HMGB1 is upregulated in the airways in asthma and potentiates airway smooth muscle contraction via TLR4.

To the Editor:

Asthma is characterized by variable airflow obstruction, airway hyperresponsiveness, and inflammation. Airway smooth muscle (ASM) contributes to asthma pathophysiology via hypercontractility, increased mass, and inflammatory mediator release. Clinical studies and animal models demonstrate a role for high-mobility group box 1 (HMGB1) and its receptors in airway inflammation and asthma.2, 3 HMGB1's activity and receptor interactions is determined by its redox state, with oxidation rendering HMGB1 inactive. We have investigated the redox state of airway HMGB1 and the role of HMGB1 in ASM function.

HMGB1 expression and/or redox state was investigated in sputum by ELISA, nonreducing electrophoresis, and Western blotting; in bronchial tissue by immunohistochemistry; and in ASM cells (ASMCs) by quantitative PCR, immunofluorescence, and flow cytometry. The effect of HMGB1 on ASMC reactive oxygen species (ROS) production, migration, proliferation, apoptosis, and contraction was evaluated. Leicestershire Research Ethics Committee approved the study, with informed consent obtained from all subjects. Statistical analysis was performed using GraphPad Prism 6.0. For detailed Methods, see this article's Online Repository at www.jacionline.org.

Sputum HMGB1 concentration was increased in those with severe asthma but not in those with mild to moderate asthma versus healthy controls (controls) (Fig 1, A; see Table E1 in this article's Online Repository at www.jacionline.org), correlating significantly with total cell counts and nonviable cell counts/g sputum (see Fig E1, A and B, in this article's Online Repository at www.jacionline.org), but not sputum differential cell counts (ie, % eosinophils, neutrophils, macrophages, lymphocytes, or epithelial cells) nor lung function (data not shown). HMGB1 concentration in sputum from those with severe asthma was unaffected by oral corticosteroid (OCS) treatment, and did not correlate with OCS dose (data not shown). Both disulphide and reduced HMGB1 were significantly increased in sputum from those with severe asthma versus controls (Fig 1, B). In sputum with detectable HMGB1, the proportion of reduced versus disulphide HMGB1 was increased in those with severe asthma versus controls (Fig 1, C). Sputum endogenous secretory receptor for advanced glycosylation end products (endogenous secretory RAGE), measured in a subset of sputum samples, was not different between groups (Fig E1, C; see Table E2 in this article's Online Repository at www.jacionline.org). HMGB1 expression was significantly increased in ASM in bronchial biopsies from those with severe asthma versus controls (Fig 1, D; see Table E3 in this article's Online Repository at www.jacionline.org), with no effect of OCS observed. No differences in ASM RAGE or epithelial HMGB1/RAGE expression in bronchial biopsies were observed (Fig E1, D-F; Table E3). Representative photomicrographs of HMGB1/RAGE staining are shown in Fig E1, G-L, in this article's Online Repository.

Fig 1

A, Sputum HMGB1 concentrations in healthy controls, patients with mild to moderate asthma, and patients with severe asthma (horizontal bar geometric mean and 95% CI). B, Disulphide and reduced redox forms of HMGB1 expressed as mean ± SEM % of standard rHMGB1 (STD), sputum from healthy controls (n = 6), patients with mild to moderate asthma (n = 4), and patients with severe asthma (n = 5) and with a representative Western blot above. C, Relative expression of reduced versus disulphide HMGB1 in sputum with detectable levels of HMGB1 from healthy controls (n = 8), patients with mild to moderate asthma (n = 12), and patients with severe asthma (n = 11), with a representative Western blot above. D, HMGB1+ cells/mm2 ASM in bronchial biopsies. Symbol key: ● = healthy control; ■ = GINA 1; ◆ = GINA 2;  = GINA 3; ▲ = GINA 4; ★ = GINA 5. GINA, Global Initiative for Asthma.

