increase of TLC, VC, IC, RV, Cchord, Cfvc50, and FRC (Figure 4(B–H), respectively) were observed in the cigarette smoking group as compared to control group. Nevertheless, we observed a significant decrease of body weight (bw) in the mice treated with cigarette smoke exposure (Figure 4(A)). Given that the TLC, VC, IC, and FRC relate to the size of body, RV and Cchord relate to the size of lung [35,36], we calculated the TLC, VC, IC, FRC of each mouse in relation to its own body weight (TCL/bw, VC/bw, IC/bw, FRC/bw), and the RV, Cchord to its own TLC (RV/TLC, Cchord/TLC). As shown in Figure 4(I–O), the normalized TLC/bw, VC/bw, IC/bw, RV/TLC, Cchord/TLC, FRC/bw, and FRC/TLC were increased in the cigarette smoking group as compared to the control group. The data above suggested that the cigarette smoke exposure induced an emphysema-like phenotype in mice.

Figure 4.

Lung function tests revealed that cigarette smoke exposure lead to emphysema-like phenotype in C57Bl/6 mice. As compared to control mice (n = 7), cigarette smoking treated mice (n = 7) showed significantly decreased (A) body weight (BW), and significantly increased (B) total lung capacity (TLC), (C) vital capacity (VC), (D) inspiratory capacity (IC), (E) residual volume (RV), (F) chord compliance (Cchord), (G) Cchord at 50% VC, (H) functional residual capacity (FRC) (I) TLC/BW, (J) VC/BW, (K) IC/BW, (L) RV/TLC, (M) Cchord/TCL, (N) FRC/BW, and (O) FRC/TLC. Data was presented as mean with 95% confidence interval (CI), the differences with p value were tested by unpaired t-test or Mann–Whitney test, p < .05 was considered statistically significant, the difference between medians/means (with 95% CI of difference) were given following p value.

Cigarette smoke exposure induced airflow limitation in C57Bl/6 mice

Figure 5 showed a significantly increase of FVC, FEV₂₀₀, PEF, and FEF75% (Figure 5(A,C,E,F), respectively) in the cigarette smoking group as compared to the control group. In particular, the FEV₁₀₀/FVC of cigarette smoking group significantly decreased as compared to control group (Figure 5(D)), indicating an airflow limitation. These data suggested that the cigarette smoke exposure induced an airflow limited phenotype in mice.

Figure 5.

Lung function tests revealed that cigarette smoke exposure lead to airflow-limited phenotype in C57Bl/6 mice. As compared to control mice (n = 7), cigarette smoking treated mice (n = 7) showed significantly increased (A) forced vital capacity (FVC), (B) forced expiratory volume at 100 ms (FEV₁₀₀), (C) forced expiratory volume at 200 ms (FEV₂₀₀), (D) FEV₁₀₀/FVC, (E) peak expiratory flow (PEF), (F) flow at 75% of FVC (FEV_75%). Data was presented as mean with 95% confidence interval (CI), the differences with p value were tested by unpaired t-test or Mann-Whitney test, p < .05 was considered statistically significant, the difference between medians/means (with 95% CI of difference) were given following p value.

Lung histology and BALF cell counts indicated profound airway inflammation, small airway fibrosis and mucus hypersecretion in cigarette smoking treated mice

Lung histology and BALF cell counts were performed to further assess the airway inflammation and airway remodeling. As compared to the control mice (Figure 6(A–C)), the cigarette smoking treated mice showed profound peribronchial and/or perivascular inflammatory infiltration, collagen deposition, and mucus secretion (Figure 6(D–F)), resulting in a significantly increased overall histological scores (Figure 6(G)). The mean linear intercept (MLI) analysis also suggested an enlargement of alveolar in cigarette smoking treated mice (Figure 6(H)). The profound airway inflammation was further confirmed by the BALF cell counts, the cigarette smoking group showed significantly increased number of total cell count, neutrophils, macrophages, and lymphocytes (Figure 6(I–L)) compared to the control group. These data, together with lung function data, suggested that the cigarette smoke exposure reproduced the main pathological features of COPD in the C57Bl/6 mice.

