Recruitment and discharge frequency of phrenic motoneurones during inspiration

Respiration

Physiology

(1976) 26, 113-128; North-Holland

Publishing

Company,

Amsterdam

RECRUITMENT AND DISCHARGE FREQUENCY OF PHRENIC MOTONEURONES DURING INSPIRATION’

STEVE ISCOE’, JERRY DANKOFF, RICHARD CAN10 POLOSA Department

MIGICOVSKY

and

of Physiology,McGill University, Montreal, Quebec H3G I Y6, Canada

Abstract. The discharges of 107 phrenic motor axons were recorded from cats under chloralose-urethane

anaesthesia with spinal cords transected at T, or with intact neuraxis. During inspiratory occlusions in spinal cats, each motoneurone was recruited at a mouth pressure constant at a given end tidal COz; no motoneurone was recruited at a pressure greater than 70% of maximum. In eupnoea (32.3 torr COz) 73 % of motoneurones were recruited during the first 30% of inspiration; during CO, rebreathing (60.8 torr CO*), 89 “/dwere recruited in the first 30 % of inspiration. Neurones recruited earlier in inspiration had a lower onset frequency than later recruited units; all increased instantaneous frequency in a linear relation to pressure. Early recruited units showed a smaller increase in frequency per unit change in pressure than did later recruited units. During CO* rebreathing. mean and peak frequencies increased on average 0.92 and 1.78 spikes . set-‘(% CO,))‘, respectively, these increases being significantly less for early than for late recruited neurones. The data show that a stable order of recruitment of phrenic motoneurones exists during inspiration, the excitability of each motoneurone likely determining its time of recruitment. Above threshold, later recruited motoneurones are more ‘sensitive’ to a change in input. Recruitment of motoneurones is responsible for pressure generation at the start of inspiration and increase in discharge frequency (rate coding) is the dominant mechanism in the second half of inspiration. Airway obstruction Inspiratory activity Rate coding

Rebreathing Recruitment Size principle

The force generated by the inspiratory muscles increases smoothly with time from the beginning to the end of inspiration. This is most easily seen by recording mouth pressure during an occluded breath in anaesthetized man and animals. Accordingly,

Acceptedfor

publication

30 October

1975.

’ This study was supported by the Medical Research Council of Canada. * Present address: Department of Physiology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, N.Y. 10461, U.S.A. 113

114

S. ISCOE,

J. DANKOFF, R. MIGICOVSKY

and

C. POLOSA

when the output of the phrenic motoneurone pool is monitored by recording the neurogram of a phrenic nerve branch, one observes that the number and size of the spikes carried by the nerve per unit time increases smoothly during inspiration. How then is the activity of the phrenic motoneurones and their associated diaphragmatic motor units regulated by the central nervous system to generate an output with the characteristics described above? Adrian and Bronk (1928, 1929) pointed out that the force of voluntary or reflex contraction of a skeletal muscle can be increased by increasing (a) the number of active motoneurones (recruitment) and (b) the discharge frequency of neurones already active (rate coding; Partridge, 1966; Mannard and Stein, 1973). Several questions related to the behaviour of diaphragmatic motor units during inspiration arise: Does recruitment occur at random or according to an orderly sequence? Do recruitment and rate coding operate throughout inspiration or are they limited to certain phases of it ; i.e., is one or the other used predominantly at certain force levels and not at others? What is the relative importance of each when both mechanisms operate? Is the ability to increase discharge frequency homogeneously distributed among the motoneurones? Although a number of observations concerning the firing patterns of single phrenic motoneurones or inspiratory motor units have been made (Adrian and Bronk, 1928; Gesell et al., 1941; Pitts, 1942; Gill, 1963; Yasargil, 1964, 1967; Budingen and Yasargil, 1972; Nail et al., 1972; Hilaire et al., 1972) clearcut answers to the above questions related to inspiration are not yet available to describe what happens during a single eupnoeic breath or under conditions of increased respiratory drive. In the present investigation, we have studied the firing patterns of individual phrenic motoneurones in the course of a single breath during eupnoea and CO, rebreathing. The results indicate that at normal end tidal CO, tensions, recruitment is the neural mechanism by which force is generated at the onset of inspiration but, as inspiration proceeds, rate coding becomes dominant. During CO2 rebreathing, little recruitment of previously inactive unit occurs and the pattern characterizing a normocapnic breath is accentuated, i.e., recruitment in early inspiration, rate coding in late inspiration.

