Brain-derived neurotrophic factor (BDNF) contributes to neuronal dysfunction in a model of allergic airway inflammation

Abstract

• Brain-derived neurotrophic factor (BDNF) is a candidate molecule for mediating functional neuronal changes in allergic bronchial asthma. Recently, enhanced production of BDNF during allergic airway inflammation caused by infiltrating T-cells and macrophages as well as by resident airway epithelial cells has been described. It was the aim of this study to investigate the effect of enhanced BDNF levels on lung function and airway inflammation in a mouse model of allergic inflammation.

• Ovalbumin-sensitised BALB/c mice were challenged in two consecutive allergen challenges. Prior to the challenge, the mice were treated with either anti-BDNF antibodies or isotype-matched control antibodies. Airway responsiveness to methacholine, capsaicin and electric field stimulation, as well as airway inflammation and chronic airway obstruction 1 week after the last allergen challenge were assessed.

• Anti-BDNF blocked enhanced reactivity in response to capsaicin, but not airway smooth muscle hyper-reactivity in vivo. Furthermore, persistent airway obstruction, as observed 1 week after the last allergen challenge, was to a large extent prevented by anti-BDNF treatment. In vitro, BDNF and anti-BDNF treatment had a profound effect on local neuronal hyper-reactivity, as shown by electric field stimulation experiments. In contrast, neither BDNF nor anti-BDNF treatment affected airway inflammation.

• Our data indicate that development of allergen-induced neuronal hyper-reactivity in mice is partially mediated by BDNF.

British Journal of Pharmacology (2004) 141, 431–440. doi:10.1038/sj.bjp.0705638

Abbreviations:

BAL(F)

bronchoalveolar lavage (fluid)

BDNF

brain-derived neurotrophic factor

BSA

bovine serum albumin

EFS

electrical field stimulation

EF₅₀

mid-expiratory airflow

HBP

head-out body plethysmography

NGF

nerve growth factor

OVA

ovalbumin

TB

time of braking

Introduction

Allergic bronchial asthma is a chronic inflammatory disease characterised by a variable degree of airway remodelling, airway hyper-responsiveness and recurrent attacks of wheezing and airway obstruction (for review, see Kay, 1996). Development of airway hyper-responsiveness to unspecific pharmacological stimuli is a pathogenic hallmark of asthma (Wills-Karp, 1999). Though the immunological network in asthma has been well characterised, the relationship between airway inflammation and changes in lung function is not completely understood (^{Wilder et al., 1999}; Wills-Karp, 1999). Airway tone is predominantly regulated by vagal cholinergic and sensory nerves (Barnes, 1996; Kesler & Canning, 1999). Over the last decade, it has been demonstrated that allergic asthma is associated with altered neuronal control of the airways, including sensory hyper-reactivity (Ellis & Undem, 1992; ^{Nieber et al., 1992}; ^{Fischer et al., 1996}; ^{Kaltreider et al., 1997}; ^{Undem et al., 1999}). However, concepts describing the link between allergic airway inflammation and neuronal dysfunction are still lacking. Several studies demonstrated that neurotrophins such as nerve growth factor (NGF) and brain-derived neurotrophic factor (BDNF) can mediate neuronal plasticity in inflammatory conditions (Lindsay & Harmar, 1989; ^{Donnerer et al., 1992}; ^{Dmitrieva et al., 1997}). There is recent evidence for a specific role of NGF in allergic asthma (^{Braun et al., 1998}; ^{Päth et al., 2002}). However, experiments on the functional role of elevated BDNF levels in asthma have not been described yet.

