THE JOURNAL OF COMPARATIVE NEUROLOGY 302:48!5499 (1990)

Medullary Projection of Nonaugmenting hspiratoxy Neurons of the Ventrolateral Medulla in the Cat KAZUYOSHI OTAKE, HIROSHI SASAKI, KAZLTHISA EZURE, AND MOTOMU MANABE Department of Anatomy, Faculty of Medicine, Tokyo Medical and Dental University, Yushima 1-5-45, Bunkyo-ku, Tokyo 113 (K.O., H.S.), and Department of Neurobiology, Tokyo Metropolitan Institute for Neurosciences, Musashidai 2-6, Fuchu-shi, Tokyo 183 (K.E., M.M.), Japan

ABSTRACT In Nembutal-anesthetized, immobilized, and artificially ventilated cats, we studied the morphological characteristics of inspiratory neurons with nonaugmenting firing patterns. HRP was injected intracellularly into a total of 22 neurons of the Botzinger complex (BOT) and the ventral respiratory group (VRG).In 20 cases somata with their axonal trajectories were stained, and in two cases only axons were stained. None of the neurons stained could be antidromically activated by stimulation of the cervical cord. The somata of 20 neurons were located in the vicinity of the nucleus ambiguus or the retrofacial nucleus (RFN) between 600 km and 2,800 km caudal to the rostra1 end of the RFN. Their axons could be traced for a distance of several millimeters on the side of the somata, and showed various projection patterns. According to these projection patterns, the 20 neurons were tentatively classified into four groups: A (8/20), B (4/20), C (6/20), and motoneurons (2/20). Group A neurons gave off extensive axon collaterals that arborized and distributed boutons predominantly in the BOT and the VRG areas. Group B neurons had less extensive axon collaterals with various projection patterns, projecting rarely to the BOT or the VRG area. Group C neurons sent their stem axons, without issuing any axonal collaterals, to the contralateral side in five cases and to the ipsilateral pons in one case. The two motoneurons had axons leaving the brainstem without any intramedullary collaterals. Thus, the nonaugmenting inspiratory neurons showed morphological variations, which may play different roles in neural control of respiration. Key words: respiratory neurons, ventral respiratory group (VRG), Botzinger complex (BOT), intracellular HRP, intramedullary collaterals

The majority of inspiratory neurons of the ventral (VRG) and dorsal respiratory group (DRG) exhibit an augmenting firing pattern (Haber et al., '57; Bianchi, '71; Merrill, '74; Cohen, '79). Intermingled with these neurons, there are other inspiratory neurons of nonaugmenting discharge patterns (Nesland and Plum, '65; Bianchi and Barillot, '82; Merrill, '72; Ezure et al., '86, '89; Cohen and Feldman, '84; Otake et al., '89a,b). Some of the nonaugmenting inspiratory neurons show burst firing which declines within the inspiratory phase and have been called early-burst or decrementing inspiratory neurons (Bianchi, '71, '74; Merrill, '74, '81). Others exhibit rapid firing which starts at the beginning of the inspiratory phase and remains constant throughout the phase. This firing pattern corresponds to the "constant" type of Cohen ('79) and also to the "tonic" type of our previous studies (Ezure et al., '86, '89; Otake et al., '89a). o 1990 WILEY-LISS, INC.

These nonaugmenting inspiratory neurons were shown to have extensive axonal arborization in the medulla and rare spinal projections by antidromic mapping or by HRP intracellular staining (Merrill, '72, '74, '81; Ezure et al., '89, '90). The decrementing inspiratoly neurons as well as the constant inspiratory neurons project to the VRG, the Botzinger complex (BOT), and the DRG (Merrill, '74; Ezure et al., '86, '89, '90; Otake et al., '89a). Spike-triggered averaging showed that spikes of the constant inspiratory neurons induced monosynaptic EPSPs in the inspiratory neurons of both the VRG and the DRG (Ezure et al., '891, and spikes of the decrementing inspiratory neurons evoked Accepted August 14,1990. Address reprint requests to Dr. K. Otake, Department of Anatomy, Faculty of Medicine, Tokyo Medical and Dental University, Yushima 1-5-45, Bunkyo-ku, Tokyo 113,Japan.

K. OTAKE ET AL.

486

HRP injections. Our main interest was to reveal the morphology of their axonal projections, in order to disclose neural circuitry formed by medullary respiratory neurons. Preliminary reports of this work were presented in abstract form (Ezure et al., '86; Otake et al., '89a).