Methods Experiments were performed on nine cats of either sex, weight 2.5-4.2 kg, anaesthetized with chloralose (40 mg/kg, i.v.) and urethane (250 mg/kg, i.v.) after induction with ether. Femoral arterial blood pressure was continuously monitored. An infrared heat lamp maintained body temperature between 3638°C. Expired COZ was measured with an infrared analyzer (Beckman LB 1). In two cats a spinal transection was made at the first thoracic segment in order to eliminate non-diaphragmatic inspiratory activity. The cervical phrenic nerves were approached ventrally, cleared of surrounding tissue, and placed on silver hook electrodes immersed in a pool of paraffin oil formed

PHRENIC MOTONEURONE DISCHARGE

115

by the surrounding skin flaps. The neurogram was recorded from one nerve while small strands, containing units whose individual spikes could be identified, were dissected from the contralateral nerve. The mass phrenic neurogram was used as a time base to which the activity of individual units could be referred. Inspiration (I) is defined as the time from onset of phrenic activity to start of rapid decrease in activity, determined either from the raw or from the integrated (time constant, 100 msec) mass phrenic neurogram. Action potentials, recorded diphasically, were led through high impedance probes, amplified (bandpass 60-10,000 Hz) and recorded on magnetic tape, together with expired CO2 and arterial blood pressure. In the two spinal cats, mouth pressure (Statham 267B) was recorded. In these cats, occlusions during room air breathing were applied at FRC by clamping, during expiration, the inspiratory side of a one way valve attached to the trachea. COZ rebreathing (5 % COZ, 50 % 02, 45 % N,) was used to increase respiratory drive. A rebreathing run usually took 6-7 min, peak CO, values at the end of rebreathing averaging 60.8 k6.4 (SD) torr, eupnoeic values 32.3 +4.5 (SD) torr. The following properties of individual motoneurones were measured: (1) the level of diaphragmatic pressure generated during an inspiratory effort against a closed airway at which the motoneurone was recruited (in the two spinal cats), (2) the time between the onset of activity in the mass phrenic neurogram and the first spike of an individual motoneurone during an unobstructed breath, (3) the mean frequency (reciprocal of mean interspike interval) during a burst, (4) the peak frequency (reciprocal of shortest interspike interval) during a burst, (5) onset frequency (reciprocal of first interspike interval) and (6) amplitude of the extracellular spike. Standard statistical methods were used throughout (Downie and Heath, 1965; Freund, 1962) to compute linear regression lines on linear and semilogarithmic plots and to test the significance of the parameters characterizing these lines.

Results A typical recording (fig. 1) shows the activity of two phrenic motoneurones, the phrenic neurogram from the contralateral nerve, its integrated activity and mouth pressure during a non-occluded and an occluded breath in a cat with spinal cord transected. The unit with the largest spike started to fire when the pressure reached -0.55 +0.15 (SD, n = 6) cm H,O while the other unit started at a pressure of -3.83f0.54 (SD, n = 6) cm H,O. Recruitment of the 16 units studied in the two spinal cats occurred at pressures ranging from - 0.13 to - 8.43 cm H,O, the latter value being approximately 70% of the maximum pressure reached during the occluded breaths. The pressure at which a unit was recruited was reproducible from one breath to another at constant Pco, , as shown by the small SD of the mean, and can be considered a fixed characteristic of that unit. Thus a fixed order of recruitment exists in the phrenic motoneurone pool. If one considers the pressure-time curve on occlusion as an analogue of the output of the motoneurone pool (Whitelaw et al.,

116

s. IscoE, J. DANKOFF,

R. ~wGIC~VSKY

and c. Po~o.3~

-0

11 cm Hz0

-10

1 set Fig. 1. Record of, from top down, activity of two phrenic motor axons, neurogram of contralateral phrenic nerve, integrated phrenic neurogram and mouth pressure. Second inspiration against occluded airway. Spinal cat.

