• The stimulation of pulmonary rapidly adapting receptors (RARs) by wood smoke was investigated. Impulses from seventy RARs were recorded in fifty-nine anaesthetized, open-chest and artificially ventilated rats; responses to delivery of 6 ml of wood smoke into the lungs were studied in sixty-one receptors whereas responses to histamine (10 or 100 μg kg⁻¹, i.v.) were studied in the other nine.
• Delivery of wood smoke stimulated fifty-two of the sixty-one RARs studied. When stimulated, an intense burst of discharge was evoked within 1 or 2 s of smoke delivery. This increased activity quickly peaked in 1-3 s (Δ= 15.8 ± 1.6 impulses s⁻¹; n = 61; mean ± s.e.m.), then declined and yet remained at a level higher than the baseline activity. The mean duration of the stimulation was 25.1 ± 2.7 s. In contrast, smoke delivery did not affect tracheal pressure.
• Peak responses of RARs to wood smoke were partially reduced by removal of smoke particulates and were largely attenuated by pretreatment with dimethylthiourea (DMTU, a hydroxyl radical scavenger), indomethacin (Indo, a cyclo-oxygenase inhibitor), or both DMTU and Indo (DMTU + Indo). Conversely, the peak responses of RARs were not significantly affected by pretreatment with isoprenaline (a bronchodilator) or vehicle for these chemicals. Additionally, pretreatment with DMTU, Indo, or DMTU + Indo did not significantly alter the RAR sensitivity to mechanical stimulation (constant-pressure lung inflation; 20 cmH₂O).
• Of the nine RARs tested, six were stimulated by histamine and their sensitivity to this chemical irritant was not altered by pretreatment with DMTU + Indo.
• The results suggest that both the particulates and gas phases are responsible for, and both the hydroxyl radical and cyclo-oxygenase products are involved in, the stimulation of RARs by wood smoke. Furthermore, changes in lung mechanics following smoke delivery are not the cause of this afferent stimulation.

We recently reported that inhalation of wood smoke via tracheostomy immediately affects breathing in anaesthetized rats (Kou & Lai, 1994; Kou, Wang & Lai, 1995; Kou, Lai, Hsu & Lin, 1997). An augmented breath is one of the two immediate ventilatory responses occurring within the first two breaths of smoke inhalation. The augmented breath is preserved when the conduction of unmyelinated C fibres is selectively blocked by perineural capsaicin treatment of cervical vagus nerves, but is totally abolished when the conduction of myelinated fibres is differentially blocked by the cooling of both vagi to 6.7°C (Kou et al. 1995). Additionally, removal of particulates from wood smoke prevents the augmented breath in 42% of the rats studied (Kou et al. 1995), whereas pretreatment with antioxidants for the hydroxyl radical (OH·) prevents the response in 80% of the animals tested (Kou et al. 1997). OH· is an extremely reactive oxygen metabolite formed in the lung tissues following inhalation of wood smoke (Pryor, 1992). The augmented breath has been postulated to be a reflex resulting from stimulation of pulmonary rapidly adapting receptors whose activity is conducted by myelinated vagal afferents (Glogowska, Richardson, Widdicombe & Winning, 1972; Coleridge & Coleridge, 1986). It has been suggested that rapidly adapting receptors can be activated directly by chemical mediators and/or indirectly through mediator-induced bronchoconstriction (Sant'Ambrogio, 1982; Coleridge & Coleridge, 1986; Ravi, Teo & Kappagoda, 1989; Chen, Lee & Kou, 1997a). Taken together, the results obtained from our reflex studies (Kou et al. 1995, 1997) suggest that rapidly adapting receptors are stimulated by both the particulate and gas phases of wood smoke and that OH· is primarily involved in this sensory stimulation. However, direct evidence provided by electrophysiological studies remains to be established.

In addition to oxygen reactive metabolites, inhalation of toxic smoke is known to cause an increase in the release of other chemical mediators including cyclo-oxygenase products (Shinozawa, Hales, Jung & Burke, 1986; Kimura, Traber, Herndon, Niehaus, Flynn & Traber, 1988; Hales, Musto, Hutchison & Mahoney, 1995). When the cyclo-oxygenase pathway is activated, arachidonic acid is metabolized to various types of prostaglandins and thromboxane (Bakhle & Ferreira, 1986). Several cyclo-oxygenase products, when administered exogenously, have been shown to stimulate rapidly adapting receptors (Bergren, Gustafson & Myers, 1984; Coleridge & Coleridge, 1986; Mohammed, Higenbottam & Adcock, 1993; Matsumoto, Takano, Nakahata & Shimizu, 1994). However, whether these arachidonate metabolites are involved in the activation of rapidly adapting receptors following smoke inhalation is not yet known.

