INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MULTIPLE CENTRAL CHEMOSENSITIVE areas with potential involvement in respiratory modulation have been identified in the medulla, pons, and midbrain by using local microinjections of acetazolamide (AZ), a carbonic anhydrase inhibitor, to cause focal tissue acidification. Nattie and his colleagues (1, 5, 6; reviewed in Ref. 17) reported that, under isocapnic in vivo conditions, an augmented respiratory output was evoked by AZ injection into the rostral, caudal, and intermediate chemosensitive areas of the ventral medullary surface (VMS), the nucleus tractus solitarii (NTS), the locus coeruleus, the medullary raphe, and a rostral ventral respiratory group. Other investigators found that AZ administration into the pre-Botzinger complex region (23) or caudal hypothalamus (19) also produced an increase in respiratory output in anesthetized rats, indicating the presence of ventilatory chemosensitivity in these regions as well. This widespread distribution of chemosensitive regions was supported by in vitro studies in which neurons recorded in these areas responded to accumulation of CO₂/H⁺ in the bath solution (7, 8, 24, 26). An interesting common characteristic of these regions is that they not only have chemosensitivity but also contain respiratory-modulated neurons (4, 9, 11, 28).

The role of the rostral fastigial nucleus (FNr) within the cerebellum in respiratory chemoreflexes has been established. Respiratory output in response to inhalation of severe hypercapnia was significantly attenuated after bilateral chemical lesions of the FNr in both spontaneously breathing and paralyzed rats (30). Consistent with this report, cerebellectomy blunted the respiratory response to moderate and severe hypercapnia in cats (32) and dogs (21). Moreover, Fos protein immunochemical studies have shown a remarkable increase in FNr neuronal activity in response to severe hypercapnia in rats (33). In support of these findings, the activity of respiratory-mediated neurons within the FNr has been extracellularly recorded in alert (10) and anesthetized cats (15, 29). Recent clinical observations also indicated fastigial nucleus (FN) involvement in respiratory chemoreflexes. Congenital central hypoventilation syndrome is characterized by a weakened ventilatory response to hypercapnia that produces dysfunction of respiratory drive during sleep. Investigators, utilizing functional magnetic resonance imaging, found that neuronal activity was comparatively decreased in the FN of these patients especially during hypercapnia (12). These results lead us to hypothesize that the FNr has CO₂/H⁺ chemosensitivity and contributes to the cerebellar facilitation of the respiratory response to CO₂. The possibility of chemosensitive sites in other cerebellar deep nuclei, such as the interposed (IN) and lateral cerebellar nucleus (LCN) that are known to be involved in respiratory modulation (10, 11, 13, 31), was also examined.

To test our hypotheses, AZ was microinjected into different cerebellar deep nuclei under isocapnic conditions and cardiorespiratory responses compared before and after administration. Two major findings were obtained from the present study. First, AZ microinjected unilaterally into the FNr significantly increased respiratory output by changing the amplitude rather than frequency (f) with little effect on arterial blood pressure (ABP). Second, the same dose of AZ delivered into the IN and LCN failed to produce any detectable respiratory response. These results suggest that the FNr, compared with the IN and LCN, uniquely has ventilatory chemosensitivity that contributes to augmentation of the ventilation during severe hypercapnia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General. Experimental protocols described in this study were approved by the Institutional Animal Care and Use Committee in compliance with the Animal Welfare Act. They were in accordance with the Guide for the Care and Use of Laboratory Animals [DHHS Publication No. 85-23 (NIH), Revised 1985, Office of Science and Health Reports, Bethesda, MD 20892]. Experiments were performed in 28 Sprague-Dawley rats (250-400 g) anesthetized with chloralose (100 mg/kg ip) and urethane (500 mg/kg ip). The left femoral vein and artery were cannulated, the former for anesthetic administration and the latter for monitoring ABP. The supplemental anesthetic was administered intravenously to suppress corneal and withdrawal reflexes. The trachea was cannulated below the larynx and connected to a one-way breathing valve. Tracheal pressure (Ptr) was measured via a pressure transducer that was connected with a side port of the tracheal cannula. Core temperature was monitored with a rectal probe and maintained at 37-38°C by a heating pad and radiant heat.

