- Email: [email protected]
Discharge patterns of bulbo-pontine respiratory unit populations in cat
Brain Research, 114 (1976) 211-225 © Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
211
DISCHARGE PATTERNS OF BULBO-PONTINE RESPIRATORY UNIT POPULATIONS IN CAT
J. F. VIBERT, F. BERTRAND, M. DENAVIT-SAUBII~and A. HUGELIN Laboratoire de Physiologie, Facultd de Mddecine St. Antoine, 75571 Paris 12 and Laboratoire de Physiologie nerveuse, C.N.R.S., 91190 Gif-sur- Yvette (France) (Accepted February 10th, 1976)
SUMMARY
Respiration related units (RRU) were recorded during a stratigraphic exploration of medulla and pons from the cervical junction to the caudal part of the pneumotaxic system in the semi-chronic locally anesthetized 'isolated respiratory centre' of the cat. Metal 'low impedance, capacitance compensated' microelectrodes recorded multi-unit signals from which unitary discharges were discriminated and processed by computer; it is suggested that using these techniques, the sample was a good representation of the total unit population. The phase relation to phrenic discharge was determined on cycle triggered time histograms. Of 23,000 units, 28700 had a definite respiratory modulation. Examined individually, each RRU showed a stable discharge pattern corresponding to one of various respiratory types, the majority of which have been described previously. Both temporal and spatial distributions of RRU discharges were analyzed. Temporal distribution of peak firing frequencies (PFF) of 5,000 RRU sampled anatomically at random showed two main populations whose modes were observed during inspiration (I) or expiration (E). Troughs were observed in the histogram at the transition from I to E and E to I, thus indicating low probability for finding phase spanning RRU in the medulla and pons up to the pneumotaxic level. These statistical results turned out to be identical to those obtained with an a priori classification method comparable to that used in most of the previous works. In addition, the PFF distribution suggested that the E population could be further divided into 3 subpopulations whose modes fall in early, mid, and late expiration respectively. Comparison of RRU temporal distribution in two regions, one rostral, another caudal to a frontal Horsley-Clarke plane situated 3 mm in front of the obex, showed that, in the caudal region, 70% of the RRU were i units, while, in the rostral medulla and pons, equal proportions of I and E neurons were found. Temporal distribution of RRU peak frequencies was studied separately in anatomical structures where the probability of finding RRU was high. No clear cor-
212 respondence between R R U types and anatomy could be found, but marked differences between structures were observed, thus suggesting nevertheless a different spatial distribution for I and E populations.
INTRODUCTION Most of the recent attempts at understanding the origin of oscillation in respiratory neuronal networks have been based on the observation of unitary burst discharges showing several different types of phase relations relative to the respiratory cycle. The main difficulty in this field is to determine the number and limits of temporal populations. The first reports 1,2,5,6,17,22,2a,27 on respiration related units (RRU) described two bulbar discharge patterns: inspiratory (I) and expiratory (E), thus leading to the postulation of a bistable system organization 18,21,33,35,~7. Later works singled out various I and E subtypes3,4,24,z°,~2, raising the question of a bulbar multistable system 32. On the other hand, exploration of the pons showed individual R R U with a peak firing frequency (PFF) near the transition between either I and E or E and I phases that were called phase spanning inspiratory-expiratory (IE) or expiratory-inspiratory (EI) units 15. It has been further hypothesized that phase spanning neurons could promote phase switching in a bulbo-pontine multistable system12,a~,z°,36. The existence of a definite IE population was demonstrated at the level of the nucleus parabrachialis medialis 8 which supports the pneumotaxic oscillator 1° and it has been supposed that phase spanning EI neurons are necessary to start inspiration in the master bulbar oscillator13,15. Authors who explored the pontine reticular formation up to VIIth motor nucleus level were not able to firtd separate IE and EI populations3,4, 26. Waldron 4~, studying the distribution of onset and cessation of RRU bursts with a graphical method, observed two 'natural clusters' that correspond to I and E populations. There was no evidence for separate phase spanning populations though the diagram showed considerable variations in the pattern of activity, even to such an extent that some early firing I or E units fit the description of phase spanning neurons. Several reasons can be advanced to explain uncertainty concerning the number of R R U populations: (1) experimental conditions varied markedly, mainly with respect to the state of anesthesia; (2) in the majority of cases, the number of analyzed units was low and did not permit quantitative comparison; (3) highly selective electrodes were used which recorded large amplitude spikes only and it can be seriously questioned if this did not introduce a considerable bias in the sampling. Combining recent technical advances allows the two latter difficulties to be overcome. Improvement of microelectrode manufacturing has enabled the number of R R U recorded extracellularly at the same point to be increased, thus resulting in a multiunit signal3, ~°. At the same time, separation of spikes in multiunit recordings can be performed by computation. In the present series where both methods were used, several thousands of R R U discharges could be analyzed in the course of a stratigraphic exploration of the brain stem from the cervical junction up to the
213 pneumotaxic system in unanesthetized cats. This allowed a statistical analysis of the number of temporal populations in the bulbo-pontine respiratory network and the subsequent study of their spatial distribution. MATERIALSAND METHODS
Preparation Experiments were performed on 60 cats using an 'Isolated Respiratory Centre' preparation (bilateral vagotomy, spinal section at the C7 level). All surgical procedures were carried out under halothane anesthesia; 5~ procaine was injected into skin margins and pressure points. Animals were immobilized with gallamine triethiodide (20 mg/kg/h in a continuous perfusion of 5~ glucose in physiological saline at the rate of 4 ml/h); general anesthesia was then stopped. Recording sessions lasted 2-3 days; EEG and phrenic nerve discharge were monitored throughout the experiment; the occurrence of spindle bursts on the EEG and the regular succession of inspiratory discharges demonstrated that the animals were not suffering from pain. Every 12 h local anesthesia was supplemented. Animals were artificially ventilated; end-tidal F•eo2 was continually monitored and normocapnia (FAco2 = 4~) was maintained by adding CO2 to the inspired gas. Rectal temperature was regulated at 38 °C. Systolic pressure ranged from 130 to 90 mmHg. Animals were fixed with conventional ear-bars in a Horsley-Clarke apparatus. The animal's head was then tilted ventrally at an angle of 45 ° relative to the HorsleyClarke vertical plane by using an adaptor plate mounted on an adjustable 45 ° sliding block to lower the eye-bars. This procedure was particularly useful when exploring the rostral pons with several microelectrodes descending perpendicularly; but, due to medulla curvature, it modified markedly the correspondence between anatomical structures and cartesian coordinates of the Horsley-Clarke reference planes (see next paper, Fig. 1A). Surgical and histological techniques have been previously described8,1°.
Recording techniques Recording was carried out using tungsten microelectrodes. Tungsten wires were electrolytically reduced in diameter from 50 to 3-4 #m and were inserted into glass micropipettes whose tip was sealed by heating. The uninsulated part of the wire was then electrolytically polished a second time to a cone, 15/zm long with a 1-3/~m base diameter and coated with platinum blackS~,42. Microelectrode impedance was lower than 0.5 M/2 at I KC. The micropipette was shielded by spraying colloidal silver on its external surface, and isolated with varnish. To obtain a signal rise time better than 10 #sec, the input capacitance of the amplifier was lowered and the capacitance effect of glass and grounded conducting medium was 'neutralized' by maintaining the shield potential at the same level as that of the microelectrode through a positive feedback system16,25,~9. Using low impedance, capacitance compensated electrodes, most units had a signal-to-noise ratio ranging from 2/1 to 7/1 and could be recorded for 40--100/~m (average = 50/~m) along the microelectrode descent. On some occasions (8~ of
214
100"~i,,,,, ! ~
II
!!:
A g
I II
o . .
1'0
2'0
N
~r25
OV 0,05s
Fig. 1. Illustration of the spike recognition technique. Left: CRO display of the amplitude histogram obtained from 512 spikes. Ordinates: amplitude; abscissa: frequency. Horizontal lines indicate limits of 3 amplitude windows, A, B and C, chosen by the experimenter. Right: example of individual spikes (5 superimposed traces) selectedusing A, B and C amplitude windows. cases) the ratio was better than 7/1 and could reach 15/1: in these cases units could be recorded for up to 600 # m of the descent• In most cases at each recording point at least 3 different units could be recognized by visual inspection on the CRO.