Table E1

Subjects' characteristics for the HMGB1 ELISA

Characteristic	Healthy control(n = 18)	Asthma mild-moderate(n = 20)	Asthma severe(n = 22)	ANOVA/χ2 test∗
GINA classification, n		GINA 1, n = 5GINA 2, n = 3GINA 3, n = 12	GINA 4, n = 13GINA 5, n = 9
Age (y)†	53.0 ± 3.2	57.6 ± 3.2	57.3 ± 2.6	P = .51
Sex: male, n (%)	7 (39)	12 (60)	11 (50)	P = .43∗
Atopy, n (%)	4 (22.2)	12 (60)	17 (77)	P = .002∗
Smoking status, current smoker, n (%)	0 (0)	1 (5)	3 (14)	P = .21∗
ICS (μg/d BDP equivalent)‡	0 (0)	200 (400)§	1600 (400)§||	P < .001
Pre-BD FEV1 % predicted‡	91.5 (35.6)	87.0 (25.0)	72.5 (41.5)§	P = .03
Pre-BD FEV1/FVC‡	74.5 (6.0)	73.0 (10.5)	72.0 (15.0)	P = .21
% Sputum neutrophils†	59.8 ± 5.3	58.1 ± 6.6	62.7 ± 4.5	P = .88
% Sputum eosinophils‡	0.75 (1.4)	1.0 (3.3)	2.8 (13.9)§||	P = .002
% Sputum epithelial cells‡	1.60 (2.3)	2.25 (4.8)	1.4 (1.7)	P = .735

BDP, Beclometasone dipropionate; FVC, forced vital capacity; ICS, inhaled corticosteroid; pre-BD, prebronchodilator.

χ2 test.

Mean ± SEM.

Median (interquartile range).

P < .05 vs healthy control.

P < .05 vs mild asthma.

Fig E1

Sputum HMGB1 correlated with sputum total cell count (A) and sputum nonviable cell count (B) in patients with asthma. C, Sputum esRAGE concentration in subjects with asthma and healthy controls. Horizontal bar represents geometric mean (95% CI). P value from Kruskall-Wallis test. Subject characteristics are presented in Table E2. D, The number of HMGB1+ cells/mm2 epithelium in bronchial biopsies. E and F, RAGE expression was assessed using a semi-quantitative scoring system: 0 = no positive staining; 1 = little positive staining; 2 = moderate positive staining; 3 = marked positive staining by a blinded observer in (Fig E1, E) ASM and (Fig E1, F) epithelium in bronchial biopsies. G-L, HMGB1/RAGE staining in bronchial tissue: G, Isotype control (rabbit immunoglobulin fraction, ×200 magnification), HMGB1 staining (brown) in a healthy control (H, ×200 magnification; I, ×400 magnification) and in a subject with severe asthma (J, ×200 magnification; K, ×400 magnification); positive staining was observed in the smooth muscle (arrow) and the epithelium (arrow head). L, RAGE staining in a tissue section from a patient with severe asthma showing RAGE expression in smooth muscle (arrow) and epithelium (arrowhead) at ×200 magnification. Inset: mouse IgG2a isotype control (×200 magnification). Subject characteristics are presented in Table E3. Symbol key: ● = healthy control; ■ = GINA 1; ◆ = GINA 2;  = GINA 3; ▲ = GINA 4; ★ = GINA 5. esRAGE, Endogenous secretory RAGE.