1

1

Cerebellar craniotomy. Animals were placed in a rigid metal frame with the head fixed in a stereotaxic apparatus (Kopf). A hole (<8-mm diameter) was drilled at the midline with the center (12.5 mm posterior to the bregma) for stereotaxically inserting a micropipette into the different cerebellar deep nuclei. Bleeding was controlled with bone wax, absorbable hemostat (Surgicel and Gelfoam), and the use of a bipolar coagulator (model 440S, Radionics). The dura was removed, and the underlying tissue covered by cotton was saturated with mineral oil.

Recording of PN. The right cranial phrenic (C₅) nerve was freed in the neck, desheathed, and cut. Efferent activity was recorded with bipolar silver electrodes in pools of mineral oil. Nerve signals were amplified (model P15 amplifier, Grass Instrument) and filtered (band pass 20-3,000 Hz). PN were integrated (PN) with a leaky resistance-capacitance circuit (0.1-s time constant) and monitored on a storage oscilloscope (model 5103n, Tektronix).

Microinjection of AZ into the cerebellar deep nuclei. After baseline cardiorespiratory variables became stable for at least 10 min, the following studies were conducted. First, the effects of focal tissue acidification of the FNr on the respiratory output were examined in spontaneously breathing rats. Stereotaxic coordinates (20) were used to unilaterally position a double-barreled micropipette (~15-µm inner tip diameter) into the FNr. One barrel of the pipette contained AZ [50 µM in warmed (38°C) mock cerebrospinal fluid, pH 7.37], and the other was prefilled with Chicago sky blue (2%). The latter (100 nl) was delivered after completion of the protocols to mark the injection sites. Only one site per side was tested in the FNr of an individual animal to determine the effects of AZ and/or vehicle on respiration in spontaneously breathing (18 rats) and paralyzed, ventilated rats (7 rats). Different doses (1 or 20 nl over 1 min) of AZ were microinjected into the FNr via a micropump or a pressure-injection device. For the latter, the volume injected was monitored and calculated by observing the displacement of the fluid meniscus via a microscope equipped with an eyepiece reticule (model E3069, Melles Griot). When respiratory responses returned to control for at least 10 min, AZ and/or vehicle injection was repeated in some cases to test the consistency of the AZ effect on respiration or serve as sham-operation controls, respectively. Second, we tested whether the presence of ventilatory chemosensitivity was specific to the FNr compared with other cerebellar deep nuclei. AZ was microinjected into the IN and/or LCN in 10 rats. In seven of these rats, AZ and/or vehicle was also microinjected into the FNr either before or after AZ administration made in the IN or LCN.

Histological examination. After completion of protocols, the animals were killed by administration of additional anesthetic and the brain stem and cerebellum removed and placed in 10% formalin. After at least 3 days of immersion fixation, the brain stem was frozen, and 50-µm sections were cut, mounted and stained for cell bodies. Tissue sections containing sites marked with Chicago sky blue were drawn with camera lucida.