Data processing Signal analysis was carried out using a PDP 12 computer operated on-line. After elimination of low frequencies up to 400 Hz using a high-pass digital filter, the first stage of spike recognition consisted of constructing a spike amplitude histogram from a sample of 512 units which was displayed on the computer CRO (Fig. 1). Several peaks were easily observed in histograms from records of most brain stem structures. It was then possible to adjust 3 windows giving the maximum amplitude range of the spikes to be kept for analysis. To avoid possible confusion between spikes, the windows were always narrow and centered on the 3 sharpest peaks. An example of 3 different families of spikes obtained by this amplitude selection is presented in Fig. 1. The spike recognition algorithm included additional elimination of short and long duration signals, thus excluding fiber discharges and compound spikes. When necessary, superimposed spikes were observed on a storage oscilloscope at a high speed sweep, making possible the discrimination of 2 spikes of identical amplitude but different shapes. In these rare cases, slightly moving the electrode permitted a clear separation of the 2 units on the usual amplitude basis. The spike recognition technique unavoidably introduced errors in sampling. According to our experience, spike discrimination based on the criterion of maximum
215 amplitude is reliable only when the amplitude window is narrow; this means that a large proportion (50-80%) of recorded spikes was rejected. It can be assumed that rejection of spikes was at random and would not therefore modify their observed discharge pattern relative to the cycle, though this led to an underestimation of firing frequency. Stability of spike pattern depended on 2 conditions: (1) rigidity of the electrode holder; (2) absence of changes in vascular bed and blood gas composition during recording. The duration of analysis of a sample was 15 min on average. Except in the cases of injured cells, there were few variations in the amplitude during the recording. During long experiments under steady conditions, the same set of units could be often recognized after several hours. Respiration modulation was determined using the cycle triggered time histogram (CTH) method, averaging 30 cycles, from which a respiration modulation index (RMI) was calculated. 0.70 was considered as the RM[ minimal value for which a unit could be considered as respiration related (full description in ref. 9). RESULTS The medulla and pons were explored with extracellular microelectrodes from the cervical junction up to the pneumotaxic complex (parabrachialis and K611iker-Fuse nuclei) in search for RRU. During the descent, every spontaneously active neuron was recorded and analyzed. Penetrations were made systematically every 0.5 mm in both caudo-rostral and latero-medial directions. In the vertical plane, the microelectrode was halted at the first point of the descent where a set of unit discharges could be recognized. After analysis, the descent was continued for 100 # m and from this point the search for neurons was resumed. Thus, sampling was performed regardless of the structures penetrated and it can be assumed that it was random relative to anatomy. In the course of microelectrode descents, numerous tonically firing units were recorded; their number was large in comparison to the figures reported after investigations carried out with highly selective electrodes in anesthetized cats. The number of units resolved over a 1 mm displacement of electrode tip ranged from 15 to 30 with an average value of 23. Assuming that a microelectrode explored a sphere with a radius of approximately 25 #m, it can be estimated that the unit density within cubes with sides of 100 # m (10 n cu.#m) ranged from 4.0 to 11.8 with an average value of 7.6. From 315 descents performed on 50 cats, 23,000 units were recorded and kept for analysis. Respiratory modulation was determined from CTH. Of the analysed cells, 28% were found to be RRU, i.e., having a RMI over 0.70. This proportion is very much less than the 47700found previously at the level of the upper ports 1°. Phasic RRU, i.e., units whose discharge stops firing during one part of the respiratory cycle, were found in 15% of cases only. They corresponded to the spikes of largest amplitude. They could be recorded anywhere in the medulla and pons, but predominated in the bulb up to a level 5 mm in front of the obex (Fig. 2).
216
VHt
12
~
J 7L
~
.KF
~
S
5M
I
I
I
I
..Q
501
o
0 4
2
0 - mm+
2
4
6
8
10
Fig. 2. Distribution of phasic R R U in the medulla. Abscissa: rostro-caudal position of the units relative to the obex. Ordinate: percentage of phasic units relative to RRU. Horizontal lines in the upper part indicate the relative level of the main respiration modulated structures. Abbreviations: 5M, motor trigeminal nucleus; 7L, facial nucleus, lateral division; 12, hypoglossal nucleus; KF, K611iker-Fuse nucleus; S, solitary tract; VH, ventral horn of the first cervical segment. TABLE I
Respiratory characteristics of bulbo-pontflte respiratory structures. UD, mean unit density per 106 cu./tm; R R U D , mean R R U density par l06 cu.#m; R R U RRUD
~ =
× 100; RMI, mean respiratory modulation indexL The separation between 3 categories A,
UD B and C results from a variance analysis of the RMI. Out of the 35 explored structures, 25 fall in the C category; data from some of them only are presented in the table.
Structures
A Motor trigeminal nucleus Hypoglossal nucleus Solitary complex Ventral horn Facial nucleus, lateral division Sensory trigeminal complex
Number of punctures
Number of units
UD
RR U% RMI ± SD log R M I
10 31 15 11 16 29
234 681 253 183 312 698
6.9 6.3 5.6 7.3 6.5 7.8
71.6 63.5 63.2 50.3 48.0 43.7
1.10 1.87 1.33 1.30 1.36 1.05
± ± ± 4± ±
1.04 1.51 1.16 1.28 1.21 1.21
122 24
4607 706
7.5 6.9
37.9 36.7
0.91 ± 1.17 0.84 ± 0.98
14 27
183 1074
10.0 7.2
34.9 32.4
0.77 ± 1.07 0.79 ± 0.97
5 25 13 19 19 35
281 408 161 288 326 569
11.8 5.3 6.8 9.4 7.3 7.3
27.1 22.0 17.3 15.2 12.3 10.4
0.48 0.48 0.67 0.45 0.51 0.50
B
Lateraltegmental field Gigantocellular tegmental field Lateral reticular nucleus of the medulla, internal division Magnocellular tegmental field C Medial longitudinal bundle Lateral vestibular nucleus Trapezoid body Inferior olive Spinal trigeminal tract Cuneate nucleus
± ± ± ± ± ±
0.60 0.88 1.01 0.76 0.89 0.83
217
R R U density In 35 anatomical structures which were explored at least 5 times, 3 parameters were defined: (1) total unit density; (2) R R U percentage relative to the total number of recorded cells and (3) mean respiratory modulation index (RMI) which was calculated for all recorded units (RRU as well as non-respiratory). Results (Table I) show 3 categories of structures. In the first (A), R R U percentage was over 43~ and R M I over 1.4; in the second group (B), R R U percentage ranged between 38~ and 32~ and RMI between 1.0 and 0.70; in the last category (C), R R U percentage was less than 28~ and RMI less than 0.70. Comparison of samples from each anatomical structure made variance analysis possible. It was performed on the RMI; the t-test was used to compare pairs of structures. Data from each of the 35 structures were tested individually against all the 34 others. Results confirmed the existenCe of 3 categories of structures by showing that: within a given group, RMI of each structure was significantly different from that of any structure of the two other categories (P < 0.001); within each category, there was no significant difference between structures. Comparable results were obtained for R R U percentages. Table I shows that the higher respiratory modulation is found in motor and sensory cranial nuclei that are known to be involved in the control of the respiratory accessory musculature, these being the motor nuclei of trigeminal, facial, vagus and hypoglossal nerves and ventral horn of the first cervical segment, and the sensory nuclei of alaminar spinal trigeminal sensory nucleus and solitary complex. It should be noted that a higher R R U percentage and R M I were found in the solitary complex which comprises the infrasolitarius nucleus (dorsal inspiratory nucleus), where recent studies have emphasized the presence of RRUll, 19. The fact that R R U were recorded in vagotomized animals confirms the inherent respiratory rhythmicity of the solitary complexT,19. In the B category fall reticular structures which have always been thought to support the control of respiratory oscillation. These are the lateral, giganto-cellular, magno-cellular and paramedian tegmental fields. The unit density in tegmental fields is comparable to that of other parts of the medulla and pons while the R R U percentage is significantly lower than in the respiratory modulated sensori-motor nuclei. The C category contains structures never before suspected of taking part in respiratory activity, even though definite R R U were observed in all of them. Included in this category are the medial longitudinal bundle, the vestibular and abducens nuclei, the lateral lemniscus, the ventral cochlear nucleus, the inferior and superior olivary nuclei and the trapezoid body; the spinal trigeminal tract, the cuneatus and gracilis nuclei and medial lemniscus and the restiform body, etc. RR U classification Examination of a large number of R R U discharges suggested that the firing patterns corresponded in most of cases to inspiratory (I) and expiratory (E) types. Within the latter category, 3 subtypes were often observed, early, mid and late. Examples of these are given in Fig. 3 for cases of either tonic or phasic activity.
218
I
eE
mE
(t)E
Fig. 3. Example of cycle triggered time histograms (CTH) corresponding to the 4 most common types of R R U discharge patterns. I, inspiratory; eE, early expiratory; mE, mid expiratory; (1)E, late expiratory. Black and white bars under each CTH indicate inspiration and expiration respectively. Duration of bins, 160 msec; number of cycles, 30.
In order: to compare our results with those of previous authors, to eliminate personal bias and to classify a large number of individual units, it was necessary to develop an algorithm for computer classification. To allow recognition of the most commonly described types, the cycle was divided into 6 parts. Owing to the usual 1:2 ratio between I and E durations, I was divided into 2 parts and E into 4, thus resulting into I and E sections of approximately the same width. The average frequency in each one was calculated from the CTH. A R R U discharge was identified as belonging to a given class when the highest average frequencies were found in two neighboring temporal sections (i.e., predominated in one-third of the cycle). Six respiratory classes could be recognized (Fig. 4A) which corresponded to types described by other authors3,4,15,32: phase spanning expiratory-inspiratory ('EI'), inspiratory (T), phase spanning inspiratory-expiratory ('IE'), early expiratory ('eE'), mid expiratory ('mE') and late expiratory ('(1)E'). CTH from 6,440 R R U were processed and classified. I units represented 47.2~ of the total respiratory population; 34.9~ of the R R U were E, of which the eE pattern comprised 18.2~, the mE 10.6~ and the (1)E 6.1~. For the phase spanning unit 5.9~o were of the EI type and 7.7~ of the IE type. Results are summarized in the diagram in Fig. 4B which indicates that the probability of observing phase spanning between bulb and upper pons is low. s
v~/J,'/,I
[
I
1
I
K/IrlA I
I
I 1 I
I
,
t
I
K/,r/j
EI
I
I
I
I
I
I
1
I
I
I
r,/,4/A
t
I
t
rl.~
40.