1975) the fixed order of recruitment means that each neurone starts to tire at a well defined level of motoneurone pool output, that is, has a pressure ‘threshold’. This threshold, in turn, can be taken as an index of the excitability of that neurone to the particular input under study (the central respiratory drive potential, CRDP; Eccles et al., 1962). Since the pressure generated by the diaphragm increases monotonically with time during both occluded and unobstructed inspirations (Whitelaw et al., 1975; Younes et al., 1975), the relationship between the pressure at which a unit was recruited during an occluded breath and the time when it was recruited during an unobstructed inspiration was studied. This was done to establish whether the use of recruitment time during a normal inspiration to express the ranking order and excitability of a motoneurone was justified. Recruitment time was expressed as a percent of inspiratory duration to eliminate the effect of variations in respiratory cycle duration between cats and in the same cat. The results from the 16 units studied in the two spinal cats are shown in fig. 2. The high negative correlation (- 0.961) between the two variables indicates that the time of recruitment (expressed as ‘A of TI) is an adequate index of the ‘threshold’ pressure level at which a motoneurone is recruited. The line of best fit intersects the ordinate above zero because neural activity must lead generation of pressure. The frequency distribution of recruitment times, expressed as % TI, obtained in 101 units in both intact and spinal transected cats, is shown in fig. 3. The recruitment time of each unit was averaged from 10 breaths. The standard deviation of the time of recruitment of all units averaged less than 50 msec, that is less than 6% of the average TI. The standard deviation of the recruitment time of individual units cor-

PHRENIC MOTONEUROIW DISCHARGE

I

I

L

-3 Threshold

L

L

-6 pressure for recruitment

(cm $0)

Fig. 2. Relationship between threshold pressure for recruitment of phrenic motoneurones during an occluded inspiration UStime of recruitment, expressed as percent of inspiratory duration during nonoccluded breaths. Each point is mean of six observations. Spinal cats.

related significantly with time of recruitment (r = 0.698, P < 0.001, n = 94). A large number of units are recruited during the first decile of inspiration and 73 % of all units, including those starting to fire before the onset of the mass phrenic neurogram, are active before 30 ‘A TI. Progressively fewer units are recruited later in inspiration until, beyond 80 % Tr, no additional units are recruited. Since VT increases until the end of inspiration, rate coding must be the mechanism responsible for force generation in late inspiration. Conversely, recruitment is the only mechanism operating at the start of inspiration, at low force levels, since the average interval between the first two spikes in a burst, for all units, is approximately 119 msec and rate coding cannot operate within this time. For eupnoeic breathing, between 30 and 80% TI, either recruitment is of minor importance compared to rate coding in contributing to force generation or the pressure contributed by the later recruited units is sufficiently greater than that generated by the early units to compensate for their fewer numbers. Is recruitment in late inspiration more important under conditions of increased respiratory drive? This point was investigated by subjecting 7 cats to CO2 rebreathing. The results are shown in fig. 4. The control distribution is provided in panel A;

118

S. ISCOE, J. DANKOFF, R. MIGICOVSKY

and C. POLOSA

spinal section N:16

L-II

0

nauraxis intact Nx65

50 Recruitment

time

100

(“A insoiratoryduration)

Fig. 3. Frequency histogram of time of recruitment expressed as percent of inspiratory duration. Hatched areas indicate units from spinal cats.

this histogram is the same as that shown in fig. 3 but without the units from the spinal cats in which rebreathing was not performed. In panel B, the frequency distribution of the recruitment times obtained during CO, rebreathing, at an average end tidal CO, of 60.8k6.4 (SD) torr, is plotted. The pattern of recruitment of high COz, in comparison with the control, has shifted towards the ordinate, 89 o/oof the units now appearing in first 30% of TI. Only six additional units, indicated by the hatched blocks, were recruited. Recruitment terminates at 60 % TI. Thus the pattern observed in eupnoea is accentuated by increased respiratory drive, recruitment increasing force at the start of inspiration, rate coding being the exclusive mechanism for increasing force over the second half of the breath. For all units, which were active in normocapnia, recruitment time decreased as CO2 concentration increased. However, the order of recruitment, in a given multiunit strand, was unaffected by CO2 rebreathing. Units which appeared first in eupnoea also appeared first at elevated levels of CO*. The data of fig. 4 are plotted on semi-logarithmic coordinates in fig. 5. During both normo- and hypercapnia, the contour of the histograms, excluding units whose