Abbreviations

VII XI1

AMB DV DX

Mv

RFN S

facial nucleus hypoglossal nucleus nucleus ambiguus descending vestibular nucleus dorsal motor nucleus of the vagus medial vestibular nucleus retrofacial nucleus solitary tract

MATERIALSANDMETHODS

monosynaptic IPSPs in the inspiratory neurons of both the VRG and the DRG (Ezure et al., '89) and in the expiratory neurons of the BOT and the VRG (Ezure and Manabe, '86; Ezure et al., '90). These observations suggested important roles for these nonaugmenting inspiratory neurons in the respiratory rhythmogenesis (Merrill, '74, '81; Richter, '82; Ezure et al., '89). In the present study, we investigated the morphological characteristics of nonaugmenting inspiratory neurons located in the area from the BOT to the VRG by intracellular

Ia1 -

Data were obtained from 17 adult cats. Some of the animals were from the same population as used in previous studies (Otake et al.,'87, '89b; Sasaki et al., '89). We only briefly describe the methods here, since almost all the experimental procedures were the same as those described in previous papers.

surgicalprocedures All the animals were initially anesthetized by intraperitoneal administration of pentobarbital sodium (Nembutal, Abbott) (40 mgkg). The anesthesia was subsequently maintained throughout the experiment and the survival period by additional intravenous doses of pentobarbital sodium. Craniotomy was performed and the caudal part of the

I

I1a

I

Ib

I

L*

IIb

IIIb

-

IIIC

Ic

Id

I

Fig. 1. Examples of intracellular recording from ten nonaugmenting inspiratory neurons stained with HRP (Ia-d, Ha-b, 1IIa-d). Ia-d: Constant-type neurons. 1IIa-d. Decrementing-type neurons. JIa-b Unclassifiable neurons. For comparison, an augmenting inspiratory

1s

I

IIId

I

neuron (IV)recorded in the same animal as that for the nonaugmenting neuron (Id) is shown. Upper traces are intracellular potentials; lower traces are rectified and integrated phrenic nerve activity. Calibrations shown in IV are applicable to the corresponding records of 1-111.

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX

487

Fig. 2. Photomicrographs. A: Stained soma ventral to the RFN. White star: the injection site of HRP; arrow: the point of emergence of an axon collateral from a stem axon. B: Axonal branches and the terminal boutons in the VRG area. Arrowheads: the boutons: asterisks: the counterstained cells.

cerebellum was aspirated to expose the brainstem 5-6 mm rostra1 to the obex. A pair of stimulus electrodes was fixed bilaterally in the spinal cord between C4 and C5 for stimulation of descending axons of the brainstem neurons. Each animal was paralyzed (pancuronium bromide 0.15 mg/kg/hour, i.v., Mioblock, Organon) and artificially ventilated. A bilateral pneumothorax was performed and an expiratory load of 1-2 cm/H,O was applied. CO, concentration of expired gas and blood pressure in the femoral artery were monitored. Rectal temperature was controlled between 37 and 385°C.

HRP injection and histologid procedures Activity was monitored from the C5 phrenic nerve. A glass micropipette filled with a 10% solution of HRP (Toyobo, Grade I-C) in 0.5 M KC1-Tris buffer (pH 7.6) was used for intracelular recording. After intracellular penetration, the firing patterns or changes of membrane potentials were examined and projection to the spinal cord was tested before the HRP injection. HRP was injected by positive current pulses (5 pulsesisecond, 100 ms duration, and 15 nA maximum current) for 1-15 minutes. Two to 32 hours after HRP injection the animals were deeply anesthetized and transcardially perfused with a solution of 1.5%glutaraldehyde and 1.5% paraformaldehyde in 0.1 M phosphatebuffered saline. The brainstem was cut transversely at 120 pm with a freezing microtome. The sections were treated with diaminobenzidine (DAB) and were counterstained with 0.1% thionin for histological examination. In some experiments, cobalt chloride and nickel ammonium sulfate were used together with DAB for intensification (Adams, ’81). The stained neurons were reconstructed by using serial photomontage as reported in previous papers (Otake et al., ’87,’89b; Sasaki et al., ’89).

RESULTS HRP was injected into respiratory neurons in the ventral respiratory group (VRG) or the Botzinger complex (B8T).

We analyzed a total of 22 inspiratory neurons whose firing patterns or trajectories of membrane potentials were of the nonaugmenting type. In 20 neurons the somata with axonal trajectories were stained, and in two neurons only the axonal trajectories were stained.

Electrophysiologicdcharacteristim All of the 22 inspiratory neurons stained were tested for their projection to the spinal cord. None of them were antidromically activated from the spinal cord at the level of C4-5. The 22 neurons showed nonaugmenting patterns of firing or membrane potential trajectory (Fig. l),although impalements of inspiratory neurons by HRP electrodes sometimes increased (Fig. 1IbJIIb) or blocked (Fig. 1IIb,IIIa,IIIc) the spike activity of the impaled neurons by depolarizing their membrane potentials, These nonaugmentingpatterns had some variation. Membrane potentials of some of the neurons abruptly depolarized at the onset of the inspiratory phase and maintained fairly constant depolarized potential levels or accompanying firing during the inspiratory phase (Fig. 1Ia-d). At the end of the inspiratory phase the membrane potentials repolarized abruptly and the firing, if any, also stopped. This type of neurons was termed “constant” (Cohen, ’791, although we have called the same type “tonic” in previous papers (Ezure et al., ’86, ’89). This constant type was differentiated from the augmenting type. In Figure lIV, a typical augmenting inspiratory neuron, which was recorded TABLE 1. Classificationof Nonaugmenting InspiratowNeurons

Morphological classification Constant Group A Group B Group C Motaneuron

Total

Firing pattern In-between

Somal sue

Decrementing (mean 5 S.D., pm)

5 0 5 1

3 1 0 0

0

1

28.7 i 7.3 (n = 8) 25.0 i 0 9 (n = 4) 29.2 i 6.7 (n = 6) 36.0 ( n = 2)