N = 6440
I
,
l//V/~
I
I
I
I/All1
I
'1 ,
%
I
I
I
,
t
•
IF
20.
eE 1~ m~
7eE
I
2 ' 3t
l
5
El
I(L)E
v/x/.a
4
i~
7mE
6' I
I
I 1
Phr
H
~ 2
I
3
,
4
(L)E I
I 5
I 6 ,
E
A
B
Fig. 4. A: Schematic illustration of the principle of the algorithm developed for automated classification. A unit is recognized as pertaining to one of the 6 classes (see text) when the highest average frequencies (hatched areas) fall into two neighboring sections. B: Diagramatic representation of R R U percentage in each of the 6 classes. Black and white bars indicate inspiration and expiration respectively.
219 N
~NxI° 1 N 10
900.
I I I i t 5
9.
00
s eSSe
_F
/ o S° toS t si
5
g t
tI
tI
o
:' J
I
E
I
E
Fig. 5. Temporal distribution of peak firing frequencies (PFF). Left: PFF histogram; abscissa, respiratory cycle. Circles indicate position of the modes of the 3 putative E sub-populations. Right: probability distribution function. The curves were obtained by plotting the cumulative distribution of left histogram on normal probability paper. Dotted curves were drawn by approximation. The remaining 4.3~ units were definitely respiration modulated, but their discharge pattern did not correspond to previous descriptions at the bulbar or pontine level.
Peak firing frequency distribution The main drawback of such a classification is that it is based on a priori or intuitive grounds. It is assumed that individuals in each class pertain to a 'natural' system. The existence of such a system can be verified if its classes fit with statistical populations. In a preceding paper 1°, it was shown that the grouping of R R U into several subpopulations could be better analyzed by comparing the temporal position of C T H peaks or peak firing frequencies (PFF) while increasing the number of temporal classes. In the present work, P F F were distributed into 15 temporal classes, 5 in I and 10 in E in order to respect the I / E duration ratio. R R U were sampled anatomically at random. Calculations could be carried out on a sample of 5,000 R R U . That the sample was representative of the population is indicated by the fact that above 3,000 units, increasing the sample size did not modify the shape of the histogram. Results shown in Fig. 5A clearly indicate one mode in inspiration. In expiration, the temporal distribution of P F F is not so well defined and
220
%
20. N=664 10_ 1%N'95
i
A ~
N=216
10~1
N=229
E
D 20
C
B
,•N:484 F
i
.
_ N=1793
N=1235
i
G H I Fig. 6. Temporal distribution of PFF in respiratory structures. Examples for individual structures : motor nuclei of the Vth (A) and XIIth (B); sensory nuclei of the trigeminal complex (D) and of the solitary complex (E); lateral (G) and giganto-cellular (H) tegmental fields. Pooled histograms for motor (C), sensory (F) and reticular (I) structures. Ordinates, number of samples in each class as a percentage of RRU of each structure. Abscissae,as in Fig. 5. the existence of 3 expiratory modes, indicated by circles in Fig. 5, is perhaps possible. Whereas statistical methods exist to test population homogeneity, heterogeneity tests are lacking. In the present study, an attempt to ascertain E population heterogeneity was made by studying the probability distribution function of the sample. Curves were obtained using a graphic method consisting of plotting the cumulative distribution on normal probability paper. Results suggested the existence of 4 populations having different variances as indicated by the different slopes of the 4 segments of the curve illustrated in Fig. 5 right. It follows that firing patterns of the R R U of the explored regions could fall into 4 sub-populations: [, eE, mE and (1)E. Though it was not possible to ascertain this hypothesis statistically, the existence of 3 E categories will be taken into consideration in the following paper where localization of neurons according to their respiratory pattern will be attempted.
Regional analysis In searching for separate functions of the various parts of respiratory structures,
221 the temporal distribution of P F F was studied in anatomical regions where R M I was over 0.70. P F F histograms were constructed separately for each one, and then pooled according to their physiological function. Examples of separate and pooled PFF histograms are shown in Fig. 6 from R R U recorded in individual motor nuclei (Fig. 6A-C), sensory nuclei (Fig. 6 D - F ) and reticular tegmental fields (Fig. 6G-I). The pooled histogram for motor nuclei (XII, X, VII, V, first segment of the ventral horn) shows that most R R U fall into the I category (Fig. 6C). The two modes appearing in early and late expiration correspond mainly to the existence of large populations of eE in the Vth nucleus (Fig. 6A) and (1)E in the ventral horn. In sensory nuclei (V and X) the 4 populations are represented, since a definite mode appeared in mid expiration. In the tegmental fields, the pooled histogram (Fig. 6I) is very similar to that of the total histogram (Fig. 5). Comparison of individual histograms from lateral and giganto-cellular tegmental fields showed, however, that I R R U predominate in the lateral tegmental field, while E units were more numerous medially. These differing temporal distributions thus suggest the possibility of a spatial organization of R R U , but no clear correspondence with anatomy can be detected. An attempt was also made to compare the P F F distribution in caudal and rostral parts of the medulla on each side of a plane transecting the brain stem 3 m m in front of the obex. Striking differences then appeared in the [ and E ratios both in the case of
~
10-
7
~
_N=1315
i
A
C
B
A
20_
N=1231
625
N =1856
10-
D
E
F
Fig. 7. Comparison of temporal distributions of PFF in caudal and rostral parts of bulbo-pontine region. The plane of separation transects the brain stem 3 mm in front of the obex. Upper section, pooled histograms of rostral structures. A: trigeminal complex, VIIth lateral nucleus, solitary complex. B: lateral, giganto-cellular and magno-cellular tegmental fields. C: total rostral RRU population. Lower section, pooled histograms of caudal structures. D: solitary complex, XIIth nucleus, ventral horn of the first cervical segment. E: lateral tegmental field, paramedian reticular nucleus. F: total caudal RRU population. Coordinates as in Fig. 6. Note that the mode in I is later in the rostral than in the caudal reticular formation. Note that the inspiratory mode is significantly earlier in sensory motor nuclei and caudal reticular formation when compared to the rostral reticular formation.
222 pooled sensory-motor and reticular histograms, as well as in the total rostral and caudal populations (Fig. 7). I represents two thirds of R R U in the bulb whereas an equal proportion o f [ and E is found rostrally. DISCUSSION The results reported in this paper have to be considered from two aspects: (1) is the unit sample representative of the whole neuronal population of the explored area? (2) What is the number of the temporal R R U populations shown to exist ? Improvement of recording techniques during the past two decades has increased progressively the number of R R U analyzed in brain stem exploration. Baumgarten 5, using 25 # m wire electrodes recorded 23 R R U in 900 penetrations, i.e., 0.025 R R U in each descent on the average. Nelson al reported 0.5 R R U in each penetration performed with extracellular micropipettes with 2 # m tips. Waldron 41 used tungsten wire in glass micropipettes with tips 3 # m in diameter and from the reported data, it can be calculated that about 2 R R U were recorded in each penetration. Lowering the impedance of the same type of electrode with platinum black was shown to increase markedly the number of units recorded at the same point, thus resulting in a multiunit signal 3°. In the present study, where signal filtering was avoided using capacitancecompensated electrodes, an average of 23 R R U could be recorded in each microelectrode descent. Two other technical points also seemed important in explaining the magnitude of the increase in number of units recorded: (1) avoidance of barbiturate anesthesia which has been reported to reduce the number of firing R R U in the reticular fields2s,29, (2) the possibility of recording in the medulla without surgical ablation of the cerebellum which drastically modified the histological appearance of reticular neurons 4° and (3) analysis of units firing tonically. Since spikes could be distinguished roughly every 50 # m during the microelectrode descent, it can be supposed that the electrode tip detected activity over a 25 # m radius. Using this assumption, it can be calculated that the number of firing neurons in a 106 cu.#m volume ranged from 4 to 11 in the medulla and from 5 to 10 in the R R U containing structures (Table [). These figures can be compared to those obtained with histological methods. Scheibel and Scheibe188 calculated from brain stem sections stained with the Golgi method that there may be 15-30 cells per 10n cu.#m in the magno-cellular portion of the medullary tegemental fields, 20-50 in parvocellular portions, and 40 to 60 in more tightly packed reticular groups. The average cell density was estimated at 33 per 106 cu.#m. These values were obtained in 10-day-old kittens; cell density is less for the adult, where brain size is greater though the total number of cells remains the same. We have calculated that to make a rough comparison of densities in the newborn and adult cat, it is necessary to multiply the volume by a factor of 10-15. On the other hand, in the Golgi preparation less than 4 0 ~ of cells are supposed to be silver impregnated. These considerations indicate that, assuming a unit volume of 10n cu.#m, the number of neurons in an adult cat's medulla could range from 2 to 10 with an
223 average value from 4 to 7. Considering the large approximation unavoidably introduced in calculations, it can be nevertheless considered that figures obtained with histological and electrophysiological methods are reasonably concordant and that the sampling carried out with low impedance, capacitance compensated microelectrodes was representative of the total population of spontaneously active neurons. The second point concerns the number of temporal sub-populations of R R U . As it was recalled in the introduction, many authors have attempted previously to classify medullary and pontine R R U according to the phase relation of bursts to the efferent respiratory motor discharges. Unfortunately, due to the limited number of units analyzed during single cell recording, the size of samples remained small in most studies. In the present work, the R R U sample turned out to be large enough to provide statistical information. It has been confirmed that medullary and pontine R R U fall into 2 broad categories [ and E. Results suggest that the E population can be further divided into 3 sub-populations: early, mid and late. These conclusions are in agreement with those of recent studies3,4,11,3°,~2. On the other hand, no evidence could be provided to support the existence of definite phase spanning populations in the brain stem caudal to the pneumotaxic system. This study finally failed to answer the important question of the number of oscillators present in the medulla. Are the R R U sub-populations pertaining to a single system, or are they parts of several oscillators? Regional analysis was not able to give any further indication about this, except that there were consistent differences in R R U temporal distribution according to anatomical structures. An entirely different approach towards solving this problem is reported in the next paper. ACKNOWLEDGEMENT This work was supported by grants from the Centre national de la Recherche scientifique (LA 204) and the D616gation g6n6rale ~ la Recherche scientifique et technique (72-7-0740).