Figure 6.

Lung histology and bronchoalveolar lavage fluid (BALF) cell counts revealed that cigarette smoke exposure lead to airway inflammation and airway remolding in C57Bl/6 mice. Representative pictures of paraffin sections (200×, scale bar = 100 μm) stained with hematoxylin and eosin solution (A and D), alcian blue-periodic acid Schiff stain (B and E) and masson’s trichrome stain (C and F) showed that cigarette smoking (CS) treated mice had profound airway inflammation, collagen deposition, and mucus secretion (D–F) as compared to control (Co) mice (A–C), resulting in increased histological scores (G). As compared to control mice (n = 7), cigarette smoking treated mice (n = 7) showed increased (H) mean linear intercept, (I) total cells, (J) neutrophils, (K) macrophages, and (L) lymphocytes counts in BALF. Data in (G–L) was presented as mean with 95% confidence interval (CI), the differences with p value were tested by unpaired t-test or Mann-Whitney test, p < .05 was considered statistically significant, the difference between medians/means (with 95% CI of difference) were given following p value.

CXCL5 and G-CSF levels increased in both serum and BALF of cigarette smoking treated mice

By using the Magnetic Luminex Screening Assay, we found that the CXCL5 levels in both serum and BALF (Figure 7(A,B)) of cigarette smoking group significantly increased as compared to healthy control. On the other hand, the G-CSF levels were also found to be increased in both the serum and BALF (Figure 7(C,D)) of cigarette smoking group as compared to control group.

Figure 7.

Increased CXCL5 and G-CSF levels in serum and bronchoalveolar lavage fluid (BALF) of cigarette smoking treated mice compared to control mice. As compared to control mice (n = 7), cigarette smoking treated mice (n = 7) showed increased (A) serum CXCL5 levels, (B) BALF CXCL5 levels, (C) serum G-CSF levels, and (D) BALF G-CSF levels. Data was presented as mean with 95% confidence interval (CI), the differences with p value were tested by unpaired t-test or Mann-Whitney test, p < .05 was considered statistically significant, the difference between medians/means (with 95% CI of difference) were given following p value.

CXCL5 levels were correlated with lung function decline and G-CSF levels in cigarette smoking treated mice

The serum CXCL5 levels were positively correlated with the TLC and FRC (Figure 8(A,B)), and negatively correlated with the FEV₁₀₀/FVC (Figure 8(E)) after adjusting for the body weight. Similarly, the BALF CXCL5 levels were positively correlated with the TLC, FRC, and FVC (Figure 8(G–I)), and negatively correlated with the FEV₁₀₀/FVC (Figure 8(K)) after adjustment. In addition, the serum and BALF CXCL5 levels were significantly correlated with the serum and BALF G-CSF levels (Figure 8(F,L)), respectively. Overall, the CXCL5 levels in both serum and BALF correlated well with the lung function and G-CSF in cigarette smoking treated mice.

Figure 8.

The correlation of CXCL5 levels to the lung function or G-CSF levels in cigarette smoking treated and control mice after adjusting for body weight. Serum CXCL5 levels was positively correlated with (A) total lung capacity (TLC, adjusted r = 0.675, p = .016), (B) functional residual capacity (FRC, adjusted r = 0.73, p = .007), (F) serum G-CSF levels (adjusted r = 0.759, p = .003), and negatively correlated with (E) the ratio of forced expiratory volume at 100 ms to forced vital capacity (FEV₁₀₀/FVC, adjusted r = –0.631, p = .028), but not correlated with (C) forced vital capacity (FVC, adjusted r = 0.55, p = .064) and (D) forced expiratory volume at 200 ms (FEV₂₀₀, adjusted r = 0.408, p = .187); Bronchoalveolar lavage fluid (BALF) CXCL5 levels was positively correlated with (G) TLC (adjusted r = 0.799, p = .002), (H) FRC (adjusted r = 0.654, p = .021), (I) FVC (adjusted r = 0.616, p = .033), (L) BALF G-CSF levels (adjusted r = 0.772, p = .002), and negatively correlated with (K) FEV₁₀₀/FVC (adjusted r = –0.681, p = .015), but not correlated with (J) FEV200 (adjusted r = 0.564, p = .056). The correlations analysis with p value were performed by Pearson’s partial correlation test to correct for the body weight of mice, and followed by multiple linear regression analysis, p < .05 was considered statistically significant.

