INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: airway; asthma; smooth muscle; bronchoconstriction

The autonomic control of lower airway (trachea and bronchi) smooth muscle, and consequently airway caliber, is primarily parasympathetic in nature. Preganglionic axons are carried to the airways by the vagus nerves and these axons form the preganglionic component of synapses on principal neurons within ganglia located within or near the airway wall (1). In the lower airway of humans, parasympathetic ganglia contain cholinergic neurons and/or nonadrenergic, noncholinergic (NANC) neurons (2). Stimulation of cholinergic and NANC neurons results in contractions and relaxations of airway smooth muscle, respectively (3). In addition, these neurons regulate, in a tonic fashion, the function of other cell types in the airway wall including glandular cells and the microvasculature (4). Although the parasympathetic nervous system has been implicated in several human airway diseases (5), no study on the cellular characteristics of lower airway intrinsic ganglia neurons in humans has been reported.

Parasympathetic ganglia neurons in the lower airway of laboratory animals display anatomical and electrophysiological properties associated with integrative function (8, 9), that is, they do not relay all signals from the central nervous system (1). Evidence that airway parasympathetic neurons integrate preganglionic stimuli was first provided by Mitchell and colleagues (10) who performed intracellular recordings of cat tracheal parasympathetic ganglia neurons in vivo. They demonstrated that during frequencies of preganglionic action potentials at 10-30 s¹, most of this synaptic input to the ganglia resulted in fast excitatory postsynaptic potentials (fEPSPs) that did not reach threshold for action potential generation, that is, most of the input to the ganglia from the central nervous system is effectively aborted (11). As cholinergic tone of lower airways smooth muscle during normal breathing represents approximately 30% of maximal cholinergic contraction (12), any shift in output from the parasympathetic ganglia could greatly affect airflow to the lungs.

The techniques used in the present study allow, for the first time, recordings from intact, unstained bronchial parasympathetic ganglia neurons from healthy human organ donors, and direct investigation of active, passive, and synaptic electrophysiological properties. Due to the small size and sparse distribution of airway parasympathetic ganglia, there have been relatively few cellular neurophysiological observations reported for these cells. All studies thus far of airway parasympathetic ganglia neurons have been limited to those in research animals (11, 13) and predominantly in tracheal ganglia, often following vital staining (11, 13, 16) or dissociation (15). The purpose of the present study was to determine membrane properties associated with integration of preganglionic stimuli. Such information is crucial in understanding the function of parasympathetic nerves in airway physiology in healthy and diseased lungs. Preliminary results from this study have been previously reported (17).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lungs from 39 human organ donors were received 12-24 h after the donor was cross-clamped and organs were removed. Once received, the bronchus was immediately transferred to and stored in chilled (4-6° C) oxygenated Krebs bicarbonate buffer for up to 12 h. The composition of Krebs buffer was (in mM) NaCl, 136; KCl, 5.6; MgCl₂, 1.2; CaCl₂, 2.2; NaH₂PO₂, 1.2; NaHCO₃, 14.3; dextrose, 11; equilibrated with 95% O₂/5% CO₂ (pH 7.4). Processing airway tissue in this manner is highly effective in maintaining neuronal responses in human airway specimens (18). The average age of the organ donors used in this study was 33 ± 6 yr (range of 7-62 yr), 19 female, 20 male, and 19 were nonsmokers. The bronchi were procured and shipped by The Anatomic Gift Foundation (Laurel, MD), The International Institute for the Advancement of Medicine (Exton, PA), or The National Disease Research Institute (Philadelphia, PA).

Bronchial ganglia were located without staining using techniques as described for guinea pig bronchial ganglia (19). Using transmitted light, the bronchial ganglia were usually near branch points of major bronchial nerve (referred to in mice as peribronchial nerves [20]) on the serosal surface, found laterally and medially at varying distances (2-50 mm) from the vagus nerve. In three preparations the ganglia were identified on secondary bronchi. For studies of synaptic transmission, ganglia on a single peribronchial nerve were chosen over those in a larger plexus of nerves with ill-defined preganglionic nerve input; choline chloride (30 µM) was added to the Krebs buffer if synaptic transmission recordings were planned (21).

