Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Carbon monoxide (CO) is formed in neurons as a byproduct of the enzymatic degradation of heme by the constitutive isoform of heme oxygenase, heme oxygenase-2 (HO-2; refs. 1-4). HO-2 has been localized to neurons throughout the central and peripheral nervous system (3, 5). The presence of HO-2 in many neurons combined with functional studies demonstrating multiple effects of CO on both neurons (3, 4, 9) and innervated tissues (6, 8, 10) has led to speculation that like nitric oxide (NO), CO might serve as an inter- and intracellular neuronal messenger molecule (3).

Like NO, CO mediates its effects in nerves and in other tissues by binding to the heme moiety of the soluble isoform of guanylate cyclase (3, 11, 14). Upon binding guanylate cyclase, CO induces formation of the second messenger cyclic guanosine monophosphate (cGMP) which has neuromodulatory effects in the central and peripheral nervous system and mediates a variety of other effects in nonneural tissues, including relaxation of smooth muscle (4, 6, 11, 12, 15, 16).

Given its pharmacologic similarities to NO and the many effects attributed to cGMP in pulmonary and airway cells, it is possible that CO, as one of only a handful of molecules that activate soluble guanylate cyclase (1), might play a modulatory and/or regulatory role in the airways and lungs similar to that suggested for NO (17). Consistent with this hypothesis, HO-2 has been localized to airway smooth muscle and to nerve fibers in the pulmonary artery in guinea pigs and pigs, respectively (6, 18). Furthermore, the inducible form of heme oxygenase, HO-1, is expressed in rat lung following hyperoxic challenge or exposure to lipopolysaccharide (LPS) (19, 20). To test the hypothesis that CO plays a neuromodulatory role in the airways and lung, the distribution of HO-2 in nerves and intrinsic neurons of the airways of both humans and guinea pigs was assessed, as were the potential neuromodulatory effects of heme oxygenase activity on the parasympathetic nerves innervating the guinea-pig trachea.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Tissue Preparation

Adult female guinea pigs (Hartley, 250-300 g; Charles River, Kisslegg, Germany) were killed by inhalation of 100% CO₂. The chest was opened and the animals were perfused with a rinsing solution (containing 0.9% NaCl, 2.5% polyvinylpyrollidon, 0.5% procaine hydrochloride, and 5,000 U/L heparin) through a cannula placed in the ascending aorta followed by fixation with 4% formaldehyde. Thoracic viscera were removed in toto, washed repeatedly in 0.1 M phosphate buffer (0.1 M NaH₂PO₄, pH = 7.4), and stored overnight in cold (4°C) 0.1 M phosphate buffer containing 18% sucrose for cryoprotection. Tissues were mounted on filter paper in optimum cutting temperature (OCT) compound, frozen in liquid nitrogen, and sectioned (12 µm) on a cryostat (Frigocut, 2000E; Leica Instruments, Nussloch, Germany).

Samples of human trachea (kindly provided by Drs. Belvisi and Yacoub, National Heart, Lung and Blood Institute, London, UK) and macroscopically healthy samples of large and small bronchi obtained from patients undergoing lung resection for tumors (kindly provided by Dr. Rabe, Krankenhaus, Grosshansdorf, Germany) were fixed by immersion in Zamboni's fixative (15% saturated picric acid and 2% paraformaldehyde in 0.1 M phosphate buffer, pH = 7.4). After fixation, tissues were washed in 0.1 M phosphate buffer, stored overnight in cold (4°C) 0.1 M phosphate buffer containing 18% sucrose for cryoprotection, frozen in liquid nitrogen after mounting on filter paper in OCT compound, and sectioned (12 µm) as described above.

Immunohistochemistry

Sections of human airway or guinea-pig thoracic viscera were placed on chromalum-coated slides and allowed to air-dry for 30 min. Nonspecific labeling was blocked by coating the slides with 0.1 M phosphate buffer containing 1% bovine serum albumin and 10% normal swine serum for 1 h. After washing in phosphate-buffered saline (PBS) the tissues were then coated with PBS containing a rabbit polyclonal antisera to HO-2 (StressGen, Victoria, Canada; 1:1,500). After overnight incubation at room temperature and washing in PBS, slides were subsequently incubated with a biotinylated goat antirabbit IgG (Amersham, Braunschweig, Germany; 1:200) for 1 h and then washed. Secondary antiserum was detected with a strepavidin-Texas Red conjugate (Amersham; 1:50).