Data acquisition and analysis. ABP, Ptr, and PET_O2 and PET_CO2 were monitored and/or recorded in all animals. We recorded and analyzed VT, f, and minute ventilation (E; product of VT and f) in spontaneously breathing rats, and peak PN (PN_peak), f, and minute PN (MPN; product of PN_peak and f) in paralyzed rats. Control (baseline) cardiorespiratory values were expressed as absolute values. For evoked respiratory responses, latency was defined as the onset time at which a detectible increase (>10% alteration compared with the control) was initiated. The t_max was defined as the time at which the maximal response of E and MPN (E_max and MPN_max, respectively) occurs. Values of ABP (mean ABP), PET_CO2, f, latency, and t_max in response to AZ or vehicle injection were expressed as absolute values. The amplitude of the respiratory responses (E, E_max, VT, MPN, MPN_max, and PN_peak) was presented as either absolute values or percent changes from control. Baseline values were obtained by averaging the relevant variables 1 min just before application of AZ. Cardiorespiratory responses were derived after AZ or vehicle injection for 2 h: every minute for the first 15 min and every 30 min thereafter. All data are presented as means ± SE. A Student's t-test was used to identify whether the data obtained at the period with maximal respiratory responses were significantly different 1) from the baseline values (before AZ injection), 2) between l- vs. 20-nl AZ injection into the FNr, and 3) between spontaneously breathing and the paralyzed and ventilated rats. Two-way ANOVA and Newman-Keuls test were utilized to examine whether the respiratory responses induced by AZ injection into the FNr significantly differed from those evoked by AZ injection into the IN and LCN or from those produced by vehicle injection into the FNr. A P value <0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Overall effects of microinjection of AZ into the FNr on ventilation. We evaluated the effect of microinjection of AZ (1 or 20 nl) and vehicle (20 nl) into the FNr on ventilation in the spontaneously breathing preparations (n = 27 trials). In 21 trials, the maximal ventilation was elevated by 46.0 ± 6.7% (P < 0.05). The response was initiated at 6.3 ± 0.8 min and reached the maximal at 19.7 ± 4.1 min after AZ injection. Vehicle injection (6 trials) did not significantly affect ventilation, with E_max increased by only 3.5 ± 2.1% (P > 0.01). Experimental recordings obtained in a spontaneously breathing rat are depicted in Fig. 1. As shown, vehicle injection did not affect ventilation (top), whereas AZ injection increased ventilation under isocapnic condition (bottom). Statistical results from the group data (Fig. 2) showed that the E_max in response to AZ injection into the FNr (220.4 ± 12.2 ml) was significantly greater than that obtained under control conditions (152.7 ± 7.3 ml; P < 0.01) and that obtained after vehicle injection (136.6 ± 12.5 ml; P < 0.01; not shown in Fig. 2). This augmented response was primarily due to an elevation of VT rather than f with little effect on ABP.

View larger version (39K): [in this window] [in a new window]   Fig. 1.   Effect of microinjection of vehicle (top) and acetazolamide (AZ; bottom) into the rostral fastigial nucleus (FNr) on ventilation in a spontaneously breathing rat. In each panel, the traces from top to bottom are end-tidal PCO2 (PETCO2) and tidal volume (VT). Recordings presented before and after the arrows reflect the baseline VT and its responses to AZ injection across the experimental time course. Horizontal bar, of 5 s.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Group data of AZ injection into the FNr-induced cardiorespiratory response obtained at the period with maximal ventilatory response. Data of minute ventilation (E; top left), mean arterial blood pressure (ABP; top right), tidal volume (VT; bottom left), and respiratory frequency (f; bottom right) were averaged from 21 trials. Values are means ± SE. *P < 0.05 between before (baseline ventilation) and after AZ administration.

Dynamic ventilatory response to AZ injection into the FNr in spontaneously breathing rats. Ventilatory responses to AZ injection into the FNr were not uniform, with the values of t_max showing either an early or late characteristic. In the present study, we classified ventilatory responses with a t_max shorter than 15 min as an early and that longer as a late response. Figure 3 contains typical examples of dynamic early response (A) and late response (B) to AZ (20 nl) injection into the FNr. Early response was observed in most trials (15 of 21 trials; 71%) in which the latency, t_max, and E_max were 4.7 ± 0.5 min, 9.6 ± 1.0 min, and 38.8 ± 6.3%, respectively. These values were all statistically different from those observed in late response (9.5 ± 1.2 min, 45.0 ± 6.7 min, and 64.4 ± 16.2%).

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Tracings of 2 different characteristics of the ventilatory responses, i.e., early response (A) and late response (B), to microinjection of AZ into the FNr.