Methods for electrophysiological recording and neurobiotin injection have recently been described (19, 22). Thick-walled electrodes were used in most preparations in order to penetrate the perineural sheath and connective tissue over the neuronal soma. Active and passive membrane properties were recorded in a total of 82 neurons from 42 bronchial ganglia. Only neurons with resting membrane potentials < 30 mV and input resistance  10 M were characterized and used in this analysis (57 of 82 neurons). After establishment of a stable resting membrane potential, active and passive membrane properties were determined as previously described (14, 19). Total input capacitance (C_i) was calculated from the equation C_i = /R_i where  is equal to the time constant and R_i is equal to the input resistance for each neuron. Synaptic potentials were elicited by electrical stimulation of the preganglionic peribronchial nerve at a point between the vagus nerve and the ganglion (14). To test the calcium dependence of neurotransmitter release or fluctuations in the resting potential, nominally calcium-free buffer was used; its composition was the same as Krebs buffer (above) except CaCl₂ was replaced on an equimolar basis with MgCl₂. Hexamethonium was applied for at least 10 min prior to drug application or peribronchial nerve stimulation.

All reagents used to prepare the Krebs solution were purchased from J.T. Baker Chem. Co. (Phillipsburg, NJ). Choline chloride, 1,1-dimethyl-4-phenylpiperazinium (DMPP), and hexamethonium bromide were obtained from Sigma Chemical Co (St. Louis, MO).

Results are presented as means ± 1 standard error of the mean. Sample mean values were analyzed by analysis of variance and the values comparing properties of tonic- and phasic-type neurons, and effect of high-frequency stimulation of fEPSP amplitude were compared using Student's t statistics for two means. Means were considered to differ significantly if p values were < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Distribution and Morphology of Human Bronchial Ganglia Neurons

The ganglia used in this study were located on or in peribronchial nerves and occasionally on finer (< 200 µm; Figure 1A) nerve branches away from the peribronchial nerves. Ganglia were located on different layers of the extrachondral (peribronchial) nerve plexus (terminology reviewed in [23]), and never in or below the smooth muscle or cartilage. Ganglia contained as few as 4 and as many as 50 neuronal soma. Examples of unstained, whole-mount ganglia are shown in Figure 1. Unstained ganglia were identified as swellings containing translucent cell bodies (Figure 1A and 1B) along peribronchial nerves; the body of the ganglion usually excluded the surrounding adipocytes, nerve fibers, and/or connective tissue (Figure 1A). Occasionally, larger ganglia contained a darker cell or cells (arrowhead in Figure 1B); these cells were not further studied.

View larger version (44K): [in this window] [in a new window]   Figure 1.   Photomicrographs of unstained human bronchial ganglia in whole-mount (fresh) tissue and a filled neuron. (A) A ganglion (arrow) is located at a division of a peribronchial nerve. Ganglia could be identified as containing clear-core somas of the neurons when viewed with transmitted light. Clear, round structures with dark, defined edges observed near ganglia are adipocytes (a). (B) A higher magnification of an unstained ganglion in fresh tissue showing a darker cell (arrowhead), often found in larger ganglia. (C ) A drawing of a single neuron injected with neurobiotin and processed for peroxidase histochemistry; asterisk indicates axon. Calibration bar is 100 µm in A and 20 µm in B and C.

To determine neuronal shape and presence of neurites, five cells were successfully filled with neurobiotin and stained with avidin-peroxidase-DAB. Neurobiotin-filled neurons had somal shapes that ranged from nearly spherical to oval with a minimum axis ranging from 16 to 26 µm and maximum axis ranging from 22 to 30 µm. One neuron had relatively few short processes (not shown) and four neurons had longer, multiple, and branching dendrites (e.g., Figure 1C). All filled neurons had a single, nontapering, process (presumed axon) extending several millimeters from the ganglion.