11

4

5

28.8 i 6 5 ( n = 20)

3

488

K. OTAKE ET AL.

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX from the same animal as the constant inspiratory neuron shown in Figure lId, is presented for comparison. Another nonaugmenting pattern which some of the 22 neurons exhibited was the decrementing type and is exemplified in IIIa-d in Figure 1. Their membrane potentials started depolarization at the onset of the inspiratory phase, quickly attained their most depolarized levels, and then repolarized gradually to the previous levels (Figs. 1IIIa-c). When spike activity was observed, the firing frequency exhibited a decrementing pattern during the inspiratory phase (Fig. lIIIb,IIId). In some cases, we could not classify the neurons as either constant or decrementing (Fig. lIIa,IIb). We treated them as indeterminate (in-between type) without venturing to classify them, as we did in a previous study (Ezure et al., '89). Of the 20 neurons whose somata were stained, 11were of the constant type and five were of the decrementing type. The remaining four were indeterminate. The two axom exhibited a decrementing firing pattern.

Correlationof morphologywith electrophysiologicalcharacteristics We could trace the axonal trajectories of stained neurons for a distance of several millimeters on the injection side, although we could not observe the contralateral morphology. The 20 neurons whose cell somata were stained had variations in axonal morphology (Fig. 2). We classified them into four tentative groups according to the pattern of their axonal projections. Eight of them had axonal branches arborizing extensively in the BOT or the VRG areas (which we call the BOTFRG area for convenience). These were classified as Group A neurons. Four neurons had less extensjve axonal collaterals which projected only rarely to the BOTFRG area, and were classified as Group B neurons. Six neurons revealed no axon collaterals and were classified as Group C neurons. The remaining two were motoneurons, whose axons left the brainstem without issuing any collaterals. This classification is not applicable to the two axons whose somata were not stained. We examined the correlation between these morphological characteristics and their firing patterns as shown in Table 1. In general, the Group A neurons showed the constant firing pattern and the Group B neurons exhibited the decrementing firing pattern. Most of the Group C neurons showed the constant firing pattern. The motoneurons showed either the constant or the decrementing firing pattern.

Neuronswithcollaterak i ) Group A. The somata were located about 600 to 1,600 pm caudal to the rostral end of the retrofacial nucleus (RFN) and ventral to the RFN, corresponding almost exactly to the caudal part of the BOT area. The RFN here is defined as the ventrolaterally located external division of rostral nucleus ambiguus, which corresponds to the definition by Taber ('61) and Berman ('68). The rostral end of the

Fig. 3. Morphology of an inspiratory neuron of the Group A in a transverse (A) and a horizontal (B) plane. Collaterals a 4 in A correspond to a 4 in B, respectively. C: histogram of the rostrocaudal distribution of boutons. Each bar indicates the number of boutons observed on one 120 IJ-m section. The thick bar along the ordinate indicates the rostrocaudal extension of the RFN. The level indicated with an asterisk shows the site of the cell soma.

489

RFN corresponds to the caudal end of the facial nucleus. Since the eight Group A neurons showed similar axonal morphology, we describe here three examples (Figs. 3-5). The firing patterns of the neurons of Figures 3 and 5 are presented in Figure 1Ia and lIIa, respectively. The stem axon of most of them (7/8)bifurcated into a crossing and a descending axon (Figs. 3, 4). The descending axons descended in the reticular formation around the nucleus ambiguus (AMB). The neuron shown in Figure 5, on the other hand, had only a crossing axon that crossed the dorsal part of the midline. This neuron did not give off any descending axons, although two thinner collaterals directed dorsocaudally (arrowheads in the figure) were issued from the stem axon. From the stem axons of these eight neurons, three to four long axonal branches were given off, which ran both rostrally and caudally to the level of the soma of origin and ramified successively in the close vicinity of the RFN and the AMB, distributing numbers of boutons. The boutons were distributed in the vicinity of the RFN and/or the AMB, mainly ventral to these nuclei, around the somata of origin. A few boutons were also found within the RFN or the AMB. Some of the neurons revealed axonal collaterals running dorsomedially around the solitary tract or the dorsal motor nucleus of the vagus (indicated with arrowheads in Figs. 3-5). These collaterals distributed a few boutons in the reticular formation of the dorsomedial medulla. Sometimes a few branches were issued from the stem axon, the crossing axon, or the descending axon, in the middle part of the medullary reticular formation, and few boutons were observed. The rostrocaudal distribution of the terminals of the neurons of this type is presented in Figures 3C, 4C, and 5C. The boutons were found from the level of the rostral end of the RFN as far as the level of the rostral part of the AMB. In each case, the distribution of terminals made a peak almost at the same level of its soma. The ratio of the number of boutons en passant to that of boutons terminaux in the BOTFRG area was 2.47 (n = 8). i i ) Group B. The four Group B neurons revealed various projection patterns. The axonal morphology of all four is shown in Figures 6-8. The somata of these neurons were scattered in the VRG between 1,100 pm and 2,400 km caudal to the rostral end of the RFN. Case 1 (Fig. 6). The neuron shown in Figure 6 exhibited a decrementing firing pattern. The stem axon of this neuron bifurcated into a crossing and a descending axon. The crossing axon issued a long ascending collateral which gave off several branches in and around the RFN and around the facial nucleus. A few terminals were distributed within the RFN and the reticular formation dorsomedial to the RFN (Fig. 6B-a). The descending axon gave off successively at 500-800 km intervals four long collaterals directed medially or dorsomedially in the medial part of the reticular formation. One of these collaterals distributed boutons in the reticular formation between the AMB and the dorsal motor nucleus of the vagus (Fig. 6B-b). Two collaterals given off at a more caudal level ran toward the dorsal motor nucleus of the vagus and the hypoglossal nucleus and arborized in or near these nuclei, but the terminal boutons were not stained (Fig. 6B-b). Cases 2 and 3 (Fig. 7). Both of two neurons revealed no boutons in the close vicinity of the somata. These neurons had only crossing stem axons, and no descending axons. The firing patterns of these neurons are presented in Figure lIIb (neuron a) and lIIIa (neuron b).