To evaluate the influence of BDNF on lung function in vivo, body plethysmography was performed as described before (^{Päth et al., 2002}; ^{Kerzel et al., 2003}). This system analyses the breathing pattern of mice by continuously registering changes in airflow during ins- and expiration. By using specific stimuli, it can distinguish between bronchoconstriction and sensory irritation. For determination of the sensory function, we used the reflex prolongation of stage I of expiration (‘time of braking (TB)’) that is caused by stimulating sensory airway nerves. Electrical or chemical stimulations of laryngeal, vagus or carotid sinus nerves were shown to prolong the period of depolarisation in postinspiratory neurons without changing the duration of expiratory or inspiratory stages, indicating a fairly selective prolongation of the first stage of expiration (^{Remmers et al., 1986}). This characteristic modification of the normal breathing pattern has been demonstrated as a sensitive and specific parameter for sensory nerve function in the lung. This system was validated and the results documented in a series of papers (^{Vijayaraghavan et al., 1993}; 1994; Alarie, 1998). It was adapted for asthma research recently (^{Braun et al., 1999}; ^{Neuhaus-Steinmetz et al., 2000}; ^{Glaab et al., 2001}). In this study, capsaicin was used as a stimulus for sensory neurons, since it is known to act specifically on sensory neurons via a vanilloid receptor (VR-1) (Michael & Priestley, 1999; Szallasi & Blumberg, 1999). In contrast, methacholine acts via direct stimulation of airways smooth muscle cells, resulting in a characteristic dose-dependent decrease in midexpiratory flow rate (EF₅₀). EF₅₀ represents an established parameter of airway obstruction (^{Neuhaus-Steinmetz et al., 2000}; ^{Glaab et al., 2001}).

Experimental data from animal models suggest that the effect of neurotrophins may be responsible for neuronal hyper-reactivity following allergic inflammation of the airways (^{Braun et al., 1998}; ^{Hoyle et al., 1998}; ^{de Vries et al., 1999}; ^{Undem et al., 1999}; ^{Päth et al., 2002}; ^{Kerzel et al., 2003}). Indeed, allergic asthmatic patients respond with increased levels of the neurotrophins NGF and BDNF in the bronchoalveolar lavage fluid (BALF) following allergen challenge (^{Virchow et al., 1998}). Utilising a well-characterised mouse model of asthma, we have identified T cells and macrophages as additional sources of BDNF in allergic airway inflammation (^{Braun et al., 1999}). Based on these observations, we have developed the hypothesis that BDNF may provide a link between airway inflammation and neuronal dysfunction in allergic asthma. It was the aim of this study to test this hypothesis in our mouse model of allergic airway inflammation.

Methods

Animals

Female BALB/c mice, 6–8 weeks old, were obtained from Harlan Winkelmann, Borchen, Germany and maintained under controlled conditions. Figure 1 and Table 1 show all experiments including the treatment scheme and total animal number.

Figure 1

Animal treatment protocol. Mice were sensitised to OVA adsorbed to Al(OH)₃ or vehicle alone by intraperitoneal injections on days 1, 14 and 21. Prior to analysis, animals received two consecutive local aerosol challenges of 1% OVA (w v⁻¹) diluted in PBS or PBS alone and delivered by 20-min aerosolisation on days 26 and 27. Intranasal application of polyclonal chicken IgY (isotype antibody) or anti-mouse BDNF was performed 3 h before each airway allergen challenge. In addition, animals received the antibodies i.p. on day 25. The response to acute allergen exposure was measured on day 35 in the body plethysmograph. All animals were analysed 24 h after the last challenge. Abbreviations: intraperitoneally (i.p.), intranasally (i.n.), aerosol challenge (aerosol).

Table 1. Overview of experiments and animal numbers

	Allergen		Treatment		Analysis
Figure	i.p.	Aerosol	i.p.	i.n.	Method	Animals
—	—	—	—	—	EFS	8
—	—	—	—	—	EFS	8
—	OVA	OVA	—	—	EFS	10
3 A	—	—	—	BDNF	EFS	12
3 A	—	—	—	BSA	EFS	12
3 B	OVA	OVA	Anti-BDNF	Anti-BDNF	EFS	12
3 B	OVA	OVA	IgY	IgY	EFS	12
5	—	—	—	—	HBP	8
5	OVA	OVA	Anti-BDNF	Anti-BDNF	HBP	10
5	OVA	OVA	IgY	IgY	HBP	10
6+7	—	OVA	—	—	HBP/BAL	11
6+7	OVA	OVA	Anti-BDNF	Anti-BDNF	HBP/BAL	11
6+7	OVA	OVA	IgY	IgY	HBP/BAL	11
7	—	—	—	BDNF	BAL	12
7	—	—	—	BSA	BAL	12

• Figures showing the results of these experiments are detailed in the first column. Sensitisation (intraperitoneally, i.p.) to allergen (ovalbumin, OVA) and allergen challenge (aerosol) are detailed in the following two columns. Intraperitoneal (i.p.) or intranasal (i.n.) treatment with anti-BDNF, isotype-matched control antibodies (IgY), BDNF or bovine serum albumin (BSA) is specified in the two columns ‘treatment’. The last two columns ‘analysis’ detail the methods (electric field stimulation, EFS; head-out body plethysmography, HBP; and bronchoalveolar lavage, BAL) and the number of animals used for analysis in each experiment. Ganglia for neuropeptide experiments were taken from animals of the last two experiments.