In this study, we recorded afferent activity arising from rapidly adapting receptors in anaesthetized rats to determine (1) whether these pulmonary receptors are stimulated by delivery of wood smoke into the lungs, (2) whether both the gas and smoke particulate phases of wood smoke are responsible for this afferent stimulation, (3) whether OH· and cyclo-oxygenase products are involved in this afferent stimulation and (4) whether changes in lung mechanics following smoke delivery are involved in this afferent stimulation. To accomplish these objectives, we compared the receptor responses to wood smoke before and after removal of smoke particulates, and after pretreatment with dimethylthiourea (DMTU; a OH·scavenger), indomethacin (a cyclo-oxygenase inhibitor), a combination of DMTU and indomethacin (DMTU + indomethacin), or isoprenaline (a bronchodilator).

Sprague-Dawley rats (weight, 312 ± 6 g) of either sex were anaesthetized with an intraperitoneal injection of chloralose (100 mg kg⁻¹; Sigma) and urethane (500 mg kg⁻¹; Sigma). A polyethylene catheter was inserted into the jugular vein and advanced until the tip was close to the right atrium for intravenous administration of pharmacological agents. The right femoral artery was cannulated for measuring arterial blood pressure. During the course of the experiments, supplemental doses of chloralose (20 mg kg⁻¹ h⁻¹) and urethane (100 mg kg⁻¹ h⁻¹) were administered to maintain the abolition of the corneal reflex and pain reflexes induced by pinching the animal's tail. During the recording of vagal action potentials, the rats were paralysed with pancuronium bromide (0.05 mg kg⁻¹, i.v.; Orgnon Teknika B. V., Boxtel, The Netherlands). Periodically, the effect of pancuronium was allowed to wear off so that the depth of anaesthesia could be checked.

The rats were tethered in a supine position and the trachea was cannulated below the larynx with a short tracheal tube via a tracheotomy. A mid-line thoracotomy was performed and the edges of the rib cage were retracted. The lungs were ventilated by a rodent respirator (Harvard 683, South Natick, MA, USA) at a constant tidal volume of 2 ml. The frequency of the respirator was set at 65-75 breaths min⁻¹ and was kept constant in each experiment. The expiratory outlet of the respirator was placed under 3-4 cm of water to maintain a nearly normal functional residual capacity. Tracheal pressure (P_tr) was monitored by a pressure transducer (Validyne MP45-28, Northridge, CA, USA) via a side tap of the tracheal cannula. Body temperature was maintained at ~36°C throughout the experiment by a servo-controlled heating blanket.

Lung inflation was used as the first step to search for rapidly adapting receptors. The lungs were hyperinflated in a step-like manner to 4 times the tidal volume (4 ×V_T) or by constant pressure inflation (~20 cmH₂O; Fig. 1A), which was maintained for 15-25 s. The response of these receptors to lung deflation was also studied by exposing the expiratory outlet of the respirator to atmospheric pressure for a period of 10-20 s (Fig. 1A). The conduction velocity of the afferent fibre arising from the individual receptor was measured in the majority (48 of 70) of the receptors studied by the method described by Bergren & Peterson (1993). Two criteria were used to identify the rapidly adapting receptors in this study: (1) an adaptation index to maintained lung inflation was > 70% (Widdicombe, 1954), and (2) conduction velocity, whenever measured, was within the range of myelinated fibres. Prior to the end of each experiment, the general locations of the receptors studied were identified within the lung structure by gently probing the tissues with a polyethylene rod (diameter, 2 mm).