K. OTAKE ET AL.

490

Figure 4

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX

49 1

Fig. 5. A Group A inspiratory neuron without descending axons in a transverse (A) and a horizontal (B) plane. Two collaterals indicated with arrowheads run dorsomediocaudally toward the dorsal motor nucleus of the vagus. C: Histogram of the rostrocaudal distribution of boutons.

One neuron (neuron a) gave off from the stem axon a long ascending axon collateral in the reticular formation along the medial or dorsomedial border of the AMB. From both the crossing stem axon and the ascending collateral, many short branches with boutons were issued. Characteristically, these branches ran only for a short distance and therefore terminals were distributed in the immediate vicinity of the stem axon or ascending collateral. The boutons from the ascending collaterals were distributed in the area medial and dorsal to the rostral part of the AMB and a few also within the AMB, at a level from 1,500 pm to 3,000 km caudal to the RFN rostral end (Fig. 7C-a). Terminals were also found in the medial part of the reticular formation along the stem axon at a level slightly caudal to the soma of origin (Fig. 7C-a’). The stem axon of neuron b gave off a long axon collateral, which was directed dorsomedially toward the hypoglossal nucleus. In the vicinity of the nucleus the axon changed its course in a ventrolateral direction. Along its trajectory, it issued some short thin branches and distributed a few terminal boutons in the reticular formation between the AMB and the hypoglossal nucleus, especially in the vicinity of the latter nucleus (Fig. 7B,C-b). Case 4 (Fig. 8).This neuron exhibited the decrementing firing pattern which is shown in Figure 1IIIb. The stem axon of this neuron, after giving off a long thin collateral which descended along the ventrolateral surface of the medulla (arrowheads in Fig. 81, bifurcated into a crossing and a descending axon. Further axonal arborization was not observed and no terminal boutons were found.

Neuronswith no ~Ilaterals i) Group C. As shown in Figure 9, six neurons had no ipsilateral axon collaterals within the observable range and were classified as Group C. Most of them (5/6) showed a constant firing pattern, and examples of their firing pattern are presented in Figure lIb, Ic, and Id. Their somata were located in the VRG area, especially ventral or lateral to the AMB about 1,500 pm and 2,600 pm from the rostral end of the RFN (Fig. 9). Five of the six had stem axons which coursed medially and crossed the midline. The crossing levels were in the ventral part of the medulla at the same level as the somata or a little more rostrally. The one remaining neuron, whose firing pattern is shown in Figure lIc, was situated relatively caudally, lateral to the AMB, and had a stem axon which coursed laterally, then turned rostrally, and ascended in the rostral medulla. This axon ascended along the lateral surface of the brainstem into the restiform body and could be traced as far as the pontine level about 2,500 pm from the rostral end of the RFN. ii) Motoneurons. Two neurons were motoneurons, judging from their axonal trajectories (Fig. 10). The firing pattern of one of them (neuron a) is presented in Figure 1IIIc. Their stem axons, after leaving the somata, coursed dorsomedially and reached the dorsal motor nucleus of the vagus (a) or its vicinity (b). Then they turned laterally, forming loops, joined the radix nervi vagi, and left the brainstem. In their entire trajectories, no axon collaterals were issued.

Axonal arborization and terminal distribution

in the contralateralside

Fig. 4. Morphology of an inspiratory neuron of Group A in a horizontal (A) and a transverse (B)plane. Arrowheads indicate an axon collateral toward the dorsal motor nucleus of the vagus. Arrows indicate the descending axons. C: Histogram of the rostrocaudal distribution of boutons.

Figure 11shows the superimposed representation of two neurons whose axons only were stained on the side contralatera1 to the somata of origin. Both neurons showed the typical decrementing firing pattern, which is presented in Figure lIIId for the neuron b. Their stem axons came from

K. OTAKE ET AL.

492

Fig. 6. One neuron belonging to Group B is shown in a horizontal (A) and two transverse (B) planes. The levels a and b in A correspond to the sections a and b in B.

the contralateral side at the level 600 km (neuron a) and 3,200 pm (neuron b) caudal to the rostral end of the RFN. The axons, on arrival in the area ventral to the RFN or AMB, turned caudally and descended through the reticular formation ventral to the RFN or the AMB toward the caudal medulla (Fig. llA,Ba,Ba’,Bb). Both of the two neurons also had medullary axon collatera l ~From . axon a, five collaterals were issued in the vicinity of the RFN and the rostral part of the AMB. The other axon (b) gave off a long ascending collateral, although the stem axon disappeared after it turned caudalward due to faint staining. The collaterals of these axons arborized extensively in the vicinity of the RFN and distributed terminals ventral to the RFN (Fig. 11Ba,b). The rostrocaudal distribution of boutons is shown in the histogram in Figure 11C. Each axon distributed boutons with a peak at the level of the caudal part of the RFN. Boutons of axon a were also found in the area ventral and ventrolateral to the AMB about 3 mm caudal to the rostral end of the RFN (Fig. 11Ba’).