119

PHRENIC MOTONEURONE DISCHARGE = 32.3 TORR pEk02

P

N = 85

N = 81

ETco2

=00.3TORR

40

n 3

30

5 % 5

20

z

10 :

.-E

Recruitment

time

(96 inrpirataydumtion)

Fig. 4. Frequency distribution of time of recruitment as percent of inspiratory duration in cats with intact neuraxis at an end-tidal CO* of 32.3 torr (A) and at an average end-tidal CO2 of 60.8 torr during CO, rebreathing (B). Newly recruited units indicated by hatched columns. IO?

0

0 N=

71

Y = YLs.-*‘X rr-W

N 178 y = 45.si~~x r=-35

\

to -

10 -

s-

2-

IL 0

Fig. 5. Data of fig. 4 plotted on semilogarithmic coordinates. Units recruited before onset of activity in phrenic neurogram excluded.

120

S. ISCOE, J. DANKOFF, R. MIGICOVSKY

and C. POLOSA

activity started before the onset of mass phrenic nerve activity, can be fitted by negative exponentials. The half-time for recruitment during eupnoea is 0.20 TI and at 60.8 torr CO2 is reduced to 0.12 TI. Although there is some evidence that diaphragmatic motor units recruited later in inspiration produce larger tetanic tensions than earlier ones (Budingen and Yasargil, 1972) the relationship between ‘threshold’ level of time or pressure at which a unit is recruited and its corresponding contribution to diaphragmatic pressure is unknown. Some idea of the role of rate coding in generating pressure may, however, be obtained by plotting the instantaneous frequency of discharge of individual units us the output of the system (pressure). The results obtained from the two units of fig. 1 are shown in fig. 6. The average time when each given value of pressure was attained is also provided on the abscissa. The first unit to tire did so at an average pressure of -0.55 cm H,O with an average onset frequency of 10.8 spikes . set- ’ (open circle). The second unit started to fire at an average pressure of -3.83 cm H,O and a frequency of 10.3 spikes . set-’ (open square). Thereafter, both units (filled symbols) showed a linear increase in frequency against the pressure generated

/

/

/

/

/time ( msec)

133

1

299

1

507

-6I

774

4

1225 -10 I

1873 -12 I

mouth pressure (cm H20)

Fig. 6. Firing rate of two motor axons at threshold (open symbols) and when pas.sing predetermined pressure values (tilled symbols). Frequency and time are averaged from 6 occluded breaths. Bars indicate 1 SD. Spinal cat.

121

PHRENIC MOTONEURONB DISCHARGE

by the diaphragm, the later recruited unit showing a greater change in frequency per unit change in pressure.3 A summary of the results obtained from the 16 units studied in the two spinal cats is given in table 1. The units are arranged in order of decreasing pressure for recruitment. As stated earlier, the pressure at which the units were recruited ranged between - 0.13 and - 8.43 cm H,O, the latter value being approximately 70% of the maximum pressure attained. For all units, instantaneous discharge frequency was linearly related to pressure, the average increase (slope) being 1.48 spikes’ see-’ (cm H20)- ‘. The correlation coefficients between these two variables, frequency

TABLE 1 Properties of 16 phrenic motoneurones, arranged in order of decreasing threshold pressure of recruitment (Prec) during occluded breaths, of two spinal cats Prec

Tree

Slope

f

(c=r KG)

(“/,D)

(imp/set/cm

(impisec)

f’ (imp/set)

f0 (imp/see)

;I 0s P)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

-0.13 -0.14 -0.16 - 0.22 - 0.46 -0.55 -0.60 -1.07 -2.54 -2.87 -3.39 -3.83 -3.96 -5.80 -5.85 -8.43