Figure 1 Afferent responses of a rapidly adapting receptor (large spikes) to lung deflation and inflation (A), unfiltered wood smoke (B) and gas-phase smoke (C)

The mean baseline activity of the rapidly adapting receptors studied was 0.4 ± 0.1 impulses s⁻¹ (n = 70). Only eleven receptors exhibited a baseline activity in phase with ventilatory cycles; the others had irregular or no baseline activity. The evoked discharge of each receptor in response to maintained lung inflation adapted rapidly (Fig. 1A) and the mean adaptation index reached 88.2 ± 1.6% (n = 70). A majority (54 of 70) of these receptors was also activated by lung deflation (Fig. 1A). The mean conduction velocity of the afferent fibres conducting impulses from forty-eight of these receptors was 13.4 ± 0.5 m s⁻¹ (range, 6.8-21.0 m s⁻¹); the conduction velocity of the remaining twenty-two was not measured. All receptors were localized within the lung structure and their physiological properties were consistent with those reported in rats (Bergren & Peterson, 1993) and in other species (Coleridge & Coleridge, 1986).

Of the sixty-one rapidly adapting receptors tested, fifty-two were stimulated by delivery of unfiltered wood smoke. When stimulated, an intense burst of discharge was evoked within 1 or 2 s of smoke delivery and quickly peaked in 1-3 s (Figs 1B and 2). After reaching its peak, the evoked discharge declined, yet remained at a level higher than the baseline activity (Figs 1B and 2). In general, the smoke-evoked discharge was not modulated by ventilatory cycles (Fig. 1B). For the whole group of sixty-one receptors, the mean duration of the afferent stimulation was 25.1 ± 2.7 s (range, 0-68 s).

Figure 2 Mean responses of rapidly adapting receptors to unfiltered wood smoke

In ten receptors that were stimulated by unfiltered wood smoke, afferent responses were compared with those evoked by the gas phase of wood smoke. Gas-phase smoke evoked a milder stimulation in each of the receptors tested compared with unfiltered smoke (Fig. 1). Statistical analysis revealed that the peak receptor response to gas-phase smoke was significantly smaller than that to unfiltered smoke (Fig. 3A). However, the mean duration of the afferent stimulation evoked by gas-phase smoke was not significantly different from that evoked by unfiltered smoke (Fig. 4A).

Figure 3 Effects of various experimental interventions on average peak responses of rapidly adapting receptor to unfiltered wood smoke

Figure 4 Effects of various experimental interventions on mean duration of stimulation of rapidly adapting receptors produced by unfiltered wood smoke

In thirty-nine receptors initially stimulated by unfiltered smoke, smoke challenges were repeated after the animals had been pretreated with vehicle (n = 8), DMTU (n = 9), indomethacin (n = 8), DMTU + indomethacin (n = 8), or isoprenaline (n = 6). Twenty minutes after pretreatment with vehicle, DMTU, indomethacin, or DMTU + indomethacin and 10 min after pretreatment with isoprenaline, the baseline activity of these receptors did not change significantly (Fig. 3). In the vehicle- or the isoprenaline-treated group (Fig. 5A), a repeated smoke challenge evoked afferent responses of similar amplitude and time course in the same receptors compared with their control responses. In contrast, in the DMTU- (Fig. 5B), indomethacin- (Fig. 6A) or DMTU + indomethacin-treated groups (Fig. 6B), a repeated smoke challenge evoked a milder afferent stimulation in each of the receptors tested. Statistical analysis revealed that the peak receptor response evoked by unfiltered smoke was not significantly altered by pretreatment with vehicle or isoprenaline, but was largely attenuated by pretreatment with DMTU, indomethacin or DMTU + indomethacin (Fig. 3). Additionally, the mean duration of the afferent stimulation was not altered by pretreatment with vehicle (Fig. 4B), isoprenaline (Fig. 4C), DMTU (Fig. 4D), or indomethacin (Fig. 4E), but was markedly shortened by pretreatment with DMTU + indomethacin (Fig. 4F). At the end of the test period, all the receptors in the vehicle-, isoprenaline-, DMTU-, indomethacin- or DMTU + indomethacin-treated group could still respond to lung inflation and their mean responses were not significantly affected by pretreatment with vehicle or these chemicals (Table 1).