Protocol of allergic sensitisation

Mice were sensitised to ovalbumin (OVA; 10 μg per injection) (Sigma, Deisenhofen, Germany) adsorbed to 1.5 mg Al(OH)₃ (Pierce, Rockford, U.S.A.) or vehicle alone by intraperitoneal injections, as detailed in Figure 1. Aerosol challenges were performed with 1% OVA (w v⁻¹) diluted in PBS or PBS alone delivered by a Pari-Jet nebuliser aerosolisation (median mass diameter MMD=3 μm). All animals were analysed 24 h after the last challenge.

Intranasal BDNF or anti-BDNF treatment

Intranasal application of antibodies was performed as described before (^{Braun et al., 1998}), and lung distribution of nasally administered substances was controlled by evans blue instillation (data not shown). Polyclonal chicken IgY anti-mouse BDNF (2 × 25 μl) (500 μg ml⁻¹) or isotype antibody (500 μg ml⁻¹) (Promega, Madison, U.S.A.) solved in sterile PBS was instilled intranasally 3 h before each airway allergen challenge. In addition, 1 × 100 μl polyclonal chicken IgY anti-mouse BDNF (500 μg ml⁻¹) or isotype antibody (500 μg ml⁻¹) solved in sterile PBS was given intraperitoneally (i.p.) 24 h before first allergen challenge (Figure 1). Recombinant BDNF (5 μg ml⁻¹) (Biosource) or BSA (0.1%) was administered intranasally. Before application, mice were anaesthetised by subcutaneous injection of 2.6 mg ketamin hydrochloride (Ketanest©, Parke Davis, Berlin, Germany) and 0.18 mg xylazin hydrochloride (Rompun©, Bayer, Leverkusen, Germany) for approx. 10 min.

Bronchoalveolar lavage (BAL)

Animals were killed by cervical dislocation and their tracheae cannulated. Airways were lavaged twice with 0.8 ml ice-cold PBS and cell numbers, and cytokine contents were determined as described previously (^{Braun et al., 1999}). BAL recovery was 1.4±0.2 ml in all groups. Cell-free BALF was frozen at −20°C until analysis.

Assessment of airway responsiveness by electrical field stimulation (EFS)

Airway smooth muscle responsiveness was assessed by EFS as described before (^{Braun et al., 1998}; ^{Herz et al., 1998}). Mice were killed by cervical dislocation 24 h after the last allergen challenge. Tracheal smooth muscle segments (∼0.5 cm) were removed and hung between stainless steel wire triangular supports. There was a standard pretreatment and treatment regime in all EFS experiments. During the 90 min equilibration in the bath, tracheal segments were stimulated with KCl as a viability control. KCl stimulation was performed at three time points, with a final concentration in the organ bath of 120 mM KCl. Afterwards, tissues were rinsed with fresh buffer and allowed to relax to their initial tension. Then, the maximum contractile response was assessed using the following parameters: 16 V, 200 mA, 2 ms pulse duration, 1 ms delay, 30 Hz. The degree of maximal contraction in response to EFS and to KCl stimulation was comparable for each animal. Finally, the stimulus–response curves were generated by stimulating the tracheal segments with 0.5–30 Hz (12 V, 200 mA), with 2 min recovery time between each stimulation step (^{Braun et al., 1998}; ^{Herz et al., 1998}). In each stimulation step, tracheal segments were stimulated until reaching the peak of the contractile response (for approximately 20–25 s). The contraction in response to EFS stimulus was measured via an isometric force transducer (Grass Instruments, Quincy, U.S.A.). The frequency that caused 50% of the maximal contraction was calculated from logarithmic plots of the contractile response versus the frequency of EFS, and expressed as the ES₅₀.