Discussion

COPD increasingly contributes to the disability and mortality around the world, bringing a great of social and economic burden [37]. Earlier diagnose and intervention of COPD benefits patients by slowing down the rate of lung function decline and enhancing the quality of late life [38]. However, the unawareness of COPD and nonstandard operation of spirometer in some developing regions might hamper the in-time diagnose of COPD. Here, we demonstrated that the circulating CXCL5, a strong chemokine to recruit neutrophils during inflammation, might sever as a potential biomarker to add additional modest value in the preliminary screening and diagnosis of COPD.

Inflammatory immune response underlies the pathological characters of COPD. Cigarette smoking elicits innate immune response through releasing of chemokines from epithelium and residual macrophages to recruit neutrophils and other inflammatory cells to the lungs. The persistent local inflammation contributes to the repeated injury and repair of airway and alveoli structure. On the other hand, COPD is characterized by low-grade chronic systemic inflammation reflected by the increased circulating white blood cells [39], platelet [40], and inflammatory markers [41]. These inflammatory changes, especially the neutrophilic inflammation driven by the innate immune response, increase with disease severity and persist on smoking cessation, thus might provide potential inflammatory biomarkers that help in assessing the disease risk and severity of COPD.

CXCL5 was thought to be produced by the alveolar type II epithelial cells to elicit neutrophil accumulation in response to cigarette smoking stimulation [22,24]. CXCL5 derived from platelets regulated chemokine scavenging by binding to duffy antigen receptor for chemokines of erythrocyte, and accounted for the neutrophil homeostasis in basal condition [23]. The deficiency of CXCL5 in mouse decreased the cigarette smoke induced-pulmonary inflammation [22], suggesting that the recruitment of neutrophil driven by the CXCL5 is crucial to the development of COPD. Here, we observed a significant increase of CXCL5 levels in both BALF and plasma of COPD patients (Figure 1), similar to previous reports [28,29] Importantly, we found that the plasma CXCL5 levels correlated with the FEV₁/FVC, FEV₁%Pred and FVC in COPD patients (Figure 2(A–C)), and might provide an additional reference to distinguish COPD patients with relatively high sensitivity and specificity as shown in the receiver operating characteristic analysis.

However, our results suggested that the BALF CXCL5 levels did not correlate with the lung function decline in COPD patients (Figure 2(F–H)). The divergence of these correlations could be partially due to the limited cases recruited in BALF testing group. In addition, the exposure of active and passive cigarette smoke (the main inducements to elicit releasing of CXCL5 in the lung) were varied among COPD patients. As shown in Supplementary Table 1, part of current smokers, former smokers and nonsmokers of COPD were continually exposed to secondhand smoke that might also disturb the BALF CXCL5 levels. The complicated cigarette smoke exposure could also explain in part why the BALF CXCL5 levels in either healthy and COPD smokers (Figure 1(E)) or current and former smokers (Figure 1(F)) showed no difference. In contrast, the circulating CXCL5 derived not only from lung residential cells but also from platelets in blood. Recent studies suggested that the values of platelets count, mean platelet volume, and platelet distribution width were correlated with the morbidity or pulmonary function in stable COPD or COPD exacerbation patients [40,42–45]. Our results also demonstrated that the plasma CXCL5 levels correlates with PLT in the blood of COPD patients (Figure 2(E)), indicating that platelet derived CXCL5 might contribute to the plasma CXCL5 levels in COPD. This hypothesis and the possible role of platelets derived CXCL5 in the pathogenesis of COPD needs to be further investigated.