Figure 5 Afferent responses of two rapidly adapting receptors to unfiltered wood smoke before and after pretreatment with isoprenaline (A) or dimethylthiourea (B)

Figure 6 Afferent responses of two rapidly adapting receptors (large spikes) to unfiltered wood smoke before and after pretreatment with indomethacin (A) or both dimethylthiourea and indomethacin (B)

Table 1 Average peak responses of rapidly adapting receptors to constant pressure lung inflation before and after various pharmacological pretreatments

To investigate the possible non-specific effects of DMTU and indomethacin, we studied the influences of DMTU + indomethacin on the receptor responses to intravenous injection of histamine, a chemical irritant that stimulates rapidly adapting receptors (Coleridge & Coleridge, 1986). Of the nine receptors studied, six were immediately (< 3 s) stimulated by histamine injection (Fig. 7A). After animals had been pretreated with DMTU + indomethacin, challenges of the same dose of histamine produced similar afferent responses in each of these six receptors and the mean response was not significantly different from the control (Fig. 7B).

Figure 7 Afferent responses of rapidly adapting receptors to histamine

Under controlled conditions, delivery of unfiltered wood smoke did not cause any detectable change in P_tr (Figs 5 and 6). The baseline P_tr and its peak response post smoke delivery were 8.0 ± 0.1 and 8.1 ± 0.1 cmH₂O, respectively (P > 0.05; n = 61). In contrast, mean arterial blood pressure initially increased from a baseline of 82.8 ± 2.0 mmHg to a peak of 89.5 ± 2.8 mmHg (P < 0.01; n = 61) at 4-10 s and subsequently decreased to its lowest level of 67.7 ± 2.6 mmHg (P < 0.01; n = 61) at 15-36 s following delivery of unfiltered smoke (Figs 5 and 6). During the hypertensive and ensuing hypotensive period, mean heart rate decreased from a baseline of 322.7 ± 5.3 beats min⁻¹ to 301.1 ± 5.7 and 297.8 ± 5.5 beats min⁻¹ (P < 0.01; n = 61), respectively. Overall, it took less than 4 min for these cardiovascular parameters to return to their normal baseline.

Results of this study demonstrate that pulmonary rapidly adapting receptors were promptly stimulated when three tidal breaths of wood smoke were delivered into the lower airways and lungs. This afferent stimulation seems to be linked to both the particulate and gas phase of wood smoke, since the afferent responses were partially reduced by removal of particulates from the smoke. Additionally, the afferent responses were largely attenuated by pretreatment with DMTU, indomethacin, or DMTU + indomethacin. Although the mean duration of the stimulation was not significantly affected by pretreatment with either DMTU or indomethacin, it was markedly shortened by a prior administration of DMTU + indomethacin. In contrast to these effects, pretreatment with vehicle did not affect the overall afferent responses to smoke delivery. DMTU is an effective scavenger for OH·, whereas indomethacin is a cyclo-oxygenase inhibitor that suppresses the production of prostaglandins and thromboxane. Hence, our results suggest that both OH· and cyclo-oxygenase products are actively involved in the stimulation of rapidly adapting receptors evoked by wood smoke.

That OH· and cyclo-oxygenase products are involved in the stimulation of rapidly adapting receptors is not surprising. These two metabolites have been shown to be important factors for the activation of other visceral sensory receptors under various pathophysiological conditions (Longhurst, Rotto, Kaufman & Stahl, 1991; Stahl, Pan & Longhurst, 1993; Ustinova & Schultz, 1994; Chen, Lee & Kou, 1997b). Nevertheless, the mechanisms by which these two metabolites are associated with the stimulation of rapidly adapting receptors by wood smoke remain unclear. One possibility is that these two metabolites may act directly on rapidly adapting receptors or induce further releases of other chemical mediators to stimulate these pulmonary receptors. A direct effect of prostaglandins on rapidly adapting receptors has been suggested (Bergren et al. 1984; Coleridge & Coleridge, 1986; Mohammed et al. 1993) and an excitatory effect of OH·on the nervous tissue has been demonstrated (Carratù, Masini, Mannaioni & Mitolo-Chieppa, 1990). Furthermore, oxygen-reactive metabolites are known to be involved in the release of other chemical mediators such as histamine (Mannaioni et al. 1988), which stimulates these pulmonary receptors (Vidruk, Hahn, Nadel & Sampson, 1977; Bergren et al. 1984; Ravi et al. 1989; Mohammed et al. 1993). A second possibility is that the function of OH· and cyclo-oxygenase products is to maintain and/or to potentiate the sensitivity of rapidly adapting receptors to smoke-related stimuli. Consequently, lowering the levels of these two metabolites by DMTU, indomethacin or DMTU + indomethacin may make rapidly adapting receptors less sensitive to a smoke challenge. However, pretreatment with DMTU + indomethacin does not seem to affect the receptor sensitivity to either mechanical stimulation or other chemical irritant. As shown in this study, the afferent responses to lung inflation or to histamine challenge were not altered by a prior administration of DMTU + indomethacin. These results also suggest that the attenuation of the smoke-induced afferent stimulation observed after administration of these chemicals was not likely to be due to the anaesthetic or deleterious effects on rapidly adapting receptors. A third possibility is that OH· and cyclo-oxygenase products may indirectly stimulate rapidly adapting receptors through their bronchoconstrictive effects (Bakhle & Ferreira, 1986; Katsumata et al. 1990). However, delivery of wood smoke did not cause any detectable change in P_tr (i.e. transpulmonary pressure in open-chest preparation) in this study. Furthermore, the receptor responses to wood smoke were not affected by a prior administration of a bronchodilator. The inability of the bronchodilator to modify the receptor responses is not because drug effects had worn off, since the hypotension induced by isoprenaline still persisted when wood smoke was challenged (Fig. 5A). These observations strongly suggest that changes in lung mechanics may not be the cause of the stimulation of rapidly adapting receptors by wood smoke.