Assessment of lung function by head-out bodyplethysmography

Meanwhile, airway function was assessed by head-out body plethysmography (HBP), as previously described (^{Vijayaraghavan et al., 1993}; 1994; ^{Braun et al., 1999}). Four mice were placed in four body plethysmographs attached to an exposure chamber (Crown Glass, Somerville, N.J., U.S.A.). For acclimatisation, animals were put into the exposure chamber 15 min prior to exposure. The slight decrease in respiratory flow during the baseline period is due to acclimatisation of mice in the body plethysmograph (Figure 6). There was no further decrease in respiratory flow after 10 min of acclimatisation (data not shown). Airflow was measured with a PTM 378/1.2 pneumotachograph (Hugo Sachs Electronics, March-Hugstetten, FRG) and a 8-T2 differential pressure transducer (Gaeltec, Dunvegan, U.K.). For determination of bronchoconstriction, mid-expiratory airflow (EF₅₀, i.e. the expiratory airflow when 50% of the tidal volume is exhaled) was measured in response to various concentrations of methacholine (4–75 mg ml⁻¹, 1 min) delivered by a jet nebuliser (Pari-Boy®; Pari-Werke, Starnberg, Germany) (^{Neuhaus-Steinmetz et al., 2000}; ^{Glaab et al., 2001}). For determination of sensory irritation, ‘TB’ was determined according to ^{Vijayaraghavan et al. (1993)}; (1994), in response to various concentrations of capsaicin (10, 50, 100, 500, 1000 μg ml⁻¹, 1 min) delivered by a jet nebuliser (Pari-Boy®; Pari-Werke, Starnberg, Germany). OVA (20 μl, 1%) was given intranasally to mice placed in the body plethysmographs, to measure allergen-induced airway obstruction.

Figure 6

Effect of anti-BDNF treatment on allergen-induced airway obstruction by HBP. Measurement of expiratory flow in response to OVA (intranasal application during continuous measurement of airflow in the HBP). (a) Integration of EF₅₀ values between 15 and 25 min after OVA application. (b) Time course of n=11 animals in each group. Data from three separate experiments are shown. Grey bar: possible artefacts from OVA application. The error bars represent standard error of the mean (s.e.m.). Student́s t-test: ***P<0.001.

ELISA, RIA and immunohistochemistry

Cytokine levels were measured in cell-free BAL supernatants using ELISA, as previously described (^{Braun et al., 1999}). Neuropeptide concentrations and distribution were analysed by RIA and immunohistochemistry, respectively. The methods are described in detail in an earlier study (^{Fischer et al., 1996}). In brief, tissue concentrations of substance P/neurokinin A and CGRP were measured by a combination of HPLC and RIA. Immunohistochemistry was performed using Zamboni's fixed vagal sensory ganglia and antigen preabsorption (10 ng peptide ml⁻¹ diluted primary antiserum) for the control of specificity. Figure 2 shows substance P/neurokinin A staining of vagal sensory ganglia (a) as well as a control section incubated with antigen-preabsorbed antiserum (b). Ganglia for neuropeptide experiments were taken from animals of the experiments for Figure 7.

Figure 2

Immunohistochemical detection of tachykinins. Substance P/neurokinin A immunohistochemistry was performed using Zamboni's fixed vagal sensory ganglia and antigen preabsorption for the control of specificity. The upper picture (a) shows substance P/neurokinin A staining of vagal sensory ganglia, the lower picture a control section incubated with antigen-preabsorbed antiserum (b). Bars represent 50 μm.

Figure 7

Effect of BDNF and anti-BDNF treatment on inflammation. (a, b) Cytokine levels in BALF. IL-4, IL-5 and IFN-γ were measured in BALF from nonsensitised or OVA-sensitised mice 24 h after aerosol challenge by ELISA. The BAL recovery was 1.4±0.2 ml in all groups. (c, d) Cell differentiation in BALF after OVA aerosol challenge on day 28. Lymphocytes, macrophages, eosinophils and neutrophils were differentiated according to morphological criteria in BALF from nonsensitised or OVA-sensitised mice 24 h after the last aerosol challenge. (e) Cell differentiation in BALF 24 h after nasal OVA challenge on day 36. Student's t-test did not reveal any statistically significant difference (P<0.05) between groups. In all, 12 mice were analysed in each group, in two separate experiments. The error bars represent standard deviation (s.d.).

Statistical analysis

Results are presented as mean values±s.d., unless otherwise stated. Student's t-test was used to determine the level of difference between animal groups, unless otherwise stated. P<0.05 was regarded as a statistically significant difference.