Table E2

Subjects' characteristics for esRAGE ELISA

Characteristic	Healthy control(n = 15)	Asthma mild-moderate(n = 16)	Asthma severe(n = 19)	ANOVA/χ2 test∗
GINA classification, n		GINA 1, n = 4GINA 2, n = 2GINA 3, n = 10	GINA 4, n = 12GINA 5, n = 7
Age (y)†	53.1 ± 3.9	58.2 ± 3.7	59.6 ± 2.5	P = .37
Sex: male, n (%)	7 (46.7)	8 (50)	10 (52.6)	P = .94∗
Atopy, n (%)	4 (26.7)	9 (56.25)	15 (78.9)	P = .01∗
Smoking status, current smoker, n (%)	0 (0)	1 (6.25)	3 (15.8)	P = .23∗
ICS (μg/d BDP equivalent)‡	0 (0)	200 (400)	2000 (400)§||	P < .001
Pre-BD FEV1 % predicted‡	94.0 (50.5)	86.5 (32.3)	75.0 (16.0)§	P = .030
Pre-BD FEV1/FVC‡	75.0 (6.0)	73.5 (13.8)	72.0 (14.0)	P = .09
% Sputum neutrophils†	62.3 ± 5.8	59.5 ± 7.0	59.4 ± 4.6	P = .93
% Sputum eosinophils‡	0.75 (1.96)	0.75 (1.75)	4.25 (13.3)§||	P = .001
% Sputum epithelial cells‡	1.2 (2.1)	1.5 (4.8)	1.5 (3.0)	P = .90

Samples used are subsets of those used for the HMGB1 ELISA.

BDP, Beclometasone dipropionate; esRAGE, endogenous secretory RAGE; ICS, inhaled corticosteroid; FVC, forced vital capacity; pre-BD, prebronchodilator.

χ2 test.

Mean ± SEM.

Median (interquartile range).

P < .05 vs healthy control.

P < .05 vs mild asthma.

Table E3

Subjects' characteristics for immunohistochemistry

Characteristic	Healthy control(n = 10)	Asthma severe(n = 19)	t test/χ2 test∗
GINA classification, n		GINA 4, n = 11GINA 5, n = 5
Age (y)†	46.7 ± 6.2	55 ± 2.8	P = .17
Sex: male, n (%)	5 (50)	8 (42.1)	P = .71∗
Smoking status, current smoker, n (%)	1 (10)	5 (26.3)	P = .63∗
ICS (μg/d BDP equivalent)‡	0 (0)	1600 (125)§	P < .001
Pre-BD FEV1 % predicted‡	93.5 (28.6)	76.0 (54.7)	P = .065
Pre-BD FEV1/FVC‡	90.5 (13.3)	71.1 (24.9)§	P = .0009
% Sputum neutrophils†	50.5 ± 18.7	52.6 ± 6.1	P = .9
% Sputum eosinophils‡	0.4 (0.8)	5.3 (16.8)§	P = .002

BDP, Beclometasone dipropionate; ICS, inhaled corticosteroids; FVC, forced vital capacity; pre-BD, pre-bronchodilator.

χ2 test.

Mean ± SEM.

Median (interquartile range).

P < .05 vs healthy control.

HMGB1 expression was investigated in primary ASMCs. Although there was no differential expression of HMGB1 mRNA (Fig 2, A), HMGB1 protein expression was significantly reduced in ASMCs from those with asthma versus controls, assessed by flow cytometry (Fig 2, B) and immunofluorescence (see Fig E2, A and B, in this article's Online Repository at www.jacionline.org). Release of HMGB1 extracellularly by ASMCs from those with asthma and/or the absence in culture of proinflammatory mediators present in asthmatic airways could explain the anomalous in vitro and in vivo data. Indeed HMGB1 protein expression was significantly upregulated in ASMCs from those with asthma but not controls following stimulation with TNF-α, IL-1β and IFN-γ, or poly(I:C) (Fig 2, C); however, HMGB1 expression poststimulation was not different between asthma and health. HMGB1 in ASMC supernatants was below the limit of detection.