The source of the origin of OH· and cyclo-oxygenase products cannot be determined in this study. Several investigators (Traber & Herndon, 1986; Pryor, 1992; Youn, Lalonde & Demling, 1992) have suggested that oxygen-reactive metabolites formed in the lung tissues following inhalation of toxic smoke might originate from an exogenous source. Both the particulate and the gas phases of wood smoke are known to contain high concentrations of free radicals and radical precursors, which may continuously generate OH· either in the smoke or when they reach the lung tissues following smoke inhalation (Pryor, 1992). On the other hand, OH· and cyclo-oxygenase products may be formed and released endogenously in the lungs following delivery of wood smoke. Certain lung cells, such as polymorphonuclear leucocytes and alveolar macrophages, are primary oxygen radical releasers (Hoidal, Beall & Repine, 1979; Traber & Herndon, 1986) and have been found to be activated following acute inhalation of toxic smoke (Traber & Herndon, 1986; Riyami et al. 1990). Additionally, it is known that the lungs are a rich source of arachidonate products and the enzymes necessary for their metabolism (Bakhle & Ferreira, 1986). Indeed, several studies have demonstrated activation of the cyclo-oxygenase pathway following acute inhalation of toxic smoke (Shinozawa et al. 1986; Kimura et al. 1988; Hales et al. 1995). Previous studies indicated that significant amounts of oxygen-reactive metabolites are formed during the metabolism of arachidonic acid via cyclo-oxygenase (Egan, Paxton & Kuehl, 1976) and that oxygen-reactive metabolites can increase the production of cyclo-oxygenase products in the lungs (Sanderud, Bjoro & Saugstad, 1993). Therefore, these two metabolites may be interrelated in their biosynthesis. On the other hand, it has been postulated that OH· must act synergistically with cyclo-oxygenase products in order to manifest the functional contribution of either metabolite to visceral C fibre responses to abdominal ischaemia and reperfusion (Stahl et al. 1993) or to vagal pulmonary C fibre responses to pulmonary air embolism (Chen et al. 1997b). Accordingly, these two metabolites may be interrelated in their physiological effects.

In this study, pretreatment with DMTU + indomethacin did not completely abolish the stimulation of rapidly adapting receptors by wood smoke, suggesting that factors other than OH· and cyclo-oxygenase products may be involved in this afferent stimulation. These factors may include smoke components and/or chemical mediators that are not related to the generation of these two metabolites. Alternatively, the doses of DMTU and indomethacin may not have been sufficient to eliminate completely OH· and cyclo-oxygenase products formed after smoke delivery.