Results

Effect of BDNF on airway hyper-responsiveness

Electrical field stimulation

To investigate the specific contribution of BDNF to the development of airway hyper-responsiveness in our animal model of allergic asthma, airway hyper-responsiveness was assessed with various methods and stimuli. First, we assessed airway hyper-responsiveness in vitro by EFS of tracheal segments. The frequency that causes 50% of maximal airway smooth muscle constriction (ES₅₀) was at 4.1 Hz in nonsensitised animals after nasal BSA control treatment and at 4.2 Hz in completely untreated mice. In contrast, OVA-sensitised and challenged mice developed airway hyper-responsiveness, as indicated by a significant decrease in ES₅₀ values to 2.4 Hz (data not shown). In a further experiment, allergen-sensitised and challenged animals were treated with anti-BDNF antibodies or isotype-matched control antibodies by intranasal application of an anti-BDNF polyclonal antibody delivered on days 26 and 27. There was one additional i.p. application of anti-BDNF polyclonal antibodies or isotype-matched control antibodies on day 25 (Figure 1). Mice treated with isotype-matched control antibodies showed a regular hyper-responsiveness following allergen challenge (Figure 3b). Anti-BDNF treatment almost completely eliminated the decrease in ES₅₀ values (Figure 3b). We then investigated whether BDNF treatment by itself could be sufficient to alter airway responsiveness. Nonsensitised naive mice were treated nasally with BDNF or BSA on two consecutive days, and airway responsiveness was assessed 24 h later. BDNF-treated mice showed marked airway hyper-responsiveness in response to EFS as compared to BSA-treated control mice (Figure 3a). This airway hyper-responsiveness seemed to be even stronger than that observed in sham-treated allergen-challenged mice, though a direct comparison of these data is not possible due to the different experimental settings (Figure 3a, b).

Figure 3

Effect of BDNF on airway hyper-responsiveness in response to EFS. Airway hyper-responsiveness was measured in response to EFS. The frequency that caused 50% of maximal airway smooth muscle constriction was defined as ES₅₀. Airway contractility was expressed as mean±s.d. Statistics were performed using t-test with Welch's correction. P<0.05 was regarded as a statistically significant difference. (a) Nonsensitised BALB/c mice were treated intranasally either with recombinant human BDNF (5 μg ml⁻¹, 50 μl) or BSA (0.1%, 50 μl) as a control on two consecutive days. AHR was measured 24 h after the last treatment. In all, 12 animals were examined in each group in two separate experiments. (a) OVA-sensitised Balb/c mice were treated intranasally either with anti-BDNF (500 μg ml⁻¹, 50 μl) or istotype IgY (500 μg ml⁻¹, 50 μl) as a control on two consecutive days, 3 h before OVA aerosol challenge. In addition, the animals received 100 μl of the antibody solution i.p. 24 h before the first OVA challenge (day 25). AHR was measured 24 h after the last treatment. A total of 12 animals were examined in each group in two separate experiments.

Head-out body plethysmography

To further analyse the underlying mechanisms, a mouse head-out body plethysmographic system was used. This system analyses the breathing pattern of each animal by continuously registering changes in airflow during ins- and expiration. By using specific stimuli, it can distinguish between bronchoconstriction due to direct stimulation of airway smooth muscle and changes in the breathing pattern due to sensory nerve activation. Capsaicin is a potent and selective stimulant of sensory nerves and induces a characteristic modification of the normal breathing pattern, which is quantitatively and qualitatively assessed by measuring the duration of braking (‘TB’) (see Introduction for further explanation). As shown in Figures 4 and 5d, capsaicin induced a marked dose-dependent increase in ‘TB’ in normal mice. In contrast, methacholine acts via direct stimulation of airways smooth muscle cells. As shown in Figure 5c, methacholine has no effect on ‘TB’, but results in a characteristic dose-dependent decrease in mid-expiratory flow rate (EF₅₀) (Figure 5a).

Figure 4

Characteristic modifications to the normal breathing pattern in response to capsaicin measured by HBP. (a) Normal breathing pattern (airflow) of nonsensitised BALB/c mice while breathing room air. Normal breathing patterns were virtually identical in sensitised and nonsensitised mice. (b) Characteristic pattern of sensory irritation (lengthening of TB) of nonsensitised BALB/c mice in response to capsaicin (500 μg ml⁻¹). The breathing pattern before capsaicin challenge is identical to breathing pattern shown in (a). (c) Characteristic pattern of sensory irritation (lengthening of TB) of OVA-sensitised BALB/c mice in response to capsaicin (500 μg ml⁻¹).