Comparison of the ventilatory response to different AZ doses injected into the FNr. Microinjection of 20 nl (14 trials) and/or 1 nl AZ (7 trials) into the FNr was conducted. Compared with 1 nl, microinjection of 20 nl increased the average latency (from 5.1 ± 0.6 to 7.5 ± 1.1 min) and the t_max of the ventilatory response (from 11.9 ± 0.8 to 23.6 ± 5.9 min). As shown in Fig. 4, the values of E_max, VT, f, and ABP were slightly greater in response to 20-nl injection than those evoked by 1-nl administration. However, these changes did not reach significance (all P values > 0.05). Moreover, early- and late-response patterns were not dependent on the different AZ dosage used. For example, the early response was evoked by both 1-nl (7 trials) or 20-nl (8 trials) AZ injection.

View larger version (49K): [in this window] [in a new window]   Fig. 4.   Comparison of cardiorespiratory responses to different dosages of AZ in spontaneously breathing rats. Data of E (top left), ABP (top right), VT (bottom left), and f (bottom right) were averaged from 21 trials: 1 nl = 7 trials; 20 nl = 14 trials. Values are means ± SE.

Comparison of the ventilatory responses to AZ injection into the cerebellar deep nuclei. Neither ventilation nor ABP was altered by AZ (20 nl) injection into the IN (8 trials) and/or LCN (7 trials) in 10 spontaneously breathing rats. Typical experimental recordings are presented in Fig. 5A. Microinjection of AZ into the IN (top) or subsequently into the LCN (bottom) failed to significantly affect ABP and ventilation. Group data indicated similar results (Fig. 5B). Changes in ABP and E induced by AZ injections across the experimental time course were not significantly different from control. Values of E_max in response to AZ injected into different cerebellar deep nuclei were compared in Fig. 6. Compared with the control ("0" level), AZ injection into the FNr rather than into the IN and LCN significantly enhanced ventilation. In addition, among the ventilatory responses, the one induced by AZ injection into the FNr was significantly different from the other two (P < 0.01).

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Effect of microinjection of AZ into the interposed nucleus (IN) and lateral cerebellar nucleus (LCN) on ventilation in spontaneously breathing rats. A: experimental recordings showing the ventilatory responses to AZ injection (denoted by the arrow) into the IN (top) and LCN (bottom). In A, the traces from top to bottom are ABP, PETCO2, and VT. Group data (B) show dynamic respiratory responses to microinjection of AZ into these cerebellar deep nuclei. In B, the traces from top to bottom are ABP and E. These data were collected in 15 trials of 10 rats. Values are means ± SE.

View larger version (10K): [in this window] [in a new window]   Fig. 6.   Comparison of the maximal ventilatory responses (Emax) to microinjection of AZ into the FNr, IN, and LCN in 21 spontaneously breathing rats (36 trials): FNr = 21 trials, IN = 8 trials, and LCN = 7 trials. Values are means ± SE. Note: Emax in the cases with AZ injected into the IN and LCN was usually <10% as illustrated. *P < 0.05 between before ("0") and after AZ administrations.  P < 0.05 among the ventilatory responses obtained from AZ injections into the FNr, IN, and LCN.

PN in response to AZ injection into the FNr. To determine whether the ventilatory response to AZ injection into the FNr resulted from an increase in respiratory central drive, we repeated the experiments in seven paralyzed, ventilated rats using the PN as an index of respiratory output. Five of the animals displayed a significant increase in the phrenic motor output with both early-response (7 trials) and late-response (2 trials) characteristics. A typical experimental recording and group data are presented in Fig. 7, respectively. Under isocapnic conditions, AZ injection into the FNr clearly elevated PN in a paralyzed and ventilated rat (Fig. 7A). Group data (Fig. 7B) indicated that this augmentation of MPN_max was due to an elevation of PN_peak. Interestingly, the augmentation observed in MPN_max (38.9% ± 8.9) was not statistically different from that found in E_max (46.0 ± 6.7%). Furthermore, the latency and t_max in the paralyzed rats (5.3 ± 0.8 and 15.3 ± 6.1 min) were not markedly different from those in the spontaneously breathing rats (6.3 ± 0.8 and 19.7 ± 4.1 min; all P values > 0.05). With respect to baseline values, PET_CO2 level was statistically higher and f lower in the paralyzed and ventilated rats than those in the spontaneously breathing rats (Table 1).