REFERENCES 1 Achard, O. et Bucher, V. M., Courants d'action bulbaires ~t rythme respiratoire, Helv. physioL pharmacoL Acta, 12 (1954) 265-283. 2 Amoroso, E. C., Bell, F. R. and Rosenberg, H., The localization of respiratory regions in the rhombencephalon of the sheep, Proc. roy. Soc. B, 139 (1951) 128-140. 3 Batsel, H. L., Localization of bulbar respiratory center by microelectrode sounding, Exp. Neurol., 9 (1964) 410426. 4 Batsel, H. L., Some functional properties of bulbar respiratory units, Exp. NeuroL, 11 (1965) 341-366. 5 Baumgarten, R. Von, Koordinationsformen einzelner Ganglienzellen der rhombencephalen Atemzentren, Pfliigers Arch. ges. PhysioL, 262 (1956) 573-594. 6 Baumgarten, R. Von, Balthasar, K. und Koepchen, H. P., Ober ein Substrat atmungsrhythmischer Erregungsbildung im Rautenhirn der Katze, Pfliigers Arch. ges. PhysioL, 270 (1960) 504-528. 7 Baumgarten, R. Von and Kanzow, E., The interaction of two types of inspiratory neurons in the region of the tractus solitarius of the cat, Arch. ital. BioL, 96 (1958) 361-373.
224 8 Bertrand, F. and Hugelin, A., Respiratory synchronizing function of nucleus parabrachialis medialis: pneumotaxic mechanisms, J. NeurophysioL, 34 (1971) 189-207. 9 Bertrand, F., Hugelin, A. and Vibert, J. F., Quantitative study of anatomical distribution of respiration related neurons in the pons, Exp. Brain Res., 16 (1973) 383-399. 10 Bertrand, F., Hugelin, A. and Vibert, J. F., A stereologic model of pneumotaxic oscillator based on spatial and temporal distributions of neuronal bursts, J. Neurophysiol., 37 (1974) 91-107. 11 Bianchi, A. L., Localisation et 6tude des neurones respiratoires bulbaires. Mise en jeu antidromique par stimulation spinale ou vagale, J. Physiol. (Paris), 63 (1971) 5-40. 12 Cohen, M. I., Discharge patterns of brain stem respiratory neurons in relation to carbon dioxide tension, J. Neurophysiol., 31 (1968) 142-165. 13 Cohen, M. I., Discharge patterns of brain-stem respiratory neurons during Hering-Breuer reflex evoked by lung inflation, J. Neurophysiol., 32 (1969) 356-374. 14 Cohen, M. I., How respiratory rhythm originates : evidence from discharge patterns of brainstem respiratory neurones. In R. Porter (Ed.), Breathing: Hering-Breuer Centenary Symposium, Churchill, London, 1970, pp. 125-150. 15 Cohen, M. I. and Wang, S. C., Respiratory neuronal activity in pons of cat, J. NeurophysioL, 22 (1959) 33-50. 16 Deville, J., Un syst6me ~t hautes performances pour l'analyse intra-cellulaire, Eleetroenceph. elin. Neurophysiok, 40 (1976) 521-528. 17 Dirken, M. N. J. and Woldring, S., Unit activity in bulbar respiratory centre, J. Neurophysiol., 14 (1951) 211-225. 18 Duffin, J., The functional organization of respiratory neurones. A review, Canad. Anaesth. Soc. J., 19 (1972) 1-7. 19 Euler, C. von, Hayward, J. N., Marttila, I. and Wyman, R. J., Respiratory neurones of the ventrolateral nucleus of the solitary tract of cat: vagal input, spinal connections and morphological identification, Brain Research, 61 (1973) 1-22. 20 Feldman, J. L. and Cowan, J. D., Large-scale activity in neural nets. II. A model for the brainstem respiratory oscillator, Biol. Cybern., 17 (1975) 39-51. 21 Geman, S. and Miller, M., A mathematical model of the respiratory centers, The Physiologist, 16 (1973) 320. 22 Gesell, R., Bricker, J. and Magee, C., Structural and functional organization of the central mechanism controlling breathing, Amer. J. Physiol., 117 (1936) 423452. 23 Gesell, R., Magee, C. and Bricker, J., Activity patterns of the respiratory neurons and muscles, Amer. J. Physiol., 128 (1940) 615-628. 24 Haber, E., Kohn, K. W., Ngai, S. H., Holaday, D. A. and Wang, S. C., Localization of spontaneous respiratory neuronal activities in the medulla oblongata of the cat: a new location of the expiratory center, Amer. J. Physiol., 190 (1957) 350-355. 25 Holmer, N. G., An electrometer amplifier for electrophysiology with low input capacitance, Med. biol. Engineer., 8 (1970) 509-511. 26 Hukuhara, T. Jr., Neuronal organization of the central respiratory mechanisms in the brain stem of the cat, Acta Neurobiol. exp., 33 (1973) 219-244. 27 Hukuhara, T., Nakayama, S. and Okada, H., Action potentials in the normal respiratory centers and its centrifugal pathways in the medulla oblongata and spinal cord, Jap. J. Physiol., 4 (1954) 145153. 28 Hukuhara, T. Jr., Saji, Y., Kumadaki, N., Kojima, H., Tamaki, H., Takeda, R., und Sakai F., Die Lokalisation von Atemsynehron entladenden Neuronen in der retikul~iren Formation des Hirnstammes der Katze unter verschiedenen experimentellen Bedingungen, Naunyn-Schmiedebergs Arch. exp. Pathol. Pharmak., 263 (1969) 462-484. 29 Karczewski, W., Contemporary views concerning central control of respiration, Acta physiol. polon., 21, Suppl. 1 (1970) 61-80. 30 Merrill, E. G., The lateral respiratory neurones of the medulla : their associations with nucleus ambiguus, nucleus retroambigualis, the spinal accessory nucleus and the spinal cord, Brain Research, 24 (1970) 11-28. 31 Nelson, J. R., Single unit activity in medullary respiratory centers of cat, J. Neurophysiol., 22 (1959) 590-598. 32 Nesland R. and Plum, F., Subtypes of medullary respiratory neurons, Exp. Neurol., 12 (1965) 337-348. 33 Palmay, F. v., Davison, E. J. and Duffin, J., The simulation of multineurone networks: modelling
225
34 35 36 37 38
39 40 41 42
of the lateral inhibition of the eye and the generation of respiratory rhythm, Bull. math. Biol., 36 (1974) 77-89. Robinson, D. A., The electrical properties of metal microelectrodes, Proc. LE.E.E. reed. Engineer., 56 (1968) 1005-1071. Rubio, J. E., A mathematical model of the respiratory center, Bull. math. Biophys., 29 (1967) 719-736. Rubio, J. E., A new mathematical model of the respiratory center, Bull. math. Biophys., 34 (1972) 467-481. Salmoiraghi, G. C. and Burns, B. D., Notes on mechanism of rhythmic respiration, J. NeurophysioL, 23 (1960) 14-26. Scheibel, M. E. and Scheibel, A. B., Structural substrates for integrative patterns in the brain stem reticular core. In H. H. Jasper, L. D. Proctor, R. S. Knighton, W. C. Noshay and R. T. Costello (Eds.) Reticular Formation of the Brain, Little, Brown, Boston Mass., 1958, pp. 31-55. Strickholm, A. and Winston, S., An electrometer amplifier for electrophysiology, Med. biol. Engineer., 7 (1969) 99-102. Tyc-Dumont, S., Histopathological alterations of brain stem neurons provoked by cerebellar ablation, in preparation. Waldron, I., Activity patterns in respiratory muscles and in respiratory neurones of the rostral medulla of the cat, J. Physiol. (Lond.), 208 (1970) 373-383. Wolbarsht, M. L. and Wagner, H. G., Glass insulated platinum microelectrode design and fabrication. In F. Bostein (Ed.), Medical Electronics (Proceedings of the 5th Intern. Conf on Med, Electronics, Likge, July 22-26, 1963), 1963, pp. 510-515.