Membrane Properties

The resting membrane potential and membrane resistance averaged 47 ± 2 mV (range 31 to 68 mV; n = 57) and 28 ± 3 M (range 10-70 M; n = 57), respectively. The mean time constant for these neurons was 8 ± 1 ms (range 3.9-16.0 ms). The action potential amplitudes and durations averaged 55 ± 4 mV (range 35-70 mV) and 6.8 ± 0.2 ms (range 3.6-9.0 ms), respectively; most action potentials had an overshooting (> 0 mV) depolarization phase of their action potentials. There was no break or "hump" in the depolarizing or repolarizing phase of the action potentials. The amplitude and duration of the afterhyperpolarization (AHP) that followed each action potential were 7.3 ± 1 mV and 64 ± 12 ms, respectively.

Human bronchial ganglia neurons exhibited two distinct action potential firing patterns in response to prolonged suprathreshold depolarizing steps. Phasic neurons (47 of 57 neurons) responded to threshold (0.8 ± 0.1 nA) rectangular anodal constant-current steps, 500 ms in duration, with one action potential at the onset of the stimulus; two to five times threshold stimuli (1.6-4 nA) elicited either one (28 of 47 neurons, Figure 2A) or a burst of two to five action potentials (19 of 47 neurons) in the initial 50 ms of the stimulus followed by accommodation. Tonic neurons (10 of 57 neurons) responded with either one action potential at the onset of a threshold (0.2 ± 0.1 nA) stimulus (6 of 10 neurons; not shown) or several action potentials of equal size throughout the duration of the threshold stimulus. Tonic neurons responded to suprathreshold (0.4-2 nA) current steps with repetitive action potentials for 200 ms or longer during the depolarization (Figure 2B-2D). Increasing stimulus amplitude (current) was associated with an increase in number and frequency of action potentials (Figure 2E): the mean minimum frequency of repetitive action potentials at threshold stimuli was 10 ± 1.5 Hz (range 6-20 Hz) at 0.2 nA and maximum frequency was 28 ± 5 Hz (range 12-56 Hz) at 2.0 nA (n = 10; Figure 2E). When two or more neurons were recorded in a single ganglion, both tonic and phasic neurons were observed (five of seven ganglia). Throughout the length of any recording (up to 2 h), neither phasic nor tonic neurons altered their accommodation properties (i.e., tonic-to-phasic or phasic-to-tonic pattern). The duration of single action potentials (measured at the level of the resting potential; between arrowheads in Figure 3A) was 21% longer in phasic when compared with tonic neurons (compare Figure 3A with 3B; p < 0.05; Table 1). Threshold for eliciting an action potential was 2.2 times lower in tonic than phasic neurons (Table 1); the mean current (using 150 ms steps) for evoking an action potential in phasic neurons was 0.6 ± 0.2 nA (range 0.4-1.0 nA) and tonic neurons was 0.2 ± 0.1 (range 0.1-0.6 nA; p < 0.05). Single action potential AHP duration differed between tonic (Figure 3A) and most phasic neurons (Figure 3B; Table 1); 10 phasic neurons had AHP durations in the same range as tonic neurons (not shown; note range in Table 1). The duration of the spike AHP following multiple action potentials in tonic neurons was nearly half the duration of most phasic neurons (Table 1); examples of AHP responses to four consecutive action potentials were evoked by 2 ms, 3 nA steps (50 Hz) for a tonic (Figure 3C) and a phasic neuron (Figure 3D).