View larger version (24K): [in this window] [in a new window]   Fig. 7.   Phrenic efferent nerve activity (PN) responses to AZ injection into the FNr. Typical examples of experimental recording (A) and group data (B) are shown. In A, the traces from top to bottom are PETCO2 and integrated PN (PN). Horizontal bar, 5 s. In B, data obtained from 9 trials in 7 paralyzed rats are shown. Values are means ± SE. *P < 0.05 between before and after AZ injection into the FNr.

Localization of AZ injections. Sites of injections are schematically presented in Fig. 8. Different symbols are used to mark the sites where injections were made. As shown, although there was some mingling, the sites where AZ injection produced early-respiratory augmentation appear to be clustered in the ventromedial region of the FNr.

View larger version (31K): [in this window] [in a new window]   Fig. 8.   Schematic presentation of the cross section of brain stem and cerebellum depicting AZ and/or vehicle injection sites in both spontaneously breathing and paralyzed rats. Different symbols are used to mark the sites where microinjections produce different responses: 1) the early response >25% increase in amplitude (; 17 trials in 13 sites) and <15% (; 5 trials in 4 sites); 2) the late response (; 6 trials in 6 sites); 3) the response to AZ injection into the IN and/or LCN (; 15 trials in 15 sites). One site (marked by *) initially displayed an early response (<15%, 1 nl), but second injection (20 nl) produced a late response. FN, rostral fastigial nucleus; IN, interposed nucleus; LCN, lateral cerebellar nucleus; Amb, ambiguous nucleus; VN, vestibular nucleus; LPGi, lateral paragigantocellular nucleus; RM, raphe magus nucleus; Py, pyramidal tract; DC, dorsal cochlear nucleus; and 4th, fourth ventricle.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Focal acidosis within the FNr increased ventilation significantly. In the present study, we found that microinjection of a low volume of AZ into the FNr significantly increased respiratory output in spontaneously breathing rats under isocapnic conditions. Because nonspecific injectate (vehicle) into the FNr did not produce remarkable respiratory alterations, this finding suggests that the FNr has intrinsic chemosensitive sites capable of facilitating the ventilatory response to hypercapnia. Cerebellar role in the control of skeletal muscle tone (respiratory muscles) has been well established. Our data showed that AZ injection into the FNr significantly increased PN activity in paralyzed and ventilated preparations with the response characteristics and amplitude similar to those observed in the spontaneously breathing animals. We conclude that AZ-induced ventilatory augmentation was primarily achieved by increasing respiratory central drive rather than by altering respiratory muscle tone. These data also suggest that AZ-induced respiratory response is independent of the inputs elicited by active contraction of respiratory muscles. Our finding is consistent with the previous observation that ablation of the whole cerebellum or selective destruction of FNr neurons significantly attenuated the respiratory response to hypercapnia in cats, rats, and dogs (21, 30, 32). There is accumulating evidence in both clinical and basic investigations that support the FNr involvement in respiratory chemoreflexes. It has been reported that FN neuronal activity was significantly lower in patients with congenital central hypoventilation syndromes and that the situation exaggerated during hypercapnia (12). Recently, investigators, using the perforated patch-clamp approach, have found that some cultured cerebellar neurons from postnatal rats exhibit CO₂/H⁺ responsiveness (27). These observations suggest the presence of cerebellar chemosensitive neurons, although it is uncertain whether these chemosensitive neurons are involved in respiratory chemoreflexes. Two important points have been raised by these authors: 1) the population of cerebellar neurons having chemosensitivity is lower compared with putative central chemoreception regions, and 2) some cerebellar chemosensitive neurons exhibit an inhibitory response to acidification. Our results indicated that the FNr is uniquely chemosensitive compared with other cerebellar deep nuclei. The fact that the FN neuronal population represents a relatively small percentage of the whole cerebellum (14) could account for the low number of chemosensitive neurons identified in the in vitro preparation containing all types of cerebellar neurons. With respect to the few neurons inhibited by acidification, we infer that those chemosensitive neurons might be Purkinje cells because Fos immunoreactivity in the Purkinje cell layer was reduced in rats exposed to hypercapnia compared with normocapnia, implying an inhibitory effect of hypercapnia on these cells (33).