211
DISCHARGE PATTERNS OF BULBO-PONTINE RESPIRATORY UNIT POPULATIONS IN CAT
J. F. VIBERT, F. BERTRAND, M. DENAVIT-SAUBII~and A. HUGELIN Laboratoire de Physiologie, Facultd de Mddecine St. Antoine, 75571 Paris 12 and Laboratoire de Physiologie nerveuse, C.N.R.S., 91190 Gif-sur- Yvette (France) (Accepted February 10th, 1976)
SUMMARY
Respiration related units (RRU) were recorded during a stratigraphic exploration of medulla and pons from the cervical junction to the caudal part of the pneumotaxic system in the semi-chronic locally anesthetized 'isolated respiratory centre' of the cat. Metal 'low impedance, capacitance compensated' microelectrodes recorded multi-unit signals from which unitary discharges were discriminated and processed by computer; it is suggested that using these techniques, the sample was a good representation of the total unit population. The phase relation to phrenic discharge was determined on cycle triggered time histograms. Of 23,000 units, 28700 had a definite respiratory modulation. Examined individually, each RRU showed a stable discharge pattern corresponding to one of various respiratory types, the majority of which have been described previously. Both temporal and spatial distributions of RRU discharges were analyzed. Temporal distribution of peak firing frequencies (PFF) of 5,000 RRU sampled anatomically at random showed two main populations whose modes were observed during inspiration (I) or expiration (E). Troughs were observed in the histogram at the transition from I to E and E to I, thus indicating low probability for finding phase spanning RRU in the medulla and pons up to the pneumotaxic level. These statistical results turned out to be identical to those obtained with an a priori classification method comparable to that used in most of the previous works. In addition, the PFF distribution suggested that the E population could be further divided into 3 subpopulations whose modes fall in early, mid, and late expiration respectively. Comparison of RRU temporal distribution in two regions, one rostral, another caudal to a frontal Horsley-Clarke plane situated 3 mm in front of the obex, showed that, in the caudal region, 70% of the RRU were i units, while, in the rostral medulla and pons, equal proportions of I and E neurons were found. Temporal distribution of RRU peak frequencies was studied separately in anatomical structures where the probability of finding RRU was high. No clear cor-
212 respondence between R R U types and anatomy could be found, but marked differences between structures were observed, thus suggesting nevertheless a different spatial distribution for I and E populations.
INTRODUCTION Most of the recent attempts at understanding the origin of oscillation in respiratory neuronal networks have been based on the observation of unitary burst discharges showing several different types of phase relations relative to the respiratory cycle. The main difficulty in this field is to determine the number and limits of temporal populations. The first reports 1,2,5,6,17,22,2a,27 on respiration related units (RRU) described two bulbar discharge patterns: inspiratory (I) and expiratory (E), thus leading to the postulation of a bistable system organization 18,21,33,35,~7. Later works singled out various I and E subtypes3,4,24,z°,~2, raising the question of a bulbar multistable system 32. On the other hand, exploration of the pons showed individual R R U with a peak firing frequency (PFF) near the transition between either I and E or E and I phases that were called phase spanning inspiratory-expiratory (IE) or expiratory-inspiratory (EI) units 15. It has been further hypothesized that phase spanning neurons could promote phase switching in a bulbo-pontine multistable system12,a~,z°,36. The existence of a definite IE population was demonstrated at the level of the nucleus parabrachialis medialis 8 which supports the pneumotaxic oscillator 1° and it has been supposed that phase spanning EI neurons are necessary to start inspiration in the master bulbar oscillator13,15. Authors who explored the pontine reticular formation up to VIIth motor nucleus level were not able to firtd separate IE and EI populations3,4, 26. Waldron 4~, studying the distribution of onset and cessation of RRU bursts with a graphical method, observed two 'natural clusters' that correspond to I and E populations. There was no evidence for separate phase spanning populations though the diagram showed considerable variations in the pattern of activity, even to such an extent that some early firing I or E units fit the description of phase spanning neurons. Several reasons can be advanced to explain uncertainty concerning the number of R R U populations: (1) experimental conditions varied markedly, mainly with respect to the state of anesthesia; (2) in the majority of cases, the number of analyzed units was low and did not permit quantitative comparison; (3) highly selective electrodes were used which recorded large amplitude spikes only and it can be seriously questioned if this did not introduce a considerable bias in the sampling. Combining recent technical advances allows the two latter difficulties to be overcome. Improvement of microelectrode manufacturing has enabled the number of R R U recorded extracellularly at the same point to be increased, thus resulting in a multiunit signal3, ~°. At the same time, separation of spikes in multiunit recordings can be performed by computation. In the present series where both methods were used, several thousands of R R U discharges could be analyzed in the course of a stratigraphic exploration of the brain stem from the cervical junction up to the
213 pneumotaxic system in unanesthetized cats. This allowed a statistical analysis of the number of temporal populations in the bulbo-pontine respiratory network and the subsequent study of their spatial distribution. MATERIALSAND METHODS
Preparation Experiments were performed on 60 cats using an 'Isolated Respiratory Centre' preparation (bilateral vagotomy, spinal section at the C7 level). All surgical procedures were carried out under halothane anesthesia; 5~ procaine was injected into skin margins and pressure points. Animals were immobilized with gallamine triethiodide (20 mg/kg/h in a continuous perfusion of 5~ glucose in physiological saline at the rate of 4 ml/h); general anesthesia was then stopped. Recording sessions lasted 2-3 days; EEG and phrenic nerve discharge were monitored throughout the experiment; the occurrence of spindle bursts on the EEG and the regular succession of inspiratory discharges demonstrated that the animals were not suffering from pain. Every 12 h local anesthesia was supplemented. Animals were artificially ventilated; end-tidal F•eo2 was continually monitored and normocapnia (FAco2 = 4~) was maintained by adding CO2 to the inspired gas. Rectal temperature was regulated at 38 °C. Systolic pressure ranged from 130 to 90 mmHg. Animals were fixed with conventional ear-bars in a Horsley-Clarke apparatus. The animal's head was then tilted ventrally at an angle of 45 ° relative to the HorsleyClarke vertical plane by using an adaptor plate mounted on an adjustable 45 ° sliding block to lower the eye-bars. This procedure was particularly useful when exploring the rostral pons with several microelectrodes descending perpendicularly; but, due to medulla curvature, it modified markedly the correspondence between anatomical structures and cartesian coordinates of the Horsley-Clarke reference planes (see next paper, Fig. 1A). Surgical and histological techniques have been previously described8,1°.
Recording techniques Recording was carried out using tungsten microelectrodes. Tungsten wires were electrolytically reduced in diameter from 50 to 3-4 #m and were inserted into glass micropipettes whose tip was sealed by heating. The uninsulated part of the wire was then electrolytically polished a second time to a cone, 15/zm long with a 1-3/~m base diameter and coated with platinum blackS~,42. Microelectrode impedance was lower than 0.5 M/2 at I KC. The micropipette was shielded by spraying colloidal silver on its external surface, and isolated with varnish. To obtain a signal rise time better than 10 #sec, the input capacitance of the amplifier was lowered and the capacitance effect of glass and grounded conducting medium was 'neutralized' by maintaining the shield potential at the same level as that of the microelectrode through a positive feedback system16,25,~9. Using low impedance, capacitance compensated electrodes, most units had a signal-to-noise ratio ranging from 2/1 to 7/1 and could be recorded for 40--100/~m (average = 50/~m) along the microelectrode descent. On some occasions (8~ of
214
100"~i,,,,, ! ~
II
!!:
A g
I II
o . .
1'0
2'0
N
~r25
OV 0,05s
Fig. 1. Illustration of the spike recognition technique. Left: CRO display of the amplitude histogram obtained from 512 spikes. Ordinates: amplitude; abscissa: frequency. Horizontal lines indicate limits of 3 amplitude windows, A, B and C, chosen by the experimenter. Right: example of individual spikes (5 superimposed traces) selectedusing A, B and C amplitude windows. cases) the ratio was better than 7/1 and could reach 15/1: in these cases units could be recorded for up to 600 # m of the descent• In most cases at each recording point at least 3 different units could be recognized by visual inspection on the CRO.