View larger version (32K): [in this window] [in a new window]   Figure 2.   Two types of action potential patterns by neurons in human bronchial ganglia. (A) A trace showing the action potential response by a phasic-type neuron to a prolonged (500 ms) suprathreshold (2.0 nA) depolarizing stimulus: only one action potential is elicited before complete accommodation to the current (i) stimulus (2.0 nA, 500 ms, represented as the bottom-most trace). (B-D) Responses by three different tonic neurons that generate repetitive action potentials at low (B), average (C ), and high (D) frequencies throughout the same stimulus as in A (2.0 nA, 500 ms, current represented as i, below the traces), that is, no accommodation to the stimulus. Calibration bars apply to all traces. In E, stimulus-dependent increases in action potential frequency (action potentials s1) by nine tonic neurons evoked by increasing current steps (500 ms, 0.5-2.0 nA).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Action potential characteristics of bronchial ganglia tonic and phasic neurons. (A) The action potential of a tonic neuron has a relatively short duration (measured between arrowheads) compared with the durations in phasic neurons (B) (data summarized in Table 1). Also note the longer action potential afterhyperpolarization (AHP, at arrow) by the phasic neuron (B). Scale bar in (B) is for traces A and B. The AHP response to consecutive action potentials (four stimuli, 2 ms, 3 nA steps, 50 Hz) is also shorter in tonic neurons (C ) than in phasic neurons (D, arrow). The calibration bars in (D) apply to traces C and D.

Several passive membrane properties differed between tonic and phasic neurons and these are summarized in Table 1. The mean resting membrane potentials for tonic and phasic neurons used in the study were not different but the mean resting input resistance (point method) for tonic neurons was 18% greater than that for phasic neurons (p < 0.05). The time constant was 2.3 times longer in tonic than phasic neurons. The current-voltage (I-V) relationship was linear for 28 of 31 phasic neurons tested (example shown in Figure 4A); the remaining showed rectification at potentials more negative than 80 mV (not shown). Tonic neurons had a steeper I-V curve and most (five of seven) displayed voltage-dependent rectification at potentials more negative than 70 mV (Figure 4B). The input resistance calculated from the slope of the linear portion (20 mV negative from the resting potential) of the current-voltage relationship for tonic neurons was 36 ± 4 M (n = 6) and 27 ± 2 M (n = 12) for phasic neurons (p < 0.05). Anodal break excitation was evoked from potentials less than 80 mV in both types of neurons (e.g., Figure 4A). Outward rectification was observed in tonic and phasic neurons at current steps ( 0.3 nA) following initiation of action potentials. The total input capacitance for tonic neurons was nearly twice that for phasic neurons (Table 1).

View larger version (21K): [in this window] [in a new window]   Figure 4.   Current-voltage relationship in tonic and phasic neurons. (A) Current-clamp recording of a phasic neuron shows changes in membrane potential during depolarizing and hyperpolarizing incrementing current steps (± 200 pA from 1.4 nA to +0.4 nA; resting potential clamped at 50 mV). (B) The response to the same stimulus as in (A) by a tonic neuron (resting potential, 50 mV). Calibration in (B) applies to traces in (A) and (B). (C ) Steady-state current-voltage relationship measured at the end of the voltage step of the phasic neuron as in (A) (open circles) and tonic neuron in (B) (filled circles). Note the rectification by the tonic neuron (diversion from dashed line) at more hyperpolarized potentials.