Our finding supports the concept that multiple areas of the central nervous system have chemosensitivity involved in respiratory chemoreflexes. For many years, people have believed that putative central chemoreceptors are localized at the VMS because application of acidic fluid to this region increased respiration (16). However, several lines of recent evidence show a widespread distribution of central chemosensitive sites in the medulla, pons, and midbrain. First, acidification of various brain stem regions by microinjection of AZ or focal diffusion of mock cerebrospinal fluid equilibrated with 100% CO₂ has been used to locate sites of central chemoreception (6, 17, 19, 23). AZ has been applied onto the VMS or injected into the retrotrapezoid nucleus (RTN), NTS, ventral respiratory group, locus coeruleus, pre-Botzinger complex, and caudal hypothalamus (1, 4, 6, 16, 19, 23). In these studies, a dramatic elevation of the phrenic motor output associated with a focal tissue acidosis has been detected in the cat and rat when AZ was microinjected (6, 17). In agreement with theses findings, inactivation of these chemosensitive areas distinctly attenuated respiratory responses to hypercapnia (reviewed in Refs. 3 and 17). Second, it has been reported that neurons in putative chemosensitive areas and other medullary fields are labeled by Fos immunoreactivity in the rat exposed to CO₂, implying some of these labeled neurons are probably chemosensitive (22, 25). Third, with the use of the patch-clamp technique, cultured cells or tissue slices in vitro from the VMS (24), the NTS (7), the raphe nucleus (26), and the caudal hypothalamus (8) in the rat and cat display chemosensitivity. The presence of cerebellar ventilatory chemoreception demonstrated in our in vivo and chemosensitive neurons identified by other investigators' in vitro studies (27) further extends and confirms the concept of the widespread distribution of central chemosensitive regions. It is not clear by which efferent pathway(s) these cerebellar chemosensitive sites are connected with the respiratory central network to modulate respiratory chemoreflexes.

FNr chemosensitivity may selectively play a role in respiratory response to severe hypercapnia. After cerebellectomy or FNr ablation in anesthetized cats (32) and rats (30), an initial attenuation of the ventilatory response was not observed until the animal was exposed to moderate hypercapnia, and this attenuation exaggerated as the hypercapnia was intensified. Moreover, data from Fos immunoreactivity have shown that 10% but not 5% CO₂ exposure significantly activated FNr neurons (33). Specific FNr involvement in the respiratory response to severe hypercapnia is also supported by the presence of recruited respiratory-modulated neurons within the FNr. These neurons represent ~35% of FNr respiratory-modulated neurons recorded extracellularly in anesthetized cats (29). They were silent during eupneic breathing, but their respiratory rhythmic activity emerged when a greater respiratory effort was demanded, such as during severe hypercapnia. Selective FNr involvement in the ventilatory response during severe hypercapnia raised the possibility that the FNr had a higher CO₂ activation threshold than other putative chemosensitive areas. Therefore, we believe that central chemosensitive sites in the VMS, medullary raphe, RTN, and pre-Botzinger complex rather than in the FNr are essential to maintaining eupneic breathing. In contrast, FNr ventilatory chemosensitivity is triggered by severe hypercapnia, coupled with other central chemosensitive areas, to amplify the respiratory output.