In some experiments, double labeling for both HO-2 and the nonspecific neuronal marker PGP 9.5 (mouse monoclonal; Biotrend, Cologne, Germany; 1:160) was performed. PGP 9.5 immunoreactivity was visualized using fluorescein isothiocyanate-labeled antimouse IgG from goats (1:50, Amersham). Likewise, double labeling experiments were also carried out for PGP 9.5 and HO-1. HO-1 was localized using polyclonal antisera from rabbits (1:1,000; kindly provided by B. Dwyer, Los Angeles, CA). The secondary antisera for HO-1 antibody was the same as that used for HO-2.

Slides were coverslipped in carbonate-buffered glycerol (pH = 8.6) and viewed using epifluorescence microscopy (Olympus BX60F, Hamburg, Germany) as previously described (21, 22).

Negative control experiments for labeling of HO-2 immunoreactive neurons and nerve fibers were performed on sections of both human and guinea-pig airways by preabsorption of the primary antiserum with recombinant HO-2 protein from Escherichia coli (StressGen, 20 µg/ml) prior to overnight incubation on the slides.

Functional Studies

The potential modulatory effects of CO formed from heme oxygenase on synaptic neurotransmission in the guinea-pig trachea was assessed using the isolated, innervated guinea-pig trachea (23). Guinea pigs were asphyxiated in a vessel filled with 100% CO₂ and exsanguinated. The trachea, esophagus, and associated extrinsic nerves were removed in toto and placed in a water-jacketed dissecting dish continuously overfilled (20 ml/min) with warmed, oxygenated Krebs buffer of the following composition (mM): NaCl (118), KCl (5.4), NaH₂PO₄ (1), MgSO₄ (1.2), CaCl₂ (1.9), NaHCO₃ (25), and dextrose (11.1). The trachea, recurrent laryngeal nerves, and vagus nerves were dissected free from extraneous tissues (including the adjacent esophagus) and a segment of the rostral trachea (rings 6 and 7 caudal to the larynx) was prepared for isometric tension measurements. Optimal tone (1.5 g) was set and continually adjusted throughout the 90-min equilibration period. Vagally-mediated cholinergic contractions were elicited by stimulating (24 Hz, 10 s, 1-150 V, 1 msec pulse duration) the right vagus nerve caudal to the nodose ganglia with suction electrodes as previously described (23).

The effects of 8-bromo-cGMP (8-Br-cGMP; 30 µM), the heme oxygenase inhibitor zinc protoporphyrin-IX (30 µM), and the soluble guanylate cyclase inhibitor ODQ (3 µM) on vagally-mediated cholinergic contractions were assessed in unpaired experiments (the concentrations utilized were chosen based on the results of previous investigations; refs. 6, 12, 24, 25). Vagally-mediated contractions were elicited before (control) and 20-30 min after drug or vehicle administration (treated). The effects of the treatments were assessed by comparing the treated responses with those elicited under control conditions.

At the end of each experiment, 300 mM barium chloride was added to elicit a maximum contraction. Unless otherwise stated, vagally-mediated cholinergic contractions were expressed as a percentage of this maximum contraction.

All experiments were carried out in the presence of the -adrenoceptor antagonist propranolol (1 µM) and the cyclooxygenase inhibitor indomethacin (3 µM) to limit the influence of adrenergic nerve stimulation and neuromodulatory prostanoids on the vagally-evoked responses (23).

Reagents

Atropine sulphate, dl-propranolol hydrochloride, indomethacin, and barium chloride were purchased from Sigma (St. Louis, MO). 8-Br-cGMP (sodium salt) was purchased from Calbiochem (La Jolla, CA). ODQ (1H-[1,2,4]oxidiazolo[4,3-a]quinoxalin-1-one) and zinc protoporphyrin-IX were purchased from Tocris Cookson (St. Louis, MO). Trimethaphan camsylate was obtained from Roche Laboratories (Nutley, NJ). Atropine (10 mM), propranolol (10 mM), 8-Br-cGMP (0.1 M), and barium chloride (1 M) were dissolved in water. Trimethaphan was purchased dissolved in 0.013% sodium acetate. ODQ (0.1 M) was dissolved in dimethylsulfoxide. Indomethacin (30 mM) was dissolved in ethanol. Zinc protoporphyrin-IX (10 mM) was dissolved in 0.2 N NaOH and stored in the dark until use (due to its light sensitivity, all experiments with zinc protoporphyrin-IX were carried out in low light). All drug solutions were made fresh daily.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Immunohistochemistry