Rapidly adapting receptors are also known as ‘irritant’ receptors (Sellick & Widdicombe, 1971). The former and the latter terms describe their afferent responses to mechanical stimuli and chemical irritants, respectively (Sant'Ambrogio, 1982). However, the functional roles of these pulmonary receptors and mechanisms of their activation in response to chemical irritants have been equivocal (Sant'Ambrogio, 1982; Coleridge & Coleridge, 1986; Ravi et al. 1989). The fact that not all the rapidly adapting receptors studied responded to wood smoke may suggest the heterogeneous nature of these pulmonary receptors. While abundant evidence indicates that these receptors normally serve as mechanoreceptors in the airways and lungs (Sant'Ambrogio, 1982; Coleridge & Coleridge, 1986; Pisarri, Jonzon, Coleridge & Coleridge, 1990; Chen et al. 1997a), some studies suggest that they may also function as chemical-sensitive receptors (Vidruk et al. 1977; Bergren et al. 1984; Kou & Lee, 1990, 1991; Mohammed et al. 1993). For example, rapidly adapting receptors may be stimulated chemically by histamine and/or mechanically through histamine-induced bronchoconstriction (Vidruk et al. 1977; Bergren et al. 1984; Ravi et al. 1989; Mohammed et al. 1993). Similar mechanisms have also been ascribed to the stimulation of these pulmonary receptors by cigarette smoke in dogs (Kou & Lee, 1991). Although wood smoke and cigarette smoke are combustion products that have both gas and particulate phases, the constituents and their concentrations in these two types of smoke are vastly different. This leads to the possibility that the stimulation of rapidly adapting receptors by these two types of smoke is mediated through different chemical mechanisms. Indeed, Kou & Lee (1991) demonstrate that nicotine is the major smoke constituent responsible for the stimulation of rapidly adapting receptors by cigarette smoke. In a recent study in rats, Bergren & Peterson (1993) reported that only eight rapidly adapting receptors were found in nineteen animals and that these pulmonary receptors were not sensitive to histamine challenge within a dose range of 100 μg to 10 mg per rat. They concluded that rapidly adapting receptors are rare and may have little function in rats. In this study, these receptors were stimulated not only by wood smoke but also by histamine challenge at a relatively low dose level (10 or 100 μg kg⁻¹). The failure of Bergren & Peterson (1993) to observe the stimulation of rapidly adapting receptors by histamine may be due to the small sample (only 3) of receptors tested in their study. Collectively, the present findings suggest that rapidly adapting receptors in rats may function as ‘irritant’ receptors (Sellick & Widdicombe, 1971) following inhalation of wood smoke. This notion gains an additional support from the results obtained in our previous studies (Kou & Lai, 1994; Kou et al. 1995, 1997) that inhaled wood smoke effectively triggers an augmented breath, a reflex presumably resulting from stimulation of this type of pulmonary receptor (Coleridge & Coleridge, 1986).

In our previous studies of the reflex responses (Kou & Lai, 1994; Kou et al. 1995, 1997), an immediate and prominent bradycardia and hypotension were evoked within 1 or 2 s of inhalation of wood smoke. In this study, however, immediate cardiovascular responses to wood smoke were absent, probably because the right vagus nerve was sectioned. The mild hypertension and ensuing hypotension post smoke delivery observed in this study may be due to the activation of arterial chemoreceptors by the extremely low O₂ and high CO₂ concentrations in the smoke (Kou & Lai, 1994).

Inhalation of toxic smoke not only causes airway irritations but also produces lung injury (Shinozawa et al. 1986; Traber & Herndon, 1986; Kimura et al. 1988; Youn et al. 1992; Larson & Koenig, 1994; Hales et al. 1995). Although the information regarding the inhalation injury of the lungs is relatively well established (Traber & Herndon, 1986; Youn et al. 1992), the mechanisms underlying the airway irritation induced by toxic smoke have been largely overlooked. It is generally believed that rapidly adapting receptors may serve as an important sensory system in detecting the onset of pathophysiological changes in the airways (Coleridge & Coleridge, 1986). While oxygen-reactive metabolites and cyclo-oxygenase products have been strongly implicated in the pathogenesis of inhalation injury (Shinozawa et al. 1986; Traber & Herndon, 1986; Kimura et al. 1988; Youn et al. 1992; Hales et al. 1995), their involvements in eliciting the sensory irritation of the airways following smoke inhalation are obscure. The observations made in this study provide the first electrophysiological evidence that rapidly adapting receptors play an important role in detecting the airway assault by wood smoke and that their functional significance is mediated through mechanisms involving OH. and cyclo-oxygenase products.

We are grateful to Dr Lu-Yuan Lee for his valuable comments on the manuscript and to Mr Al Vendouris for his editorial assistance. This study was supported by a National Science Council of Republic of China grant 86-2314-B010-079 and by a VGHYM grant 87-S4-23.