Cellular location and size The location of the 20 nonaugmenting inspiratory neurons is summarized in Figure 12. They were located in and

around the AMB or in and around the RFN, especially ventral or ventrolateral to these nuclei (Fig. 12A), between 600 pm and 3,600 pm caudal to the rostral end of the RFN (Fig. 12B). The somata of the Group A (filled circles) and the Group B neurons (filled triangles) were located in the BOT/VRG area, whereas the Group C neurons (open symbols) were found exclusively in the VRG area. The Group A neurons were situated more laterally than the Group B neurons. The mean somal size (algebraic mean of major and minor axes) of the neurons of each group is shown in Table 1,and the plots of Figure 13 present the major and minor axes of these neurons. The mean somal size ranged from 21.0 pm to 43.5 pm (n = 20, including two motoneurons). The Group A and the Group C neurons had varied somal sizes ranging from 21.0 to 43.5 pm. The Group B neurons were all relatively small. The motoneurons had relatively larger somata. The mean somal size of all nonaugmenting neurons (excluding two motoneurons) was 28.0 6.2 p m (n = 18). This value was significantly smaller than that of the augmenting inspiratory neurons of the VRG (36.8 2 7.3 km, n = 13) of the previous work (Sasaki et al., ’89) (t-test, P < 0.01).

*

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX

493

Fig. 7. Two Group B neurons without descending axons are presented on a horizontal (A) and a transverse (B) plane. C: The rostrocaudal distribution of terminals of the neuron a is presented in

Ca (for the terminals distributed by the ascending collateral) and Ca‘ (for the terminals of the crossing stem axon), and that of neuron b is presented in Cb.

DISCUSSION

limited sample of neurons, and 3) the relatively low quality of intracellular recording, which is the major problem of penetrating small neurons with high-impedance HRP electrodes. Furthermore, our classification of the stained neurons based on their axonal projections might require some reservation, since there was the possibility that HRP failed to fill some thin collaterals. Therefore, taking these technical limitations in both electrophysiological and morphological methods into consideration, we carefully examined the correlation between axonal projections and firing patterns of the stained neurons.

The firing pattern of inspiratory neurons can be classified into three fundamental types as clearly described in earlier studies (Cohen, ’79; Berger, ’81).The first is the augmenting or late-peak type, and the second is the decrementing or early peak (early burst) type. The third is the constant type (Cohen, ’79). The dividing lines between the augmenting, constant, and decrementing types are not always apparent. Actually, we have encountered a few inspiratory neurons whose firing patterns were hard to classify into either the augmenting or the nonaugmenting pattern. We ignored such neurons in the present analysis, and studied only the nonaugmenting inspiratory neurons. Our previous extracellular study showed that the classification of nonaugmenting inspiratory neurons into subtypes, i.e., the constant and the decrementing neurons, is sometimes difficult (Ezure et al., ’89).This difficulty seemed to be amplified in the present intracellular study, because of 1)the modification of firing and membrane potential due to intracellular impalement as described in the Results, 2) the

Nonaugmenting inspiratmy neurons In previous papers (Ezure et al., ’86, ’89; Otake et al., ’89a), we used the term “tonic” for the constant type. The term “tonic” may be inadequate, since this term is usually used for respiratory neurons which fire throughout the respiratory cycle (Bianchi, ’71, ’74; Cohen, ’79). Therefore, the term “constant” or “plateau” is more appropriate; here we use constant instead of tonic.

K. OTAKE ET AL.

494

Fig. 8. A neuron of Group B is presented on a horizontal (A) and a transverse (B)plane. The stem axon, after giving off a long thin collateral (arrowheads),bifurcated into a crossing and a descending axon.

There are many studies (e.g., Merrill, '74; Bianchi, '74) on the augmenting and the decrementing inspiratory neurons, but very few on the constant inspiratory neurons. Researchers have not necessarily recognized this third type as one group. Sometimes inspiratory neurons with typical constant firing have been included in a group of augmenting or late-peak inspiratory neurons (see Fig. 19.5 of Merrill, '74). However, the present neuroanatomical study together with our previous electrophysiological study (Ezure et al., '89) has proved that the constant inspiratory neurons are differentiated from either augmenting or decrementing inspiratory neurons. In the present study, four of the 22 inspiratory neurons were of an in-between type. We did not try to classify these four neurons, but they could belong to either of the two types. For instance, the Group A neuron whose firing pattern is shown in Figure 1IIa is probably of the constant type; and the Group B neuron shown in Figure lIIb is very much like the decrementing type. Therefore, we may say that the firing pattern of the Group A neurons is basically of the constant type and that of the Group B is of the decrementing type. Although both Group A and Group C neurons showed generally constant firing patterns, their morphology was clearly different. We also found differences in the firings of the two groups. The firing frequencies of many Group A neurons were high, usually exceeding 100 Hz, and some-