Figure 5

Effect of anti-BDNF treatment on airway hyper-responsiveness in response to methacholine and capsaicin by HBP. (a) Dose–response curve of expiratory flow (EF₅₀) in response to inhaled methacholine. (b) Dose–response curve of expiratory flow (EF₅₀) in response to inhaled capsaicin. (c) Dose–response curve of sensory irritation (TB) in response to inhaled methacholine. (d) Dose–response curve of sensory irritation (TB) in response to inhaled capsaicin. The error bars represent standard error of the mean (s.e.m.), n=8–10 animals in each group in three separate experiments. Student's t-test: *P<0.05. (a) Nonsensitised against OVA-sensitised+isotype, (d) OVA-sensitised+isotype against OVA-sensitised+anti-BDNF.

Methacholine and capsaicin were used to further analyse the specific contribution of BDNF to the development of airway hyper-responsiveness (Figure 5). As shown in Figure 5a, allergen-sensitised and -challenged animals develop increased airway responsiveness to methacholine, characterised by a significant (P<0.05) leftward shift of the dose–response curves to aerosolised methacholine. This response was not altered by anti-BDNF treatment. In contrast, hyper-reactivity to capsaicin was almost completely blocked by anti-BDNF treatment.

Effect of BDNF on allergen-induced airway obstruction

As recently demonstrated (^{Neuhaus-Steinmetz et al., 2000}), a re-challenge of OVA-sensitised and allergic mice at weekly intervals causes chronic airway obstruction, as indicated by a decrease in baseline EF₅₀ values (Figure 6). In this model, additional acute intranasal OVA-challenge further decreased EF₅₀ values 10–30 min post challenge (Figure 6). This model was employed to assess the contribution of BDNF to allergen-induced airway obstruction. Anti-BDNF treatment partially prevented chronic airway obstruction, as indicated by a significant (P<0.05) increase in baseline EF₅₀ value when compared to sham-treated animals. After an acute allergen challenge, these mice showed a rapid decrease in EF₅₀ values within 10 min post challenge, which persisted for at least 30 min. The magnitude of this response was similar in both groups, indicating that anti-BDNF treatment did not affect the strength of the acute response following allergen challenge. On the contrary, the differences in the mid-expiratory flow following allergen challenge seem to reflect pre-existing differences in airway calibre between the groups.

Effect of BDNF on airway inflammation

Murine allergic airway inflammation is characterised by a characteristic change of the cytokines interleukin-4 (IL-4), interleukin-5 (IL-5) and interferon-γ (IFN-γ), as well as increased eosinophil, monocyte and lymphocyte counts in the bronchoalveolar lavage (^{Braun et al., 1999}). BDNF might decrease airflow by affecting the inflammatory response in the airways. To investigate this possibility, we assessed IL-4, IL-5 and IFN-γ levels as well as differential cell counts in BALFs obtained from anti-BDNF or sham-treated mice. Blocking of BDNF neither quantitatively nor qualitatively affected the pattern of allergic inflammation (Figures 7b, d). In addition, anti-BDNF treatment had no effect on the cellular influx of eosinophils and lymphocytes in acutely OVA-challenged mice on day 36 (Figure 7e). Furthermore, nasal administration of recombinant BDNF to nonsensitised naive mice did not induce a local inflammatory response (Figures 7a, c).

Effect of BDNF on neuronal neuropeptide content

Neurotrophins may modulate afferent nerve functions via several mechanisms, including enhanced neuropeptide production and releasability. In order to assess whether the functional changes in sensory neurons are associated with altered neuropeptide levels, tachykinin and CGRP content in sensory ganglia was measured 24 h after the last allergen challenge. The data presented in Table 2 compare the results obtained with anti-BDNF and sham-treated allergic mice. There was no detectable difference between the groups in (a) number of tachykinin or CGRP-containing neurons or (b) tachykinin or CGRP content in various ganglia.