View larger version (122K): [in this window] [in a new window]   Figure 1.   HO-2 immunoreactivity in human airway ganglia. (a) Nonspecific fluorescence due to lipofuscine in slides labeled for HO-2 was found to be minimal in the sections viewed in this study, including the tracheal ganglion photographed in a. (b) Viewing the same section with the appropriate filter combination reveals that the HO-2 antisera labeled all perikarya within this ganglion. HO-2 immunoreactivity was also detected in all perikarya of intrinsic ganglia associated with (c) small and (d) large bronchi. (e) Higher magnification of the tracheal ganglion depicted in a and b reveals that HO-2 immunolabeling of the perikarya varied from weak (arrowhead) to strong (large arrows). Occasionally, neuronal processes are labeled (small arrows). (  f  ) In the airway wall, where nerve fibers were readily identified in the epithelium (upper part of f  ), lamina propria, and smooth muscle with the nonspecific neuronal marker PGP 9.5 (see Figure 2), no HO-2 immunoreactive nerve fibers were identified. Each micrograph is representative of at least five separate experiments. Scale bars represent 50 µm in a-d and f, and 20 µm in e.

View larger version (102K): [in this window] [in a new window]   Figure 2.   Lack of immunoreactivity to the inducible form of heme oxygenase, HO-1, in (a) nerve fibers and (c) a bronchial ganglion in the wall of a human bronchus. Double labeling of nerve fibers and neuronal perikarya with the nonspecific neuronal marker PGP 9.5 in the sections negatively stained for HO-1 reveals (b) a dense network of nerve fibers in the smooth muscle of the bronchus and (d) the location of the neuronal perikarya within the ganglion in the sections photographed in a and c, respectively. Each panel is representative of at least three experiments. Bar = 20 µm.

View larger version (139K): [in this window] [in a new window]   Figure 3.   Localization of HO-2 immunoreactivity to guinea-pig airway nerves. All neuronal cell bodies in (a) tracheal and (c) bronchial ganglia are readily labeled with antisera to HO-2. Specificity of labeling is revealed in (b) where the antisera preabsorbed with the recombinant HO-2 protein prior to overnight incubation failed to label the ganglion present in the field photographed. (d) No nerve fibers displaying immunoreactivity to HO-2 were observed innervating any structure within the airway wall even though (e) airway ganglia and (  f  ) nerve fibers throughout the sections were readily labeled with antisera to the nonspecific neuronal marker PGP 9.5. Each micrograph is representative of at least five experiments. Bars are 20 µm in a-c and e, and 10 µm in d and f.

Functional Studies

The presence of HO-2 in the neuronal perikarya of the intrinsic ganglia of the airways and not in the nerve fibers innervating structures in the airway wall suggests that CO-mediated neuromodulatory effects might be limited to effects on synaptic neurotransmission mediated by preganglionic nerves. This hypothesis was tested using the isolated, innervated guinea-pig trachea. Stimulation (24 Hz, 10 s, 150 V, 1 msec pulse duration) of the right vagus nerve elicited cholinergic contractions of the guinea-pig trachea that averaged 41 ± 5% of the maximum attainable contraction elicited by 300 mM barium chloride (n = 11). Addition of the stable, pharmacologically active cGMP analog 8-Br-cGMP (30 µM) virtually abolished vagally-mediated contractions (Figure 4). By contrast, neither the heme oxygenase inhibitor zinc protoporphyrin-IX (30 µM) nor the soluble guanylate cyclase inhibitor ODQ (3 µM) had marked effects on vagally-mediated contractions (Figure 4).