To explore whether the CXCL5 levels also correlate with the lung function decline in the mouse model of COPD, C57Bl/6 mice were treated with cigarette smoke exposure for 12 weeks. We observed significant increases of TLC, VC, IC, RC, Chord, FRC, FVC, FEV₂₀₀, PEF, FEF_75%, total cells, neutrophils, macrophages, lymphocytes and histological, and significant decrease of FEV₁₀₀/FVC (Figures 4–6). These emphysematous, airflow limited and inflammatory phenotypes suggested that the mouse model reproduced the major pathological features of COPD, similar to previous reports [32,33,46]. On this basis, we further showed that the CXCL5 levels in serum of these mice significantly increased (Figure 7(A)), and correlated with their own TLC, FRC, FVC and FEV₁₀₀/FVC (Figure 8(A–C,E)), similar to what we observed in the COPD patients. These results further emphasized that the circulating CXCL5 levels correlated well with the lung function decline in COPD. Interestingly, the increased BALF CXCL5 levels (Figure 7(B)) also correlated with the lung function (Figure 8(G–I,K)) in the mouse model of COPD, differed from what we observed in COPD patients. This divergence might be resulted from that the mice were exposed to the cigarette smoke in a strict uniform condition, while the cigarette smoke exposure of COPD patients were varied.

G-CSF is a cytokine known to influence the survival, proliferation and differentiation of the neutrophil lineage under basal conditions of hematopoiesis [47], and has been described to be increased in the BALF of COPD patients and systemically manifest the destructive inflammation and complex comorbidities of COPD [48]. Similarly, we observed a significant increase of G-CSF both in the BALF and plasma of COPD patients (Figure 1(G,H)). Moreover, we showed that the G-CSF levels correlated well with the CXCL5 levels both in the plasma and BALF of COPD patients (Figure 2(D,I)). The increased pattern of G-CSF (Figure 7(C,D)) and its correlation with CXCL5 (Figure 8(F,L)) were well reproduced in the cigarette smoke induced mouse model of COPD, indicating that the recruitment of neutrophils by CXCL5 might be coordinated with the priming of neutrophils by G-CSF, and both CXCL5 and G-CSF are significant in the pathogenesis of neutrophilic inflammation in COPD.

The present study might be limited by the followings: first, we only recruited 121 subjects (including 30 healthy controls and 91 COPD patients) due to the strict inclusion and exclusion criteria, the limited number of cases (especially in the BALF testing group) might result in false-negative correlations. Second, the age of healthy subjects were younger than COPD patients, and the gender biased towards male in smokers in respective of healthy controls or COPD patients. However, we adjusted age and gender in the correlation analysis and there is no study showing any relationship between CXCL5 expression and age, or CXCL5 expression and gender. Third, only stable COPD patients were recruited in this study, thus a possible dynamic change of CXCL5 levels during the progression of acute exacerbation of COPD needs to be further investigated. Finally, the circulating CXCL5 is expected to be increased in other diseases of neutrophilic inflammation such as pneumonia, the specificity of circulating CXCL5 as a biomarker of COPD needs to be further clarified. Only when combined with the clinical symptoms and medical history, the circulating CXCL5 level might have additional modest predictive value on top of clinical variables.

In conclusion, our results indicated that the CXCL5 might coordinate with the G-CSF in the pathogenesis of neutrophilic inflammation in COPD; the circulating CXCL5 might serve as a potential blood-based biomarker to assist in the preliminary screening of COPD and offer additional modest reference for the diagnosis of COPD.

Funding Statement

This work was supported by the 1·3·5 project for disciplines of excellence, West China Hospital, Sichuan University under Grant numbers ZYGD18006, ZYJC18012; National Natural Science Foundation of China (NSFC) under Grant numbers 81670038, 81830001, 81470236, 81800015; National Key Research and Development Program in China under Grant number 2016YFC0903600; China Postdoctoral Science Foundation under Grant numbers 2018M643501, 2019T120851; Fundamental Research Funds for the Central Universities (2018SCU12028, the Postdoctoral Foundation of Sichuan University); and Post-Doctor Research Project, West China Hospital, Sichuan University under Grant number 2018HXBH056.

Disclosure statement

All authors have read the policy on disclosure of potential conflicts of interest outlined by the Annals of Medicine. The authors report no conflict of interest.