6.8 6.1 8.4 1.9 11.6 6.0 12.3 15.5 24.0 21.5 32.6 20.9 39.0 47.0 62.7 63.6

0.61 1.05 0.95 0.71 0.97 1.68 1.14 0.97 1.09 1.10 0.90 2.33 1.26 2.39 3.24 3.34

8.8 10.7 8.3 9.1 10.4 19.2 10.8 9.5 9.7 9.3 1.9 19.3 9.7 8.9 16.6 14.0

11.2 14.2 11.8 11.7 14.0 29.3 13.1 13.2 11.9 11.4 9.2 31.9 11.5 10.8 19.9 20.7

6.1 1.4 5.8 6.3 6.8 10.8 7.6 6.1 6.6 6.9 6.3 la.3 7.4 7.6 12.6 9.9

- 0.956 - 0.968 -0.948 - 0.947 - 0.976 - 0.978 -0.091 - 0.978 - 0,962 -0.939 - 0.928 -0.967 -0.947 - 0.961 -0.943 -0.959

181 188 100 144 105 334 163 238 172 48 100 1.56 660 73 195 100

X SD

- 2.50 2.55

24.5 19.5

1.48 0.87

11.4 3.8

15.4 6.7

7.8 2.0

- 0.958 0.015

185 146

Unit

E;)

Tree (% TI): time of recruitment expressed as percent of inspiratory duration of nonoccluded breaths. Slope: change in frequency per unit change of pressure. f: mean frequency. f’ : peak frequency. f,, : onset frequency. r: correlation coefficient between instantaneous frequency (f) and mouth pressure (P) during occlusions. Size: amplitude of extracellular spike. All values but size and r mean of six observations.

3 During unobstructed inspirations, the relationships between instantaneous frequency and time of each unit matched those shown in fig. 6, indicating that occfusion does not alter the input to the motoneurone pool in these spinal&d cats.

122

s. ISCOE,J. DANKOFF,R. MIGICOVSKY,and c. POLOSA

and pressure, for all units ranged between -0.928 and -0.978. Onset frequency (f,,) averaged 7.8 f 2.0 spikes . set- ‘. The relationship between slope (spikes . see- ’ (cm H,O)- ’ and the threshold pressure of recruitment for each unit is shown in fig. 7. A highly significant (P < 0.001) relationship between recruitment pressure and slope is found ; units recruited at more negative pressures, and therefore recruited later in inspiration, show a greater increase in frequency per unit change in pressure. A relationship (r = 0.525, P < 0.05) was also obtained between onset frequency (f,) and the threshold pressure for recruitment (Prec) of individual units. The equation expressing this relationship is f. = 0.41 Prec + 6.8 and is similar to that obtained by Clamann (1970) for motor units of human brachial biceps.

i

0, I

r =-.a49

OO

1

1

-3

-6

Threshotd

pressure

for recruitment

(cm

HpO)

Fig. 7. Relationshipbetweenthresholdpressureof recruitmentand increasein dischargefrequencyper unit decreasein pressure(slope,see table 1). Spinalcats.

Firing variability of five early recruited (Tree ranging from 1.8 to 6.7% TI) and five late recruited (Tree ranging from 35.9 to 5 1.O% TI) phrenic motoneurones was estimated by calculating the standard deviation of the interspike interval. The SD,

123

PHRENIC MOTONEURONE DISCHARGE

expressed as % of the mean interval, was 0.250 (range 0.214 to 0.292) for the early and 0.120 (range 0.103 to 0.143) for the late units (fig. 8). This difference is significant (P < 0.0005). However, when the analysis of firing variability of early units was restricted to that part of TI over which the late units were active (i.e. beyond 30% TI) all SDS dropped to values in a range (0.068 to 0.151) insigni~cantly different (P > 0.20) from those of later recruited units. The significance of this finding is explained in the discussion.

I

f

I

I

I

2

5

10

20

50

Recruitment

time (“A lnwiratory

100

duration)

Fig. 8. Firing variability in 5 eariy (left) and 5 late (right) recruited phrenic motor axons, Standard deviation of mean interspike interval expressed as a fraction of the mean interspike interval. For explanation, see text.