Data processing Signal analysis was carried out using a PDP 12 computer operated on-line. After elimination of low frequencies up to 400 Hz using a high-pass digital filter, the first stage of spike recognition consisted of constructing a spike amplitude histogram from a sample of 512 units which was displayed on the computer CRO (Fig. 1). Several peaks were easily observed in histograms from records of most brain stem structures. It was then possible to adjust 3 windows giving the maximum amplitude range of the spikes to be kept for analysis. To avoid possible confusion between spikes, the windows were always narrow and centered on the 3 sharpest peaks. An example of 3 different families of spikes obtained by this amplitude selection is presented in Fig. 1. The spike recognition algorithm included additional elimination of short and long duration signals, thus excluding fiber discharges and compound spikes. When necessary, superimposed spikes were observed on a storage oscilloscope at a high speed sweep, making possible the discrimination of 2 spikes of identical amplitude but different shapes. In these rare cases, slightly moving the electrode permitted a clear separation of the 2 units on the usual amplitude basis. The spike recognition technique unavoidably introduced errors in sampling. According to our experience, spike discrimination based on the criterion of maximum
215 amplitude is reliable only when the amplitude window is narrow; this means that a large proportion (50-80%) of recorded spikes was rejected. It can be assumed that rejection of spikes was at random and would not therefore modify their observed discharge pattern relative to the cycle, though this led to an underestimation of firing frequency. Stability of spike pattern depended on 2 conditions: (1) rigidity of the electrode holder; (2) absence of changes in vascular bed and blood gas composition during recording. The duration of analysis of a sample was 15 min on average. Except in the cases of injured cells, there were few variations in the amplitude during the recording. During long experiments under steady conditions, the same set of units could be often recognized after several hours. Respiration modulation was determined using the cycle triggered time histogram (CTH) method, averaging 30 cycles, from which a respiration modulation index (RMI) was calculated. 0.70 was considered as the RM[ minimal value for which a unit could be considered as respiration related (full description in ref. 9). RESULTS The medulla and pons were explored with extracellular microelectrodes from the cervical junction up to the pneumotaxic complex (parabrachialis and K611iker-Fuse nuclei) in search for RRU. During the descent, every spontaneously active neuron was recorded and analyzed. Penetrations were made systematically every 0.5 mm in both caudo-rostral and latero-medial directions. In the vertical plane, the microelectrode was halted at the first point of the descent where a set of unit discharges could be recognized. After analysis, the descent was continued for 100 # m and from this point the search for neurons was resumed. Thus, sampling was performed regardless of the structures penetrated and it can be assumed that it was random relative to anatomy. In the course of microelectrode descents, numerous tonically firing units were recorded; their number was large in comparison to the figures reported after investigations carried out with highly selective electrodes in anesthetized cats. The number of units resolved over a 1 mm displacement of electrode tip ranged from 15 to 30 with an average value of 23. Assuming that a microelectrode explored a sphere with a radius of approximately 25 #m, it can be estimated that the unit density within cubes with sides of 100 # m (10 n cu.#m) ranged from 4.0 to 11.8 with an average value of 7.6. From 315 descents performed on 50 cats, 23,000 units were recorded and kept for analysis. Respiratory modulation was determined from CTH. Of the analysed cells, 28% were found to be RRU, i.e., having a RMI over 0.70. This proportion is very much less than the 47700found previously at the level of the upper ports 1°. Phasic RRU, i.e., units whose discharge stops firing during one part of the respiratory cycle, were found in 15% of cases only. They corresponded to the spikes of largest amplitude. They could be recorded anywhere in the medulla and pons, but predominated in the bulb up to a level 5 mm in front of the obex (Fig. 2).
216
VHt
12
~
J 7L
~
.KF
~
S
5M
I
I
I
I
..Q
501
o
0 4
2
0 - mm+
2
4
6
8
10
Fig. 2. Distribution of phasic R R U in the medulla. Abscissa: rostro-caudal position of the units relative to the obex. Ordinate: percentage of phasic units relative to RRU. Horizontal lines in the upper part indicate the relative level of the main respiration modulated structures. Abbreviations: 5M, motor trigeminal nucleus; 7L, facial nucleus, lateral division; 12, hypoglossal nucleus; KF, K611iker-Fuse nucleus; S, solitary tract; VH, ventral horn of the first cervical segment. TABLE I
Respiratory characteristics of bulbo-pontflte respiratory structures. UD, mean unit density per 106 cu./tm; R R U D , mean R R U density par l06 cu.#m; R R U RRUD
~ =
× 100; RMI, mean respiratory modulation indexL The separation between 3 categories A,
UD B and C results from a variance analysis of the RMI. Out of the 35 explored structures, 25 fall in the C category; data from some of them only are presented in the table.
Structures
A Motor trigeminal nucleus Hypoglossal nucleus Solitary complex Ventral horn Facial nucleus, lateral division Sensory trigeminal complex
Number of punctures
Number of units
UD
RR U% RMI ± SD log R M I
10 31 15 11 16 29
234 681 253 183 312 698
6.9 6.3 5.6 7.3 6.5 7.8
71.6 63.5 63.2 50.3 48.0 43.7
1.10 1.87 1.33 1.30 1.36 1.05
± ± ± 4± ±
1.04 1.51 1.16 1.28 1.21 1.21
122 24
4607 706
7.5 6.9
37.9 36.7
0.91 ± 1.17 0.84 ± 0.98
14 27
183 1074
10.0 7.2
34.9 32.4
0.77 ± 1.07 0.79 ± 0.97
5 25 13 19 19 35
281 408 161 288 326 569
11.8 5.3 6.8 9.4 7.3 7.3
27.1 22.0 17.3 15.2 12.3 10.4
0.48 0.48 0.67 0.45 0.51 0.50
B
Lateraltegmental field Gigantocellular tegmental field Lateral reticular nucleus of the medulla, internal division Magnocellular tegmental field C Medial longitudinal bundle Lateral vestibular nucleus Trapezoid body Inferior olive Spinal trigeminal tract Cuneate nucleus
± ± ± ± ± ±
0.60 0.88 1.01 0.76 0.89 0.83
217
R R U density In 35 anatomical structures which were explored at least 5 times, 3 parameters were defined: (1) total unit density; (2) R R U percentage relative to the total number of recorded cells and (3) mean respiratory modulation index (RMI) which was calculated for all recorded units (RRU as well as non-respiratory). Results (Table I) show 3 categories of structures. In the first (A), R R U percentage was over 43~ and R M I over 1.4; in the second group (B), R R U percentage ranged between 38~ and 32~ and RMI between 1.0 and 0.70; in the last category (C), R R U percentage was less than 28~ and RMI less than 0.70. Comparison of samples from each anatomical structure made variance analysis possible. It was performed on the RMI; the t-test was used to compare pairs of structures. Data from each of the 35 structures were tested individually against all the 34 others. Results confirmed the existenCe of 3 categories of structures by showing that: within a given group, RMI of each structure was significantly different from that of any structure of the two other categories (P < 0.001); within each category, there was no significant difference between structures. Comparable results were obtained for R R U percentages. Table I shows that the higher respiratory modulation is found in motor and sensory cranial nuclei that are known to be involved in the control of the respiratory accessory musculature, these being the motor nuclei of trigeminal, facial, vagus and hypoglossal nerves and ventral horn of the first cervical segment, and the sensory nuclei of alaminar spinal trigeminal sensory nucleus and solitary complex. It should be noted that a higher R R U percentage and R M I were found in the solitary complex which comprises the infrasolitarius nucleus (dorsal inspiratory nucleus), where recent studies have emphasized the presence of RRUll, 19. The fact that R R U were recorded in vagotomized animals confirms the inherent respiratory rhythmicity of the solitary complexT,19. In the B category fall reticular structures which have always been thought to support the control of respiratory oscillation. These are the lateral, giganto-cellular, magno-cellular and paramedian tegmental fields. The unit density in tegmental fields is comparable to that of other parts of the medulla and pons while the R R U percentage is significantly lower than in the respiratory modulated sensori-motor nuclei. The C category contains structures never before suspected of taking part in respiratory activity, even though definite R R U were observed in all of them. Included in this category are the medial longitudinal bundle, the vestibular and abducens nuclei, the lateral lemniscus, the ventral cochlear nucleus, the inferior and superior olivary nuclei and the trapezoid body; the spinal trigeminal tract, the cuneatus and gracilis nuclei and medial lemniscus and the restiform body, etc. RR U classification Examination of a large number of R R U discharges suggested that the firing patterns corresponded in most of cases to inspiratory (I) and expiratory (E) types. Within the latter category, 3 subtypes were often observed, early, mid and late. Examples of these are given in Fig. 3 for cases of either tonic or phasic activity.
218
I
eE
mE
(t)E
Fig. 3. Example of cycle triggered time histograms (CTH) corresponding to the 4 most common types of R R U discharge patterns. I, inspiratory; eE, early expiratory; mE, mid expiratory; (1)E, late expiratory. Black and white bars under each CTH indicate inspiration and expiration respectively. Duration of bins, 160 msec; number of cycles, 30.
In order: to compare our results with those of previous authors, to eliminate personal bias and to classify a large number of individual units, it was necessary to develop an algorithm for computer classification. To allow recognition of the most commonly described types, the cycle was divided into 6 parts. Owing to the usual 1:2 ratio between I and E durations, I was divided into 2 parts and E into 4, thus resulting into I and E sections of approximately the same width. The average frequency in each one was calculated from the CTH. A R R U discharge was identified as belonging to a given class when the highest average frequencies were found in two neighboring temporal sections (i.e., predominated in one-third of the cycle). Six respiratory classes could be recognized (Fig. 4A) which corresponded to types described by other authors3,4,15,32: phase spanning expiratory-inspiratory ('EI'), inspiratory (T), phase spanning inspiratory-expiratory ('IE'), early expiratory ('eE'), mid expiratory ('mE') and late expiratory ('(1)E'). CTH from 6,440 R R U were processed and classified. I units represented 47.2~ of the total respiratory population; 34.9~ of the R R U were E, of which the eE pattern comprised 18.2~, the mE 10.6~ and the (1)E 6.1~. For the phase spanning unit 5.9~o were of the EI type and 7.7~ of the IE type. Results are summarized in the diagram in Fig. 4B which indicates that the probability of observing phase spanning between bulb and upper pons is low. s
v~/J,'/,I
[
I
1
I
K/IrlA I
I
I 1 I
I
,
t
I
K/,r/j
EI
I
I
I
I
I
I
1
I
I
I
r,/,4/A
t
I
t
rl.~
40.