Synaptic Membrane Properties

fEPSPs were elicited in 10 neurons in ganglia near isolated peribronchial nerves by electrical stimulation (1-80 V, 0.1-1 ms, 1 Hz) of that nerve with a suction electrode. In seven phasic neurons, a single stimulus of the preganglionic nerve elicited a single population of fEPSPs that fluctuated in amplitude to identical stimulations (Figure 5A). The preganglionic conduction velocity was 1.1 ± 0.3 m/s; one phasic neuron had two distinct populations of fEPSPs (not shown). Three tonic neurons had two, three, or five temporally distinct populations of fEPSPs (three populations shown in Figure 5B). The conduction velocities of the preganglionic nerves of these multiple-innervated tonic neurons ranged between 0.4 m/s and 6.1 m/s. In five phasic neurons, the single population of fEPSPs did not reach action potential threshold (Figure 5A) despite a wide range of electrical stimulation strengths, durations, and frequencies (1-80 V, 0.1- 1.5 ms, 0.1-10 Hz) applied to the preganglionic nerve; in two other phasic neurons, fEPSPs reached threshold. The average amplitude of subthreshold fEPSPs was 9 ± 1.2 mV (range 4-13 mV) and duration was 11 ± 4 ms (range 5-16 ms). When the frequency of stimulation increased, the subthreshold fEPSPs decreased in amplitude; in one cell fEPSPs reached action potential threshold at 1 Hz (Figure 5C), whereas at higher frequency (5-10 Hz) stimulation caused fEPSPs to be subthreshold (10 Hz in Figure 5D). The amplitude of fEPSPs at 1.0 Hz decreased 33% at 10 Hz (Figure 5C). The mean threshold for action potential generation was 34.2 ± 0.4 mV depolarized from an average resting potential of 48 ± 0.1 mV (n = 4; 41.5 mV in Figure 5C). In phasic cells in which fEPSPs elicited a regenerative action potential, only one action potential was observed following a single stimulus (Figure 5C); however, in two tonic neurons a single stimulus elicited up to three distinct action potentials (not shown).

View larger version (21K): [in this window] [in a new window]   Figure 5.   Fast excitatory post-synaptic potentials (fEPSPs) in human bronchial ganglia. (A) An overlay of traces showing the responses by a human bronchial ganglia neuron to 10 consecutive peribronchial nerve stimulations (shock artifact at vertical arrow; 1.0 ms, 20 V, 0.5 Hz). Fast EPSPs are subthreshold for action potential generation and graded in amplitude. (B) A single stimulus (shock artifact at arrow) to the preganglionic nerve trunk elicits three temporally distinct fEPSPs (arrowheads) indicating convergence of preganglionic axons. Dashed line is resting membrane potential of 51 mV. (C ) Shows overlaid traces of fEPSPs stimulated at 1 Hz that are at threshold for generation of action potentials. (D) Higher frequency (10 Hz) stimulation of the same cell generating action potentials as in (C ) reduced the amplitude of the fEPSPs to below threshold. (E) Fast EPSP amplitude decreased with increasing stimulation frequencies (0.1-10 Hz; *p < 0.05 when compared with 1 Hz; n = 4).

Evoked fEPSPs in bronchial ganglia were reversibly blocked by bath application of the nicotinic receptor antagonist, hexamethonium (50-100 µM, n = 4), and also by calcium-free Krebs solution (n = 3, not shown). Superfusion of the ganglion with the nicotinic agonist 1,1-dimethyl-4-phenylpiperazinium (DMPP) produced a concentration-dependent (0.5-50 µM) depolarization (1-32 mV) of the resting membrane potential (n = 6; not shown). The DMPP-induced membrane depolarization was reduced by pretreatment with hexamethonium (100 µM, n = 5).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study represents the first electrophysiological evaluation of parasympathetic ganglia neurons from lower airways of healthy human organ donors. Information on mammalian autonomic ganglia neuronal membrane properties is largely derived from laboratory animals and assumptions are generally made that such properties are similar for neurons in human ganglia. Based on the present study, properties such as the resting membrane potential, input resistance, and time constant of human bronchial neurons were within the range of those recorded in vitro for guinea pig airway (14, 16) and ferret tracheal parasympathetic neurons (13, 24) and in vivo for cat tracheal ganglion neurons (10). Unlike neurons in human gallbladder (8) and lumbar (25) autonomic ganglia, but similar to airway ganglia neurons in laboratory animals (13, 14, 16, 24), neurons located in human bronchial ganglia have anatomic, synaptic, and membrane properties that are indicative of an integrative function (reviewed in [1]). That membrane properties of airway parasympathetic ganglia neurons from human and laboratory animals were similar indicates that these cells have a similar function in the mammalian lower airways.