The discharge frequency of individual units was also studied during COz rebreathing. An example is shown in fig. 9. The time of recruitment of this unit showed a progressive decrease from 10 to 2 % TI. Both mean and peak frequency increased continuously as a function of end tidal COz. However, the increase observed was modest despite a near doubling of CO, tension. In 68 units, the average increases in mean and peak frequencies were 0.13 +O.lO (SD) impulses set- ’ (torr COJ- ’ (0.92 impulses set - 1 (% CO*)- ‘) and .025 _t 0.17(SD) impulses set- ’ (torr COz)- ’ (1.78 impulses set- ’ (% CO,)- I), respectively. The correlation coefficients between these variables and Tree were 0.597 and 0.409 respectively, both highly significant (P < 0.001) and indicate that later recruited units are driven to higher discharge frequencies by increases in the CRDP.

124

S.

ISCOE, J. DANKOFF,

and

R. MIGICOVSKY

C. F’OLOSA

+ +*

++ +

+*+ +

PETCo2 0

4

30 i

I

5

40

50

I

,

6

,

%CO2

10

60

8

9

IO0

’

Fig. 9. Effect of CO, rebreathing on mean and peak frequency (open squares dinate) and recruitment time (crosses, right ordinate) in a single motor axon.

and tilled circles, left orCat with intact neuraxis.

Earlier work on lumbar (Henneman et al., 1965) and phrenic (Nail et al., 1972; Hilaire et al., 1972) motoneurones indicates that the extracellular spike amplitude of a particular motor axon correlates with the motoneurone threshold. Units with the smallest spike amplitudes appear to have the lowest thresholds and, therefore, are the first to start firing and the last to stop during inspiration. The relation between the order of recruitment or dropout and spike amplitude is shown in table 2, which includes data from this and another study. As an example, in three unit filaments the upper left entry (23) indicates that the phrenic motoneurone with the smallest spike was the first to start firing in 23 of 35 cases. In the case of dropout in two unit strands, the lower left entry (33.5, a ‘0.5’ indicating a tie) indicates that in 33.5 of 56 cases the smaller unit was the last to stop firing. With the exception of recruitment in two unit strands, both the order of appearance and disappearance of individual motoneurone tiring is significantly related to spike amplitude; units with small spikes are the first to start and the last to stop firing. This relationship appears to be valid, however, only within a single strand since there was no significant correlation between the variables spike size and recruitment time (% TI) in all units active during eupnoea (r = 0.168, P > 0.10).

Discussion

These observations have revealed that there is a stable order in which phrenic moto-

PHRENIC MOTONIXJRONE DISCHARGE

125

TABLE 2 Relationship between size (rank) and order of recruitment and dropout. For explanation see text TWO-UNIT FILAMENTS size (rank)

size (rank) 1 2

1

order in

order out

1st 2nd x2 = 4.321 P < 0.05

x2 = 3.571 P > 0.05

THREE-UNIT

order in

1st 2nd 3rd

FILAMENTS size (rank) 1 2 23 8 4

9 23 3

x2 = 69.560 P < 0.001

size (rank) 3 3 4 38

order out

1st 2nd 3rd

x2 = 39.429 P 4 0.001

neurones are recruited in the course of inspiration. An individual phrenic unit always discharged when the output of the population (as estimated from mouth pressure during an occluded breath) was above a certain level and never fired when the output was at a slightly lower level. This critical level was constant for a given motoneurone at a fixed end tidal CO2 tension but varied for different units. As a consequence of this fixed order of recruitment, a particular motoneurone will start firing only if all the lower ranking motoneurones in the pool discharge with it. We assume that a unit which starts to tire when the pressure generated by the diaphragm is low is more excitable than one which is activated only at increasingly negative pressure levels. This assumption would explain our findings only if the additional assumption is made that the respiratory input is homogeneously distributed to the phrenic motoneurone pool, so that all units are exposed to an input signal of the same average shape and time course. Our observations provide three lines of evidence to support this latter assumption. First, within a given multiunit strand of phrenic nerve, the order of recruitment of the units is unaffected by COz rebreathing. Units which appeared first in eupnoea also appeared first at elevated levels of CO2 tension (see Results, p. 115). If different neurones were exposed to input waveforms of different shapes and time courses, one might expect that reversals of recruitment sequences would have occurred.