N = 6440
I
,
l//V/~
I
I
I
I/All1
I
'1 ,
%
I
I
I
,
t
•
IF
20.
eE 1~ m~
7eE
I
2 ' 3t
l
5
El
I(L)E
v/x/.a
4
i~
7mE
6' I
I
I 1
Phr
H
~ 2
I
3
,
4
(L)E I
I 5
I 6 ,
E
A
B
Fig. 4. A: Schematic illustration of the principle of the algorithm developed for automated classification. A unit is recognized as pertaining to one of the 6 classes (see text) when the highest average frequencies (hatched areas) fall into two neighboring sections. B: Diagramatic representation of R R U percentage in each of the 6 classes. Black and white bars indicate inspiration and expiration respectively.
219 N
~NxI° 1 N 10
900.
I I I i t 5
9.
00
s eSSe
_F
/ o S° toS t si
5
g t
tI
tI
o
:' J
I
E
I
E
Fig. 5. Temporal distribution of peak firing frequencies (PFF). Left: PFF histogram; abscissa, respiratory cycle. Circles indicate position of the modes of the 3 putative E sub-populations. Right: probability distribution function. The curves were obtained by plotting the cumulative distribution of left histogram on normal probability paper. Dotted curves were drawn by approximation. The remaining 4.3~ units were definitely respiration modulated, but their discharge pattern did not correspond to previous descriptions at the bulbar or pontine level.
Peak firing frequency distribution The main drawback of such a classification is that it is based on a priori or intuitive grounds. It is assumed that individuals in each class pertain to a 'natural' system. The existence of such a system can be verified if its classes fit with statistical populations. In a preceding paper 1°, it was shown that the grouping of R R U into several subpopulations could be better analyzed by comparing the temporal position of C T H peaks or peak firing frequencies (PFF) while increasing the number of temporal classes. In the present work, P F F were distributed into 15 temporal classes, 5 in I and 10 in E in order to respect the I / E duration ratio. R R U were sampled anatomically at random. Calculations could be carried out on a sample of 5,000 R R U . That the sample was representative of the population is indicated by the fact that above 3,000 units, increasing the sample size did not modify the shape of the histogram. Results shown in Fig. 5A clearly indicate one mode in inspiration. In expiration, the temporal distribution of P F F is not so well defined and
220
%
20. N=664 10_ 1%N'95
i
A ~
N=216
10~1
N=229
E
D 20
C
B
,•N:484 F
i
.
_ N=1793
N=1235
i
G H I Fig. 6. Temporal distribution of PFF in respiratory structures. Examples for individual structures : motor nuclei of the Vth (A) and XIIth (B); sensory nuclei of the trigeminal complex (D) and of the solitary complex (E); lateral (G) and giganto-cellular (H) tegmental fields. Pooled histograms for motor (C), sensory (F) and reticular (I) structures. Ordinates, number of samples in each class as a percentage of RRU of each structure. Abscissae,as in Fig. 5. the existence of 3 expiratory modes, indicated by circles in Fig. 5, is perhaps possible. Whereas statistical methods exist to test population homogeneity, heterogeneity tests are lacking. In the present study, an attempt to ascertain E population heterogeneity was made by studying the probability distribution function of the sample. Curves were obtained using a graphic method consisting of plotting the cumulative distribution on normal probability paper. Results suggested the existence of 4 populations having different variances as indicated by the different slopes of the 4 segments of the curve illustrated in Fig. 5 right. It follows that firing patterns of the R R U of the explored regions could fall into 4 sub-populations: [, eE, mE and (1)E. Though it was not possible to ascertain this hypothesis statistically, the existence of 3 E categories will be taken into consideration in the following paper where localization of neurons according to their respiratory pattern will be attempted.
Regional analysis In searching for separate functions of the various parts of respiratory structures,
221 the temporal distribution of P F F was studied in anatomical regions where R M I was over 0.70. P F F histograms were constructed separately for each one, and then pooled according to their physiological function. Examples of separate and pooled PFF histograms are shown in Fig. 6 from R R U recorded in individual motor nuclei (Fig. 6A-C), sensory nuclei (Fig. 6 D - F ) and reticular tegmental fields (Fig. 6G-I). The pooled histogram for motor nuclei (XII, X, VII, V, first segment of the ventral horn) shows that most R R U fall into the I category (Fig. 6C). The two modes appearing in early and late expiration correspond mainly to the existence of large populations of eE in the Vth nucleus (Fig. 6A) and (1)E in the ventral horn. In sensory nuclei (V and X) the 4 populations are represented, since a definite mode appeared in mid expiration. In the tegmental fields, the pooled histogram (Fig. 6I) is very similar to that of the total histogram (Fig. 5). Comparison of individual histograms from lateral and giganto-cellular tegmental fields showed, however, that I R R U predominate in the lateral tegmental field, while E units were more numerous medially. These differing temporal distributions thus suggest the possibility of a spatial organization of R R U , but no clear correspondence with anatomy can be detected. An attempt was also made to compare the P F F distribution in caudal and rostral parts of the medulla on each side of a plane transecting the brain stem 3 m m in front of the obex. Striking differences then appeared in the [ and E ratios both in the case of
~
10-
7
~
_N=1315
i
A
C
B
A
20_
N=1231
625
N =1856
10-
D
E
F
Fig. 7. Comparison of temporal distributions of PFF in caudal and rostral parts of bulbo-pontine region. The plane of separation transects the brain stem 3 mm in front of the obex. Upper section, pooled histograms of rostral structures. A: trigeminal complex, VIIth lateral nucleus, solitary complex. B: lateral, giganto-cellular and magno-cellular tegmental fields. C: total rostral RRU population. Lower section, pooled histograms of caudal structures. D: solitary complex, XIIth nucleus, ventral horn of the first cervical segment. E: lateral tegmental field, paramedian reticular nucleus. F: total caudal RRU population. Coordinates as in Fig. 6. Note that the mode in I is later in the rostral than in the caudal reticular formation. Note that the inspiratory mode is significantly earlier in sensory motor nuclei and caudal reticular formation when compared to the rostral reticular formation.
222 pooled sensory-motor and reticular histograms, as well as in the total rostral and caudal populations (Fig. 7). I represents two thirds of R R U in the bulb whereas an equal proportion o f [ and E is found rostrally. DISCUSSION The results reported in this paper have to be considered from two aspects: (1) is the unit sample representative of the whole neuronal population of the explored area? (2) What is the number of the temporal R R U populations shown to exist ? Improvement of recording techniques during the past two decades has increased progressively the number of R R U analyzed in brain stem exploration. Baumgarten 5, using 25 # m wire electrodes recorded 23 R R U in 900 penetrations, i.e., 0.025 R R U in each descent on the average. Nelson al reported 0.5 R R U in each penetration performed with extracellular micropipettes with 2 # m tips. Waldron 41 used tungsten wire in glass micropipettes with tips 3 # m in diameter and from the reported data, it can be calculated that about 2 R R U were recorded in each penetration. Lowering the impedance of the same type of electrode with platinum black was shown to increase markedly the number of units recorded at the same point, thus resulting in a multiunit signal 3°. In the present study, where signal filtering was avoided using capacitancecompensated electrodes, an average of 23 R R U could be recorded in each microelectrode descent. Two other technical points also seemed important in explaining the magnitude of the increase in number of units recorded: (1) avoidance of barbiturate anesthesia which has been reported to reduce the number of firing R R U in the reticular fields2s,29, (2) the possibility of recording in the medulla without surgical ablation of the cerebellum which drastically modified the histological appearance of reticular neurons 4° and (3) analysis of units firing tonically. Since spikes could be distinguished roughly every 50 # m during the microelectrode descent, it can be supposed that the electrode tip detected activity over a 25 # m radius. Using this assumption, it can be calculated that the number of firing neurons in a 106 cu.#m volume ranged from 4 to 11 in the medulla and from 5 to 10 in the R R U containing structures (Table [). These figures can be compared to those obtained with histological methods. Scheibel and Scheibe188 calculated from brain stem sections stained with the Golgi method that there may be 15-30 cells per 10n cu.#m in the magno-cellular portion of the medullary tegemental fields, 20-50 in parvocellular portions, and 40 to 60 in more tightly packed reticular groups. The average cell density was estimated at 33 per 106 cu.#m. These values were obtained in 10-day-old kittens; cell density is less for the adult, where brain size is greater though the total number of cells remains the same. We have calculated that to make a rough comparison of densities in the newborn and adult cat, it is necessary to multiply the volume by a factor of 10-15. On the other hand, in the Golgi preparation less than 4 0 ~ of cells are supposed to be silver impregnated. These considerations indicate that, assuming a unit volume of 10n cu.#m, the number of neurons in an adult cat's medulla could range from 2 to 10 with an
223 average value from 4 to 7. Considering the large approximation unavoidably introduced in calculations, it can be nevertheless considered that figures obtained with histological and electrophysiological methods are reasonably concordant and that the sampling carried out with low impedance, capacitance compensated microelectrodes was representative of the total population of spontaneously active neurons. The second point concerns the number of temporal sub-populations of R R U . As it was recalled in the introduction, many authors have attempted previously to classify medullary and pontine R R U according to the phase relation of bursts to the efferent respiratory motor discharges. Unfortunately, due to the limited number of units analyzed during single cell recording, the size of samples remained small in most studies. In the present work, the R R U sample turned out to be large enough to provide statistical information. It has been confirmed that medullary and pontine R R U fall into 2 broad categories [ and E. Results suggest that the E population can be further divided into 3 sub-populations: early, mid and late. These conclusions are in agreement with those of recent studies3,4,11,3°,~2. On the other hand, no evidence could be provided to support the existence of definite phase spanning populations in the brain stem caudal to the pneumotaxic system. This study finally failed to answer the important question of the number of oscillators present in the medulla. Are the R R U sub-populations pertaining to a single system, or are they parts of several oscillators? Regional analysis was not able to give any further indication about this, except that there were consistent differences in R R U temporal distribution according to anatomical structures. An entirely different approach towards solving this problem is reported in the next paper. ACKNOWLEDGEMENT This work was supported by grants from the Centre national de la Recherche scientifique (LA 204) and the D616gation g6n6rale ~ la Recherche scientifique et technique (72-7-0740).