View larger version (4K): [in this window] [in a new window]   View larger version (12K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   Figure 4.   Effects of (top panel ) 8-Br-cGMP, (middle panel ) zinc protoporphyrin-IX (Zn PP IX), and (bottom panel ) ODQ on vagally-mediated contractions of the guinea-pig trachea. The right vagus nerves were stimulated (24 Hz, 10 s) at various stimulus intensities to generate voltage-response curves for cholinergic contractions. Contractions are expressed as a percentage of the maximum contractile response elicited by vagus nerve stimulation (24 Hz, 10 s, 150 V). The parasympathetic and cholinergic nature of the contractions elicited were confirmed based on their sensitivity to the ganglionic blocker trimethaphan (0.1 mM; n = 3) or the muscarinic receptor antagonist atropine (1 µM; n = 3), both of which abolished the vagally-mediated contractions. (top) Representative trace illustrating the effects of the stable, pharmacologically active cGMP analog 8-Br-cGMP (30 µM) on vagally-mediated contractions of the guinea-pig trachea. The right vagus nerve was stimulated (24 Hz, 10 s; denoted by short horizontal bars) before and 20 min after addition of 8-Br-cGMP. The 8-Br-cGMP abolished vagally-mediated contractions at submaximal stimulus intensities ( 50 V) and markedly inhibited (91 ± 4% inhibition; n = 4; P < 0.05) contractions elicited at the maximum (150 V) stimulus intensity studied. By contrast, neither inhibiting (middle) heme oxygenase with Zn PP IX (30 µM; n = 3) nor inhibiting (bottom) soluble guanylate cyclase with ODQ (3 µM; n = 4) had significant effects on vagally-mediated contractions of the trachealis.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Functional and morphologic studies carried out in the brain and spinal cord and in peripheral tissues including the gastrointestinal tract and the pulmonary artery are consistent with the hypothesis that CO formed by the enzymatic cleavage of heme might have a neurotransmitter and/or neuromodulatory role both in the central and peripheral nervous system (4, 6, 8, 9, 12, 26). Thus the constitutive form of heme oxygenase, HO-2, is widely distributed in the central nervous system (3) and in peripheral neurons (6); and further, CO added exogenously and acting through soluble guanylate cyclase can mimic nerve-mediated effects that are sensitive to heme oxygenase inhibition (1, 3, 4, 6, 8, 9, 12, 29, 30).

Given the many pharmacologic properties that CO shares with NO combined with the evidence that NO plays a role at myriad autonomic synapses (4, 31), it is reasonable to speculate that CO might also be a neurotransmitter released from autonomic nerves. Indeed, Rattan and Chakdar (12) presented compelling evidence for endogenous CO-mediated effects in the internal anal sphincter of the opposum. Further evidence for a role of CO in the autonomic nervous system comes from studies with preparations of isolated tissues and cells and in vivo studies with transgenic HO-2 knockout mice (6, 8, 13, 30, 32). By contrast, we have previously presented evidence that CO may not be a neurotransmitter released from guinea-pig airway autonomic nerve endings (18).

In the present study, HO-2 immunoreactivity was found in all perikarya of the intrinsic ganglia in both human and guinea-pig airways but was conspicuously absent in nerve fibers within the ganglia and in nerve fibers innervating other structures in the airway wall. This observation is consistent with our previous studies indicating that CO is not a neurotransmitter in the airway smooth muscle but suggests that CO might modulate synaptic neurotransmission. Thus perhaps CO, acting through the soluble isoform of guanylate cyclase, could alter synaptic efficacy within the airway parasympathetic ganglia. Indeed, cGMP is known to modulate synaptic neurotransmission and NO, presumably through the actions of cGMP formed upon activation of guanylate cyclase, also has marked neuromodulatory effects at central and peripheral synapses and nerve terminals (1, 15, 16).

The immunohistochemical evidence notwithstanding, functional studies described here fail to demonstrate a role for CO in regulating synaptic neurotransmission in the airways. Thus neither inhibiting heme oxygenase activity with zinc protoporphyrin-IX nor inhibiting soluble guanylate cyclase activity with ODQ markedly affected vagally-mediated cholinergic contractions of the guinea-pig trachea. These observations do not, however, preclude the possibility that under altered conditions CO might modulate synaptic neurotransmission through formation of cGMP. Indeed, 8-Br-cGMP markedly inhibited vagally-mediated contractions of the trachealis. While the site of action (pre- or postjunctional) for this effect of the stable analog of cGMP was not determined, the observation is at least consistent with the hypothesis that CO might play a neuromodulatory role in the airways. Likewise, while HO-1 immunoreactivity was not found in neuronal perikarya or nerve fibers of the human airways, the possibility that CO formed by HO-1 might play a neuromodulatory role in the airways can also not be discounted. Indeed, HO-1 has been found in lungs from rats exposed to hyperoxic stimuli or LPS (19, 20). HO-1 gene and protein expression can also be induced in neurons (33) and preliminary evidence indicates that HO-1 may also be induced in the lung of sensitized guinea pigs following aerosolized antigen challenge (W. Kummer, personal communication).

Alternatively, heme oxygenase might not play a neuromodulatory role in the lung but may subserve only a metabolic role in airway nerves. The observation that HO-2 immunoreactivity was localized to neuronal perikarya, occasionally in the proximal portions of dendrites, and absent in nerve fibers within the ganglia and nerve fibers elsewhere in the airway wall is consistent with previous reports that HO-2 may be localized to the metabolically active endoplasmic reticulum (7, 34).