Table 2. Tachykinin containing sensory neurons from ovalbumin-sensitised mice

	Ganglion V	Ganglion IX/X	Ganglion IX/X
	SP/NKA (pg mg^−1 wet weight)	SP/NKA (pg mg^−1 wet weight)	SP/NKA positive neurons (%)
IgY	15.8±1.7	7.9±2.4	6.5±0.8
Anti-BDNF	13.0±1.8	8.0±0.8	6.0±1.1

• Immunohistochemistry for SP/NKA revealed immunoreactive perikarya that were counted for each ganglion and are expressed as % of total neurons. The neuropeptide content of each ganglion is expressed as pg mg⁻¹ wet weight. The ganglia were prepared 24 h after the last allergen provocation. Each value represents the mean of four animals±s.e.m. Student's t-test did not reveal any statistically significant difference (P<0.05) between groups.

Discussion

The aim of this study was to analyse the role of BDNF with respect to airway function in allergic asthma, using a well-established murine model of allergic airway inflammation. In order to characterise the specific contribution of BDNF to altered lung function, we measured allergen-induced airway obstruction as well as airway hyper-responsiveness to various nonsensitising stimuli in anti-BDNF treated and in sham-treated control mice. We were able to distinguish between airway dysfunctions mediated by smooth muscle changes and those mediated by neuronal alterations. This differentiation is especially important for the evaluation of airway hyper-responsiveness. Airway hyper-responsiveness following allergic inflammation is mediated by multiple independent and additive pathways working in concert (^{Herz et al., 1998}; ^{Wilder et al., 1999}; Wills-Karp, 1999). Dependent on the method chosen for the measurement of airway hyper-responsiveness, different pathways can be distinguished. They include: (a) altered neuronal regulation of airway tone, (b) increased smooth muscle content or function, and (c) increased epithelial mucus production and airway edema (Wills-Karp, 1999). Recent studies with asthmatic patients revealed an individual contribution of neuronal and smooth muscle hyper-reactivity to airway hyper-responsiveness (^{Millqvist et al., 1998}).

Neutralisation of BDNF by local and systemic treatment with specific antibodies reduced allergen-induced chronic airway obstruction in vivo, as shown by HBP. Allergen-induced airway obstruction has recently been demonstrated in mice using body plethysmography (^{Cieslewicz et al., 1999}; ^{Neuhaus-Steinmetz et al., 2000}). The time course of BDNF in BALF showed an elevation over 2–3 weeks and a late peak on day 7 after the last allergen challenge (^{Lommatzsch et al., 2003}). Therefore, we examined the effects of BDNF neutralisation on airway obstruction on day 7 after the last allergen challenge. Interestingly, EF₅₀ values were still decreased in allergen-challenged mice versus control mice at this late time point. Neutralisation of BDNF significantly increased these EF₅₀ values. The decrease of EF₅₀ in response to allergen re-challenge was concomitantly altered. However, the latter effect seems to reflect only an altered basal airway obstruction, since the differences in EF₅₀ between anti-BDNF-treated and control-antibody-treated groups are nearly the same before and after allergen re-challenge. We conclude that neutralisation of BDNF has profound effects on the basal allergen-induced airway obstruction, even 1 week after the initial allergen challenge. One explanation for this phenomenon could be an enhanced basal parasympathetic tone of the airways. An alternative explanation could be an inhibition of sensory nerve-induced plasma leakage, mucus production and oedema, which could affect airway calibre as well. Significant obstructions due to altered inflammatory conditions are unlikely, since anti-BDNF did not affect inflammatory parameters.

Using EFS in vitro, we have further shown that anti-BDNF treatment was able to prevent local neuronal hyper-reactivity in the airways. It has been previously demonstrated that in vitro EFS of tracheal segments specifically reflects neuronal airway obstruction. Administration of both atropine (disruption of cholinergic pathways) and capsaicin (depletion of sensory neurons) completely blocks responsiveness of tracheal segments to electric field stimulation (Andersson & Grundstrom, 1983; Ellis & Undem, 1992). We used, therefore, in vitro EFS to examine the effects of BDNF on neuronal dysfunction. Locally administered BDNF was sufficient to induce airway hyper-responsiveness in response to EFS in the absence of inflammation. In addition, neutralisation of BDNF in allergen-sensitised and -challenged mice prevented airway hyper-responsiveness to EFS, which normally follows allergic airway inflammation. Thus, BDNF seems to have direct effects on neuronal control of the airway diameter. These effects seem to be independent of the local inflammatory situation, since neither BDNF nor anti-BDNF affected any of the inflammatory parameters measured.