Fig 2

HMGB1 mRNA and protein expression in ASMCs assessed by (A) quantitative PCR and flow cytometry (B) at baseline with representative histograms to the right and (C) following stimulation with proinflammatory cytokines (TNF-α, IFN-γ, and IL-1β, 10 ng/mL) or the dsRNA mimic poly(I:C) (12.5 μg/mL). *P < .05 versus unstimulated ASMCs. ROS production in response to (D) HMGB1 (10-1000 ng/mL) in ASMCs, *P ≤ .05 healthy control (●) vs patient with asthma (■), unpaired t test, (n = 5-9 ASM donors), and (E) HMGB1 (1000 ng/mL) ± LPS-RS or sRAGE (10 μg/mL) in ASMCs from healthy controls. P values from unpaired t tests. F-H, Contraction of collagen gels impregnated with ASMCs from subjects with asthma in the presence of bradykinin following incubation with vehicle control or 1000 ng/mL HMGB1 ± LPS-RS or sRAGE (10 μg/mL) expressed as (Fig 2, F) time course, *P < .05 vs vehicle control, (Fig 2, G) example collagen gels at 60 minutes, and (Fig 2, H) AUC at 60 minutes. *P < .05 vs vehicle control, paired t tests. Means ± SEM are shown. GMFI, Geometric mean fluorescence intensity; PE, phycoerythrin.

Fig E2

A, HMGB1 protein expression in ASM cells was assessed by immunofluorescence. B, Representative immunofluorescent staining of cell nuclei (DAPI, upper panel), HMGB1 (middle panel), and merged DAPI and HMGB1 (lower panel). C, ASM from subjects with asthma and control subjects was shown to express different isoforms of RAGE by PCR. PCR products were isolated by gel purification, sequenced, and identified as full-length membrane-bound RAGE (RAGE), or soluble isoforms of RAGE (sRAGE 1-3). M = molecular weight marker, C = healthy control, A = asthma. D, Representative flow cytometry histograms showing cell-surface RAGE ( FITC fluorescence, upper panel) and TLR4 (Alexa Fluor 488 fluorescence, lower panel) expression in ASM cells. Open histograms represent isotype controls; dark gray and light gray shaded histograms are RAGE or TLR4-stained cells, respectively. E, ASM cell-surface receptor expression was expressed as gMFI (fold change antibody/isotype control) for healthy control subjects (●) and subjects with asthma () for RAGE or TLR4 (P = .84 and P = .99, healthy control vs patient with asthma, respectively, unpaired t test). DAPI, 4′,6′-Diamidino-2 phenylindole; FITC, fluorescein isothiocyanate.

Reduced recombinant HMGB1, at concentrations equivalent to those in sputum, caused a concentration-dependent increase in intracellular ROS production in ASMCs from controls, but not in ASMCs from those with asthma (Fig 2, D), which was reduced by the RAGE decoy receptor soluble RAGE (sRAGE) and the Toll-like receptor (TLR) 4 antagonist LPS from Rhodobacter sphaeroides (LPS-RS) (Fig 2, E). This defective response in ASMCs from those with asthma was not due to impaired ROS generation capacity; ROS production was similar at baseline, and was increased in response to hydrogen peroxide stimulation of ASMCs from those with asthma versus controls. In addition, ASMC RAGE and TLR4 expression was not different between those with asthma and controls (Fig E2, C-E). HMGB1 activity and receptor binding is dependent on its oxidation state, and is regulated by several binding partners. The complex interplay between these factors affects the cellular response to HMGB1 and might affect ROS production in ASM from those with asthma.

In the absence of bradykinin, contraction of collagen gels impregnated with ASMCs from those with asthma was not significantly different from controls (see Fig E3, A, in this article's Online Repository at www.jacionline.org), nor was it affected by HMGB1 (100, 300 [data not shown], and 1000 ng/mL [Fig E3, B and C]). However bradykinin-mediated contraction of collagen gels impregnated with ASMCs was potentiated by 1000 ng/mL HMGB1, resulting in a decrease in the area under the curve at 60 minutes (Fig E3, D and E), but not 100 to 300 ng/mL HMGB1 (data not shown). In the presence of HMGB1, bradykinin-mediated contraction of collagen gels impregnated with ASMCs from those with asthma was potentiated to a greater extent compared with controls (Fig E3, F) and was significantly inhibited by LPS-RS, but not by sRAGE (Fig 2, F-H). sRAGE and LPS-RS had no significant effect in the absence of HMGB1.