REFERENCES 1 Achard, O. et Bucher, V. M., Courants d'action bulbaires ~t rythme respiratoire, Helv. physioL pharmacoL Acta, 12 (1954) 265-283. 2 Amoroso, E. C., Bell, F. R. and Rosenberg, H., The localization of respiratory regions in the rhombencephalon of the sheep, Proc. roy. Soc. B, 139 (1951) 128-140. 3 Batsel, H. L., Localization of bulbar respiratory center by microelectrode sounding, Exp. Neurol., 9 (1964) 410426. 4 Batsel, H. L., Some functional properties of bulbar respiratory units, Exp. NeuroL, 11 (1965) 341-366. 5 Baumgarten, R. Von, Koordinationsformen einzelner Ganglienzellen der rhombencephalen Atemzentren, Pfliigers Arch. ges. PhysioL, 262 (1956) 573-594. 6 Baumgarten, R. Von, Balthasar, K. und Koepchen, H. P., Ober ein Substrat atmungsrhythmischer Erregungsbildung im Rautenhirn der Katze, Pfliigers Arch. ges. PhysioL, 270 (1960) 504-528. 7 Baumgarten, R. Von and Kanzow, E., The interaction of two types of inspiratory neurons in the region of the tractus solitarius of the cat, Arch. ital. BioL, 96 (1958) 361-373.
224 8 Bertrand, F. and Hugelin, A., Respiratory synchronizing function of nucleus parabrachialis medialis: pneumotaxic mechanisms, J. NeurophysioL, 34 (1971) 189-207. 9 Bertrand, F., Hugelin, A. and Vibert, J. F., Quantitative study of anatomical distribution of respiration related neurons in the pons, Exp. Brain Res., 16 (1973) 383-399. 10 Bertrand, F., Hugelin, A. and Vibert, J. F., A stereologic model of pneumotaxic oscillator based on spatial and temporal distributions of neuronal bursts, J. Neurophysiol., 37 (1974) 91-107. 11 Bianchi, A. L., Localisation et 6tude des neurones respiratoires bulbaires. Mise en jeu antidromique par stimulation spinale ou vagale, J. Physiol. (Paris), 63 (1971) 5-40. 12 Cohen, M. I., Discharge patterns of brain stem respiratory neurons in relation to carbon dioxide tension, J. Neurophysiol., 31 (1968) 142-165. 13 Cohen, M. I., Discharge patterns of brain-stem respiratory neurons during Hering-Breuer reflex evoked by lung inflation, J. Neurophysiol., 32 (1969) 356-374. 14 Cohen, M. I., How respiratory rhythm originates : evidence from discharge patterns of brainstem respiratory neurones. In R. Porter (Ed.), Breathing: Hering-Breuer Centenary Symposium, Churchill, London, 1970, pp. 125-150. 15 Cohen, M. I. and Wang, S. C., Respiratory neuronal activity in pons of cat, J. NeurophysioL, 22 (1959) 33-50. 16 Deville, J., Un syst6me ~t hautes performances pour l'analyse intra-cellulaire, Eleetroenceph. elin. Neurophysiok, 40 (1976) 521-528. 17 Dirken, M. N. J. and Woldring, S., Unit activity in bulbar respiratory centre, J. Neurophysiol., 14 (1951) 211-225. 18 Duffin, J., The functional organization of respiratory neurones. A review, Canad. Anaesth. Soc. J., 19 (1972) 1-7. 19 Euler, C. von, Hayward, J. N., Marttila, I. and Wyman, R. J., Respiratory neurones of the ventrolateral nucleus of the solitary tract of cat: vagal input, spinal connections and morphological identification, Brain Research, 61 (1973) 1-22. 20 Feldman, J. L. and Cowan, J. D., Large-scale activity in neural nets. II. A model for the brainstem respiratory oscillator, Biol. Cybern., 17 (1975) 39-51. 21 Geman, S. and Miller, M., A mathematical model of the respiratory centers, The Physiologist, 16 (1973) 320. 22 Gesell, R., Bricker, J. and Magee, C., Structural and functional organization of the central mechanism controlling breathing, Amer. J. Physiol., 117 (1936) 423452. 23 Gesell, R., Magee, C. and Bricker, J., Activity patterns of the respiratory neurons and muscles, Amer. J. Physiol., 128 (1940) 615-628. 24 Haber, E., Kohn, K. W., Ngai, S. H., Holaday, D. A. and Wang, S. C., Localization of spontaneous respiratory neuronal activities in the medulla oblongata of the cat: a new location of the expiratory center, Amer. J. Physiol., 190 (1957) 350-355. 25 Holmer, N. G., An electrometer amplifier for electrophysiology with low input capacitance, Med. biol. Engineer., 8 (1970) 509-511. 26 Hukuhara, T. Jr., Neuronal organization of the central respiratory mechanisms in the brain stem of the cat, Acta Neurobiol. exp., 33 (1973) 219-244. 27 Hukuhara, T., Nakayama, S. and Okada, H., Action potentials in the normal respiratory centers and its centrifugal pathways in the medulla oblongata and spinal cord, Jap. J. Physiol., 4 (1954) 145153. 28 Hukuhara, T. Jr., Saji, Y., Kumadaki, N., Kojima, H., Tamaki, H., Takeda, R., und Sakai F., Die Lokalisation von Atemsynehron entladenden Neuronen in der retikul~iren Formation des Hirnstammes der Katze unter verschiedenen experimentellen Bedingungen, Naunyn-Schmiedebergs Arch. exp. Pathol. Pharmak., 263 (1969) 462-484. 29 Karczewski, W., Contemporary views concerning central control of respiration, Acta physiol. polon., 21, Suppl. 1 (1970) 61-80. 30 Merrill, E. G., The lateral respiratory neurones of the medulla : their associations with nucleus ambiguus, nucleus retroambigualis, the spinal accessory nucleus and the spinal cord, Brain Research, 24 (1970) 11-28. 31 Nelson, J. R., Single unit activity in medullary respiratory centers of cat, J. Neurophysiol., 22 (1959) 590-598. 32 Nesland R. and Plum, F., Subtypes of medullary respiratory neurons, Exp. Neurol., 12 (1965) 337-348. 33 Palmay, F. v., Davison, E. J. and Duffin, J., The simulation of multineurone networks: modelling
225
34 35 36 37 38
39 40 41 42
of the lateral inhibition of the eye and the generation of respiratory rhythm, Bull. math. Biol., 36 (1974) 77-89. Robinson, D. A., The electrical properties of metal microelectrodes, Proc. LE.E.E. reed. Engineer., 56 (1968) 1005-1071. Rubio, J. E., A mathematical model of the respiratory center, Bull. math. Biophys., 29 (1967) 719-736. Rubio, J. E., A new mathematical model of the respiratory center, Bull. math. Biophys., 34 (1972) 467-481. Salmoiraghi, G. C. and Burns, B. D., Notes on mechanism of rhythmic respiration, J. NeurophysioL, 23 (1960) 14-26. Scheibel, M. E. and Scheibel, A. B., Structural substrates for integrative patterns in the brain stem reticular core. In H. H. Jasper, L. D. Proctor, R. S. Knighton, W. C. Noshay and R. T. Costello (Eds.) Reticular Formation of the Brain, Little, Brown, Boston Mass., 1958, pp. 31-55. Strickholm, A. and Winston, S., An electrometer amplifier for electrophysiology, Med. biol. Engineer., 7 (1969) 99-102. Tyc-Dumont, S., Histopathological alterations of brain stem neurons provoked by cerebellar ablation, in preparation. Waldron, I., Activity patterns in respiratory muscles and in respiratory neurones of the rostral medulla of the cat, J. Physiol. (Lond.), 208 (1970) 373-383. Wolbarsht, M. L. and Wagner, H. G., Glass insulated platinum microelectrode design and fabrication. In F. Bostein (Ed.), Medical Electronics (Proceedings of the 5th Intern. Conf on Med, Electronics, Likge, July 22-26, 1963), 1963, pp. 510-515.