Similarities and differences of the respiratory responses elicited by activation of chemosensitive neurons in the FNr and other central areas. An elevation of VT and PN_peak rather than f to augment the respiratory output was found after focal acidosis in the FNr consistent with the respiratory response to acidosis in the RTN, NTS, and rostral ventral respiratory group (5, 6, 17). Similar to these previous studies, we also found that AZ injection into some FNr sites failed to elicit notable respiratory augmentation, supporting the notion that the chemosensitive sites within the FNr are not distributed uniformly. In addition, the different characteristics (early vs. late response) and amplitude of the respiratory responses evoked by AZ injected into the FNr seem to be dose independent. In agreement with these findings, a previous report indicated that the amplitudes of respiratory responses to different AZ doses (10, 50, and 100 nl) injected into the same ventral medullary areas were not significantly different (5). Major differences between the respiratory response to AZ injection into the FNr and the RTN or Botzinger complex are the amplitude of maximal respiratory response and t_max. In our rat study, the averaged increase in PNpeak to AZ injection into the FNr was 38%, relatively lower than values observed after injection into the RTN (~60%) (6, 17) or the pre-Botzinger complex (~110%) (23). Because of the relative importance of these areas in respiratory rhythmogenesis and overall regulation of breathing, it is tempting to conjecture that the chemosensitive sites in the RTN and pre-Botzinger complex are more clustered or sensitive to hypercapnia than those in the FNr. In other words, a relative smaller response induced in the FNr may be related to a smaller population of cells in the FNr with a lower sensitivity to acidosis. In our experiment, the averaged t_max was 19.7 ± 4.1 min in spontaneously breathing rats and 15.3 ± 6.1 min in the paralyzed preparation. These values are smaller than those induced by AZ injection into the pre-Botzinger complex (47.5 ± 7.2 min; Ref. 23) or the RTN (30 min; Ref. 6). Why did AZ excitatory effects appear earlier in the FNr? AZ is an inhibitor of carbonic anhydrase (reviewed in Ref. 17); therefore, its presence and concentration must be a key to determine the effect of focal injection of AZ. One can infer that AZ injected into a region with higher concentration of carbonic anhydrase should produce a faster accumulation of CO₂/H⁺ in focal tissue. In fact, an extremely high concentration of carbonic anhydrase has been demonstrated in the cerebellum and its deep nuclei (18), which may contribute to the shorter onset latency and t_max induced by AZ injection into the FNr. The reason responsible for the early and late response we observed is unknown. Because the major sites to produce the early response are localized within the ventromedial region of the FNr, the difference may be partially related to the injection site.

Critique of methods. The exact mechanisms underlying AZ-induced focal tissue acidosis are not completely understood. An increase in brain PCO₂ via interference with the hydration of CO₂ or impairing transport in brain capillaries (red blood cells) and decrease in medullary extracellular fluid pH via focal accumulation of metabolically produced H⁺ were reported after intravenous injection of AZ (1, 2). Because inhibition of brain carbonic anhydrase can independently increase focal extracellular fluid H⁺ concentration and PCO₂, our data cannot distinguish whether the AZ-induced respiratory augmentation results from focal change in PCO₂ or H⁺ concentration, or both. The size of the region that is subjected to AZ-induced focal acidosis and diffusion of the injectate were not measured in the present study. However, previous studies have indicated that 1-nl AZ injection into the RTN of the rat and cat does not spread >350 µm (6). No direct evidence has been found to show the spread of 20-nl AZ injection. However, we found that some AZ injection sites in the IN ~600 µm lateral to the FNr did not affect respiration. In addition, other investigators have also reported that 20 nl AZ injected into the pre-Botzinger complex alters respiration but that injection 300-500 µm dorsal or medial to this region does not (23). These observations imply that the spread from 20-nl AZ injection is reasonably limited.

Summary. Although other cerebellar deep nuclei are also involved in respiratory modulation, the FNr uniquely presents ventilatory chemosensitivity. Similar to other central chemosensitive areas recently discovered, the FNr has ventilatory chemosensitivity and contains respiratory-modulated neurons. Because FNr facilitation of the ventilatory response emerges during severe hypercapnia (30, 32), it is highly likely that both FNr chemosensitive sites and respiratory-modulated neurons are recruited and contribute to the respiratory augmentation during severe hypercapnia.