Stimulation of peribronchial nerves near the ganglia evoked fEPSPs that had amplitudes similar to those reported for bronchial ganglia neurons in guinea pigs (14) but had shorter durations than those described in tracheal ganglia of laboratory animals (10, 13, 16). Consistent with other lower airway parasympathetic ganglia neurons (13, 14), fEPSPs recorded in human bronchial neurons were pharmacologically distinguished as nicotinic cholinergic. Half of the neurons had fEPSPs that did not reach action potential threshold despite simultaneous stimulation of preganglionic nerves with a wide range of stimulus pulse durations, amplitudes, and frequencies. Increasing the frequency of preganglionic stimuli decreased fEPSP amplitude in human bronchial ganglia neurons (Figure 5D and 5E) similar to the frequency-dependent depression of vagus nerve stimulated cholinergic tone in the guinea pig bronchus (26) and depression of fEPSP amplitude observed in human myenteric neurons (9). Frequencies of preganglionic input to cat tracheal neurons in vivo (10) were over double the frequency causing a depression in the present study. Although depression of fEPSP amplitude could be due to depletion of neurotransmitter, it is not likely due to supply of choline as this was added to the perfusion buffer (21). These neurons also have multiple, branching dendrites (Figure 1C), similar to neurons in guinea pig bronchial (19) and ferret tracheal (13) ganglia. Complex dendritic arbors, as observed for bronchial neurons, may modulate synaptic strength by interposing variable electrotonic distances between each synapse and the site of action potential generation (27), consequently affecting fEPSP amplitude.

Single stimulation of peribronchial nerves evoked multiple, temporally distinct fEPSPs (Figure 5B); such a response indicates that converging preganglionic axons innervate human bronchial ganglion neurons. In neurons that did not display multiple fEPSPs, it may be possible that not all of the preganglionic inputs were stimulated, although ganglia were selected to avoid this possibility (see METHODS). Preganglionic axons had conduction velocities that varied over a wide range; this was evident with the observation of temporally distinct populations of fEPSPs (Figure 5B). Faster conducting axons (> 5 m/s) were similar to that observed for myelinated sympathetic preganglionic nerves (e.g., [28]) while the slower conducting axons ( 1 m/s) were similar to unmyelinated parasympathetic preganglionic axons in the lower airway ganglia of laboratory animals (13, 14, 29). Although the slowest conduction velocities may be indicative of a relay connection from another ganglion between the stimulation site and the recorded ganglia, no neighboring ganglion or ganglia were observed between the stimulating electrode and the recorded neurons. This does not rule out intraganglionic communication, as has been suggested to occur within guinea pig bronchial (19) and ferret tracheal (13) ganglia. It has been speculated that differences in conduction velocities of parasympathetic preganglionic axons may be associated with different effector tissue (e.g., smooth muscle, glands, vasculature) or differences in ganglion neuron neurotransmitter phenotype (1). Different preganglionic input to airway parasympathetic ganglia have been reported for guinea pig lower airways where the parasympathetic neurons evoking tracheal relaxation are activated by both cholinergic and neurokinin-containing preganglionic fibers (3). By contrast, cholinergic ganglion neurons reported to regulate smooth muscle contractions are not overtly activated by neurokinins (30) released from surrounding sensory nerves (31).

Human bronchial ganglia neurons could be classified into two types based on response to prolonged depolarizing current steps. Most neurons displayed one action potential and accommodated in response to suprathreshold stimuli while the remaining neurons responded with repetitive action potentials without accommodation; these neurons are referred to as phasic or tonic, respectively (28). Tonic and phasic neurons have also been reported for neurons in human myenteric plexus (9) but only phasic neurons were observed in human gallbladder ganglia (8). Unlike tonic neurons in guinea pig bronchial ganglia (22), human bronchial tonic neurons displayed a greater range of action potential frequency at suprathreshold current steps (Figure 2E); furthermore, the ratio of phasic-to-tonic cells is much greater in human than guinea pig bronchial ganglia (14). Accommodative properties of phasic type neurons may greatly affect the ability of the neurons to relay preganglionic stimuli (14) and, consequently, affect parasympathetic tone in the airway (1, 12, 26).