times beyond 200 Hz as shown in Figure 1Ia. However, the frequencies of the Group C neurons were low, usually around several tens of hertz. Previously we showed that constant inspiratory neurons are excitatory neurons which project to the VRG and the DRG (Ezure et al., '89). We are almost certain that the excitatory constant inspiratory neurons belong to the present Group A: see the highfrequency burst of the constant inspiratory neuron shown in Figure 1of Ezure et al.('89). Two nonaugmenting inspiratory neurons stained were revealed to be motoneurons. This may partly explain why the inspiratory activity recorded from the whole recurrent laryngeal nerve exhibited a nonaugmenting, slightly decrementing, pattern in contrast to the augmenting pattern of simultaneously recorded phrenic nerve activity (Cohen, '75). However, we do not mean that medullary inspiratory motoneurons in general show decrementing or constant firing patterns. Many inspiratory motoneurons within the medulla exhibit augmenting firing patterns (Bianchi, '74; Bianchi et al., '88). This underscores the difficulty in identifying propriobulbar neurons or motoneurons simply from their firing patterns or the trajectories of membrane potential.

Cellular location and morphology The present 20 nonaugmenting inspiratoy neurons were distributed from the caudal part of the BOT to the VRG

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX

495

I I I

a+

----

I500

I

--

1

I

I

I

R~PHE Fig. 9. Axonal trajectories of six inspiratory neurons belonging to Group C are superimposed onto a horizontal (bottom)and two transverse (upper right, a, b) planes. The levels indicated as a and b in the horizontal plane correspond to a and b of the transverse planes.

rostral to the obex. Nonaugmenting inspiratory neurons are intermingled with expiratory neurons of the VRG caudal to the obex (Arita et al., '87) and with the DRG augmenting inspiratory neurons (Cohen and Feldman, '84; Otake et al., '89b). Bianchi et al. ('88) stained with HRP a decrementing inspiratory neuron which was located dorsal to the caudal portion of the facial nucleus. It is difficult, however, to relate it to the present study, since its axonal trajectory was not reported. There is not much information about the location of constant inspiratory neurons. The present study suggests that the constant neurons of Group C are distributed throughout the VRG, whereas those of Group A, which have extensive medullary projections, are distributed from the caudal BOT to the very rostral part of the VRG. Our previous neuroanatomical study with HRP stained three nonaugmenting inspiratory neurons in the DRG area, one of which issued extensive axonal branches in the medulla and showed a rather constant firing pattern (see Fig. 11 of Otake et al., '89b). These observations may indicate that the constant inspiratory neurons as well as the decrementing inspiratory neurons are distributed in the BOT, the VRG, and the DRG. Electrophysiological study showed a tendency for the constant inspiratory neurons to be distributed more later-

ally than the decrementing inspiratory neurons (Ezure et al., '89). The present Group A neurons were located more laterally than the other neurons, as shown in Figure 12. This further supports the notion that the Group A neurons correspond to the constant inspiratory neurons of the previous study. The mean somal size of bulbospinal inspiratory neurons of the VRG, whose firing patterns were augmenting, was 36.8 & 7.3 pm (n = 13) (Sasaki et al., '89). The mean somal size of the present nonaugmenting inspiratory neurons was 28.0 -t 6.2 pm (n = 18) and significantly smaller than the former. As shown in Figure 13 and Table 1, the Group B neurons were especially small, although the somal sizes of Group A and C neurons were distributed in a rather wide range, and some of them were relatively large. This is consistent with Merrill's observation ('74) that decrementing inspiratory neurons are much smaller than other inspiratory neurons in the VRG because of the short extracellular recording times and the small spike amplitudes. Kreuter et al. ('77) investigated the morphology of the VRG inspiratory neurons by intracellular injection of procion yellow. Although they did not identify either constant or decrementing inspiratory neurons, they showed that the somata of propriobulbar neurons were signifi-

K. OTAKE ET AL.

496

The present neuroanatomical study, jointly with the previous electrophysiological study (Ezure et al., '891, shows that nonaugmenting inspiratory neurons form a heterogeneous population. Although we classified them into four tentative groups, it seems that there are further subgroups