Fig E3

A, Time course of contraction of collagen gels impregnated with ASM cells from healthy controls and subjects with asthma in the absence of bradykinin, with the area under the curve (AUC) over a 24-hour period as an inset. Time course (B) and AUC over a 60-minute period (C) of contraction of collagen gels impregnated with ASM from subjects with asthma in the absence of bradykinin, following incubation with HMGB1 (1000 ng/mL, n = 5). D, Time course of contraction of collagen gels impregnated with ASM cells in the presence of bradykinin following incubation with HMGB1 (1000 ng/mL), n = 12 ASM donors, *P < .05 HMGB1 versus vehicle control, paired t test at each time point. E, AUC over a 60-minute period of contraction of collagen gels impregnated with ASM cells in the presence of bradykinin following incubation with HMGB1 (1000 ng/mL), healthy control (●), asthma (), P < .05, paired t test. F, Reduction in AUC at 60 minutes (% vehicle control) in gels impregnated with ASM derived from both healthy controls and subjects with asthma, *P < .05 versus vehicle control. G, Compared with the premigration control significant ASM cell migration, measured by counting Hoechst-positive cell nuclei in the migration zone, was seen (*P < .05 vs premigration control) but was not significantly different following treatment for 24 hours with exogenous 1000 ng/mL HMGB1, or by blocking the activity of endogenously released HMGB1 with 10 μg/mL sRAGE or 10 μg/mL LPS-RS. H, Cell proliferation as measured by the CellTiter AQueous One Solution Cell Proliferation Assay was unaffected following treatment with 1000 ng/mL HMGB1 for 72 hours in the presence or absence of serum. The percentage of cells undergoing apoptosis as detected by Annexin V staining (I) and necrosis as detected by propidium iodide (J) was not significantly different following treatment with 1000 ng/mL HMGB1. Staurosporine (STS) was used as a positive control.

HMGB1 and/or sRAGE or LPS-RS (10 μg/mL) had no effect on ASMC migration (wound healing assay: HMGB1 [100-1000 ng/mL] ± CXCL12 [10-100 ng/mL], data not shown, ORIS assay: 3-1000 ng/mL HMGB1, Fig E3, G), proliferation in the presence/absence of serum (Fig E3, H), and apoptosis or necrosis (Fig E3, I and J).

Our data support and extend previous studies suggesting an imbalance between HMGB1 and endogenous secretory RAGE in the asthmatic airways might have implications for HMGB1 in asthma pathophysiology. Because the increased HMGB1 we see in the sputum in asthma correlates with sputum total cell and nonviable cell counts, we propose that HMGB1 can be upregulated in the airways in asthma because of inflammatory and stress stimuli that can result in HMGB1 secretion actively by activated immune cells and passively by necrotic cells.7, 8

We propose that the ROS produced in response to HMGB1 in ASMCs from controls, in a RAGE/TLR4-dependent manner, terminally oxidize HMGB1, rendering it inactive or alter Ca2+ homeostasis, leading to reduced contractility via a TLR4/ROS-dependent mechanism as in murine cardiomyocytes, thus limiting the potentiation of contraction of collagen gels impregnated with ASMCs from controls. Because of the ROS-generating capacity of ASMCs in response to HMGB1 being defective in asthma, these ROS-mediated responses would be reduced. Therefore, HMGB1 can potentiate contraction of collagen gels impregnated with ASMCs from those with asthma, in a TLR4-dependent manner, to a greater extent than those impregnated with ASMCs from controls. Thus, HMGB1 could contribute to ASM dysfunction and airway hyperresponsiveness in asthma, as supported by animal models,3, 10 possibly representing a potential therapeutic target.