Tonic and phasic neurons in sympathetic (28), parasympathetic (22), and myenteric ganglia (9) are reported to differ in their active and passive membrane properties. In human bronchial ganglia, tonic neurons had a greater input resistance, longer time constant, and steeper current-voltage relationship when compared with phasic neurons. Differences in the mean values of time constant may be due to the differences in specific membrane capacitance and the shape and size of the neurons. Although the higher input resistance and tonic firing pattern could represent artifacts from recording from healthier cells (or a better impaled cells), the fact that other membrane properties were similar (resting membrane potential, action potential amplitude, and AHP amplitude) indicates that these two distinct firing patterns may represent normal characteristics of two populations of neurons within these ganglia. It is unlikely that the difference between tonic and phasic neurons was due to the health of the tissue or donor as both tonic and phasic responses were observed in different neurons within the same ganglion from different donors. Whether tonic and phasic neurons innervate distinct target tissue and are anatomically distinct, as are tonic and phasic neurons in sympathetic ganglia (32), remains to be determined. It would be difficult to conclusively state whether neuronal geometry underlies the differences between these two cell types as only phasic neurons that were successfully filled had multiple dendrites.

Single action potential properties also differed between tonic and phasic neurons. Spike width was consistently broader in phasic than tonic cells (Figure 3), and relatively broader for both cell types when compared with human gallbladder (8) or myenteric (9) neurons, or with airway ganglia neurons from other species, for example, two times that reported for guinea pig bronchial neurons (14). Although the AHP amplitude that followed single action potentials was approximately the same in tonic and phasic neurons, the duration of the AHP was nearly twice as long in most phasic than tonic neurons. The AHP duration following four action potentials was also longer in phasic than tonic neurons. The broader spike duration and longer AHP in phasic neurons may be indicative of a calcium-activated potassium current similar to that observed in guinea pig bronchial ganglia neurons (22); much longer durations ( 1000 ms AHP_slow) associated with guinea pig sympathetic (33) or airway sensory (34) neurons were not observed. Nonetheless, the longer AHP in phasic neurons may be responsible for the accommodative properties in most of these neurons (33) and could limit the firing frequency and, consequently, the output from these ganglia.

Preganglionic and synaptic membrane properties of phasic and tonic cells also differed. Tonic neurons had a higher level of converging preganglionic input than did phasic neurons; most phasic neurons had only one population of synaptic potentials (fEPSPs; Figure 5A), indicative of one preganglionic input, whereas all tonic neurons had more converging preganglionic inputs with as many as five converging on one cell. Threshold for action potential generation, reached either by cathodal current injection or by fEPSPs, was lower in tonic than phasic neurons. In several phasic neurons, fEPSPs did not reach action potential threshold; it is possible that one or more presynaptic pathways to this ganglion was not stimulated, although every attempt was made to select ganglia on one peribronchial nerve trunk (see METHODS). Based on the findings in human bronchial ganglia it is more likely that phasic ganglia neurons attenuate preganglionic stimuli as has been reported for cat tracheal (10) and guinea pig bronchial (14) ganglia neurons. By contrast, tonic neurons, due to increased synaptic input, lower action potential threshold, and their inability to accommodate, may amplify preganglionic input.

Neurons in human bronchial parasympathetic ganglia display diverse and complex electrophysiological and anatomical properties suggesting that these neurons may integrate information received from the central nervous system. This is not unlike other mammalian autonomic ganglia neurons, most of which integrate input from the central nervous system (1). In human bronchial ganglia, preganglionic stimuli evoked either subthreshold fEPSPs, resulting in an input-to-output ratio less than one, or multiple, threshold fEPSPs that could amplify preganglionic input. The existence of tonic and phasic firing patterns, multiple fEPSPs, relatively long, additive action potential AHPs, in addition to anatomical features such as a complex dendritic arbor, may contribute to the integration of signals from the central nervous system. Changes in the integrative properties, as has been reported for other airway parasympathetic ganglia neurons (22, 35), could greatly affect the output from these ganglia, and, consequently, airway caliber (12).