in them. For instance, all of the four Group B neurons which exhibited more or less decrementing firing patterns were morphologically different from each other. One Group C neuron which had the axon ascending ipsilaterally was different from the other five of the same group, whose axons crossed the midline to the contralateral side. On the other hand, the Group A neurons may form a morphologically homogeneous group. Recent electrophysiological study with spike-triggered averaging of intracellular potentials gave direct evidence that there were at least two types of nonaugmenting inspiratory neurons in the BOTNRG area, i.e., excitatory and inhibitory neurons (Ezure et al., '89). The excitatory neurons tended to exhibit constant firing and made monosynaptic connections with inspiratory neurons of the VRG and the DRG, while the inhibitory neurons exhibited decrementing firing and made monosynaptic connections with inspiratory neurons of the VRG and the DRG and with expiratory neurons of the BOT and the caudal VRG (Ezure and Manabe, '86; Ezure et al., '89, '90). Judging from the firing patterns and locations, these excitatory and inhibitory neurons presumably correspond to the present Group A and Group B neurons, respectively. The above synaptic connections were made with respiratory neurons on the contralateral side. In the present study, however, only ipsilateral axonal branches of the inspiratory neurons were found, except for two decrementing axons. Therefore, the correspondence is not so direct. We are almost certain that the morphology of the excitatory neurons is represented in the Group A neurons, which show characteristic high-frequency burst firing and make extensive arborizations in the ipsilateral BOTNRG area. The two axons showing decrementing firing and ha-ng extensive axonal arborizations in the contralateral BOTNRG area were presumably the axons of inhibitory neurons. On the other hand, ,,projections from Group B neurons to the ipsilateral BOTNRG area were rare. This raises the question whether the axons of these four Group B neurons project to the BOTNRG area or the DRG on the contralatera1 side. There are three possibilities. First, decrementing inspiratory neurons may make extensive arborizations on the contralateral side, but have only sparse projections to the ipsilateral VRG area, as suggested by Merrill ('72, '74). Second, there may be decrementing inspiratory neurons which are inhibitory and project to both the ipsilateral and contralateral BOTNRG and DRG areas. Third, some inbetween-type neurons of Group A, such as shown in Figures lIIa and 5, may be the inhibitory neurons. Where in the brainstem do the Group C neurons project? One of the six neurons, which ascended the ipsilateral medulla into the pons, may transmit rhythmic activity to the structures rostral to the medulla. The same role may be played by the ascending collaterals observed in some of the Group A and B neurons as well as some augmenting VRG neurons (Sasaki et al., '89). These ascending projections may correspond to the bulbopontine projections revealed by neuroanatomical (Kalia, '77, '81; King, '80; Smith et al., '89) and electrophysiological (Bianchi and St. John, '81) studies. It remains to be studied where the other five of the

Fig. 11. A: Horizontal projection of trajectories and branches of the two axons whose somata were not stained. B: The distribution of the terminals of the neuron a is superimposed in two transverse planes (a for the level of the BOT area, and a' for the level of the rostral VRG

area) and that of the neuron b in one transverse plane (b). Open circles with arrows in Ba, a', and b indicate the pathways of the stem axon. C: The rostroeaudal distribution of the terminals is presented in the histogram (neuron a on the left and neuron b on the right).

Fig. 10. Morphology of two inspiratory motoneurons in transverse planes.

cantly smaller than those of bulbospinal neurons. Our data basically agree with these studies.

Functionalconsiderations

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX

497

3 3

$3

E

K. OTAKJ3 ET AL.

498

B c

T

a -I1

I/

b _II

/I

C

-I'

/I

d

-e 11

/I

-1)

/I

C

f 1

Fig. 12. Location of 20 inspiratory neurons is shown in transverse (A) and horizontal (B) planes. The levels indicated as a,b,c,d,f in A correspond to a,b,c,d,f, in B. The neurons belonging to Group A are indicated with filled circles; the neurons of Group B with filled triangles; neurons of the Group C with open circles (for the neurons of

*

Group C neurons project and what their functional roles are. Some fine collateral branches of both Group A and Group B neurons projected to the hypoglossal nucleus, the dorsal motor nucleus of the vagus, and the reticular formation dorsomedial to the AMB. These areas are outside the typical respiration-related areas. Similar projections were observed from augmenting inspiratory neurons of the VRG and the DRG, some of which projected to these areas as well as to the BOT, the VRG, or the spinal cord (Sasaki et al., '89; Otake et al.,'89b). This suggested that they influence the accessory respiratory muscles of the upper airways or the peripheral visceral organs. Both previous and present neuroanatomical studies with HRP revealed that some of the medullary inspiratory neurons, irrespective of their type, have collateral branches to the hypoglossal nucleus, the dorsal motor nucleus of the vagus, and the surrounding reticular formation.

0

0

O

A oAA A

Ot

a0

0

* 0

1

.

.

20

30

40

,

.

50

60

constant type) and an open triangle (for the neuron of decrementing type); and motoneurons with asterisks. The level indicated with an arrow in B corresponds to the rostral end of the RFN. R: rostral, C: caudal, M: medial, L: lateral.

Pm

major axes Fig. 13. The soma1 size of the stained neurons. The abscissa shows the major axes and the ordinate the minor axes. The symbols are the same as those used in Figure 12.

ACKNOWLEDGMEWTS we express Our thanks to Mrs. M. Ta@chi for her technical assistance. This work was Partly supported bY Grants-in-Aid for Scientific Research (Nos. 63570027,

INSPIRATORY NEURONS OF VRG AND BOTZINGER COMPLEX 01570080, 01770038) from the Japan Ministry of Education. Science and Culture.