In vivo experiments strengthened the idea that anti-BDNF treatment primarily affects neuronal parameters. Anti-BDNF treatment did affect the altered reactivity of sensory neurons (to capsaicin), but not the altered reactivity of airway smooth muscle (to methacholine). It needs to be emphasised that increased sensory irritation in response to capsaicin, as shown by a lengthening of the first phase of expiration, does not necessarily imply airway obstruction. In fact, it appears that capsaicin and neuropeptides such as substance P are bronchodilators in rodents (Manzini, 1992; ^{Szarek et al., 1998}). In addition, altered capsaicin reactivity does not necessarily imply altered neuropeptide levels in sensory neurons, as shown by our immunohistochemical studies. The capsaicin experiments in our present study were solely carried out to answer the general question whether anti-BDNF treatment can affect the reactivity of sensory neurons. The mechanisms of enhanced reactivity in response to capsaicin and the effects on airway diameter are species-specific and cannot be extrapolated from our results.

Our data support the idea that BDNF could primarily act on the reactivity of innervating neurons. Since BDNF neither affected inflammatory conditions in allergen-challenged mice, nor induced any inflammation by itself, it seems unlikely that it affects airway functions indirectly via inflammatory alterations. In addition, direct effects on airway smooth muscle cells seem to be unlikely, since anti-BDNF treatment did not affect airway hyper-responsiveness in response to methacholine. In contrast, the profound effects of BDNF and anti-BDNF on airway hyper-responsiveness measured by in vitro EFS, as well as the effects of BDNF on the capsaicin response in vivo, demonstrate that BDNF could directly affect neuronal control. We hypothesise, therefore, that enhanced BDNF production by immune cells in the lung (e.g., T-cells or macrophages) (^{Virchow et al., 1998}; ^{Braun et al., 1999}) could substantially contribute to neuronal dysfunction.

It has been well established that a variety of sensory and motor neurons innervating the adult viscera express the high-affinity BDNF receptor trkB (Wetmore & Olson, 1995; ^{Kashiba et al., 1997}; ^{Michael et al., 1997}; ^{Zhou et al., 1998}). On the other hand, BDNF receptors are not detectable on resident non-neuronal cells in the adult lung (^{Lommatzsch et al., 1999}). The abundance of BDNF in the adult viscera, especially in the lung, and the retrograde transport of BDNF in sensory and motor neurons with their axons in the vagus nerve indicate that target-derived BDNF could indeed influence innervating neurons (^{Helke et al., 1998}; ^{Lommatzsch et al., 1999}). Effects on both sensory and motor neurons have to be considered. For example, altered reactivity of sensory neurons to capsaicin could be mediated by BDNF-induced VR-1 receptor changes in vagal sensory neurons (Winter, 1998). Effects of BDNF on motor neurons may be especially important regarding airway obstruction. BDNF is retrogradely transported in vagal motor neurons and can affect a variety of motor neuron functions, including excitability, ion channel expression, synaptic composition and neurotransmitter production (Gonzalez & Collins, 1997; ^{Fernandes et al., 1998}; ^{Helke et al., 1998}; ^{Baldelli et al., 1999}; ^{Novikov et al., 2000}). Effects of the closely related neurotrophin NGF on enhanced reactivity in response to capsaicin have already been shown (^{Hoyle et al., 1998}; ^{Winston et al., 2001}). Interestingly, NGF seems to have similar effects as BDNF on airway sensory nerves (^{Päth et al., 2002}). This could be due to the fact that NGF affects airway sensory neurons by activating the pan-neurotrophin (p75) receptor (^{Kerzel et al., 2003}). Notably, NGF does have profound effects on allergic inflammation (^{Päth et al., 2002}), whereas BDNF does not. This could be due to specific effects of NGF on neuronal tachykinin production, as well as direct trkA-mediated effects on immune cells (^{Päth et al., 2002}).

In conclusion, our experiments demonstrate that BDNF could serve as a link between allergic airway inflammation and neuronal dysfunction in allergic asthma. We have found evidence that BDNF is directly acting on neuronal tissues in the lung, inducing neuronal hyper-reactivity and chronically enhancing the basal airway tone.

Acknowledgments

This work was supported by the DFG, the SFB 587, B4 and the Volkswagen Foundation. We thank Margarita Strozynski and Christine Seib for excellent technical assistance. Special thanks to Mary Haak-Frendscho from Promega Corporation for providing the anti-BDNF antibody.