Adams, J.C. (1981) Heavy metal intensification of DAB-based HRP reaction product. J. Histochem. Cytochem. 29:775. Arita, H., N. Kogo, and N. Koshiya (1987) Morphological and physiological properties of caudal medullary expiratory neurons of the cat. Brain Res. 401:258-266. Berger, A.J. (1981) Properties of medullary respiratory neurons. Fed. Proc. 40.2378-2383. Berman, A.L. (1968) The Brain Stem of the Cat. A Cytoarchitectonic Atlas With Stereotaxic Coordinates. Madison: University of Wisconsin Press. Bianchi, A.L. (1971) Localisation et etude des neurones respiratoires bulbaires. Mise en jeu antidromique par stimulation spinal ou vagale. J. Physiol. (Paris) 63:540. Bianchi, A.L. (1974) Modalite de decharge et proprietes anatomo-fonctionnelles des neurones respiratoires bulbaires. J. Physiol. (Paris) 68:555587. Bianchi, A.L., and W.M. St. John (1981) Pontile axonal projections of medullary respiratory neurons. Respir. Physiol. 45167-183, Bianchi, A.L., and J.C. Barillot (1982) Respiratory neurons in the region of the retrofacial nucleus: Pontile, medullary, spinal and vagal projections. Neurosci. Lett. 31.977-282. Bianchi, A.L., L. Grelot, S. Iscoe, and J.E. Remmers (1988) Electrophysiological properties of rostra1 medullary respiratory neurones in the cat: An intracellular study. J. Physiol. (Lond.) 407r293-310. Cohen, M.I. (1975) Phrenic and recurrent laryngeal discharge patterns and the Hering-Breuer reflex. Am. J. Physiol. 228:1489-1496. Cohen, M.I. (1979) Neurogenesis of respiratory rhythm in the mammal. Physiol. Rev. 59:1105-1173. Cohen, M.I., and J.L. Feldman (1984) Discharge properties of dorsal medullary inspiratory neurons: Relation to pulmonary afferents and phrenic efferent discharge. J. Neurophysiol. 51:753-776. Ezure, K., and M. Manabe (1986) An origin of inspiratory inhibition of expiratory neurons in the Botzinger complex. J. Physiol. SOC.Jpn. 48:384. Ezure, K., M. Manabe, and K. Otake (1989) Excitation and inhibition of medullary inspiratory neurons by two types of burst inspiratory neurons in the cat. Neurosci. Lett. 104r303-308. Ezure, K., M. Manabe, K. Otake, and H. Sasaki (1990) Synaptic connections from decrementing and constant inspiratory neurons in cat medulla. Neurosci. Res. [Suppl.] 11:521.

499

Ezure, K., M. Manabe, K. Otake, H. Sasaki, and H. Mannen (1986) The axonal projection of .two types of burst inspiratory neurons to the Botzinger complex (BOT). Neuroscience Res. [Supp1.]3rS79. Haber, E., K.W. Kohn, S.H. Ngai, D.A. Holaday, and S.C. Wang (1957) Localization of spontaneous respiratory neuronal activities in the medulla oblongata of the cat: A new location of the expiratory center. Am. J. Physiol. 190:350-355. Kalia, M.P. (1977) Neuroanatomical organization of the respiratory centers. Fed. Proc. 36.2405-2411. Kalia, M.P. (1981) Anatomical organization of central respiratory neurons. Annu. Rev. Physiol. 43:105-120. King, G.W. (1980) Topology of ascending brainstem projections to nucleus parabrachialis in the cat. J. Comp. Neurol. 191:615-638. Kreuter, F., D.W. Richter, H. Cameron, and R. Senekowitch (1977) Morphological and electorical description of medullary respiratory neurons of the cat. Pfliigers Arch. 3727-16. Merrill, E.G. (1972) Interactions between medullary respiratory neurones in the cat. J. Physiol. (Lond.)226t72P. Merrill, E.G. (1974) Finding a respiratory function for the medullary respiratory neurons. In R. Bellairs and E.G. Gray (eds): Essays on the Nervous System. Oxford: Clarendon, pp. 4 5 1 4 8 6 . Merrill, E.G. (1981) Where are the real respiratory neurons? Fed. Proc. 40:2389-2394. Nesland, R.S., and F. Plum (1965) Subtypes of medullary respiratory neurons. Exp. Neurol. 12337-348. Otake, K., H. Sasaki, H. Mannen, and K. Ezure (1987) Morphology of expiratory neurons of the Botzinger complex: an HRP study in the cat. J. Comp. Neurol. 258:565-579. Otake, K., H. Sasaki, K. Ezure, and M. Manabe (1989a) Morphological analysis of burst inspiratory neurons in cat medulla. ActaAnat. Nippon. 64:456. Otake, K., H. Sasaki, K. Ezure, and M. Manabe (198913) Axonal trajectory and terminal distribution of inspiratory neurons of the dorsal respiratory group in the cat's medulla. J. Comp. Neurol. 286.218-230. Richter, D.W. (1982)Generation and maintenance of the respiratoryrhythm. J. Exp. Biol. 100:93-103. Sasaki, H., K. Otake, H. Mannen, K. Ezure, and M. Manabe (1989) Morphology of augmenting inspiratory neurons of the ventral respiratory group in the cat. J Comp. Neurol. 282:157-168. Smith, J.C., D.E. Morrison, H.H. Ellenberger, M.R. Otto, and J.L. Feldman (1989) Brainstem projections to major respiratory neuron populations in the medulla of the cat. J. Comp. Neurol. 281r69-96. Taber, E. (1961) The cytoarchitecture of the brain stem of the cat. 1. Brain stem nuclei of cat. J. Comp. Neurol. 116.2749.