126

s. Isco~,J.DANKOFF, R. MIGIC~VSKY and c. mLos.4

Second, the standard deviation of the time of recruitment of individual units increased significantly as their time of recruitment increased (see Results, p. 115). If all units were exposed to an input waveform (CRDP) whose rate of change of amplitude decreased with time, the crossing of the firing thresholds of low excitability (i.e. late) units would occur on progressively flatter segments of the input waveform. On the further assumption that variability of both the tiring threshold levels and the trajectory of the driving waveform are similar for all motoneurones, a greater range of possible threshold-crossing times results for neurones of low rather than of high excitability. A greater standard deviation of recruitment time for late recruited units would not necessarily be expected if, for instance, their ‘lateness’ was due to their being driven by an input waveform, similar to that impinging on the earlier units, but starting later. Third, late units have a smaller standard deviation of the mean interspike interval than early units (see Results, p. 115). Our explanation for this finding is that late units, due to their low excitability, translate into tiring only the late, plateau portion of the CRDP, the initial, steeper part being subthreshold. That this is the likely explanation is shown by the fact that when the analysis of firing variability of early units was restricted to the same fraction of inspiration over which the late units fired, the firing variability of early units was found to be similar to that of late units. Again, this set of observations is consistent with the hypothesis that both early and late units are exposed to the same CRDP. Our results therefore indicate that phrenic motoneurones, active in eupnoeic breathing, constitute a homogeneous population in terms of time of recruitment. These results do not support an earlier proposal (Hilaire et al., 1972) that the factor determining whether a phrenic motoneurone is ‘early’ or ‘late’ is its connexion(s) to specific early and late inspiratory neurone types in the respiratory centre. A number of other properties of the neurone have been found to be correlated with recruitment time. Units of low threshold tend to have a lower onset frequency, a smaller increase in firing frequency per unit decrease in mouth pressure, and a smaller increase in firing frequency per unit increase in end-tidal COz than late units, Some of these properties are reminiscent of those reported for motor units of human dorsal interosseous muscle (Milner-Brown et al., 1973b). Thus, these early units seem more excitable on the basis of recruitment time, but at the same time less sensitive to a given change in input than late units. Although we have no data concerning the possible mechanisms underlying these properties, Kernel1 (1966) has found lumbar alpha-motoneurones with lower thresholds to intracellular current injection to be less sensitive in terms of an increase in discharge frequency, to a given increase in current. During inspiration, an exponential decline in the number of recruited motoneurones occurs. Phrenic motoneurones behave in a manner strikingly similar to those innervating the triceps surae (Henneman et al., 1965) and soleus (Grillner and Udo, 1971) of cat and human dorsal interosseous (Milner-Brown et al., 1973a) in that the majority (> 75 %) were recruited by the time occlusion pressure had reached

PHRENIC

MOTONBURONE

127

DISCHARGE

half of maximum (table 1, maximum pressure developed E - 12 cm HzO) or 50 % TI had been reached (fig. 3). The increased pressure generated during the second half of the breath, therefore, must be developed through increases in spike frequency (rate coding). Moreover, it is the units recruited later in the inspiration that show the greatest changes in tiring frequency per unit change in diaphragmatic output (fig. 7). Because these later recruited units show greater increases in firing frequency as inspiration proceeds (fig. 6) the overall sensitivity of the motoneurone pool, and therefore of the diaphragm, to the CRDP increases during inspiration because of the recruitment of these units. This is useful since the contraction times of later recruited diaphragmatic motor units are faster than those of earlier ones (Budingen and Yasargil, 1972) and greater increments in discharge frequency would be required to produce sustained tetanic tension. This together with the fact that later recruited motor units generate greater twitch tensions than early ones (Budingen and Yasargil, 1972) is particularly important for the diaphragm not only because of the force-length realtionship of muscle which makes the diaphragm less effective in generating tension at higher lung volumes (shorter muscle length) but also because of the effect of the diaphragm’s increasing radius of curvature and the resulting reduction in effectiveness of conversion of tension to pressure at higher lung volumes (Law of Laplace, Marshall, 1962). Much higher frequencies than those obtained here in response to CO* rebreathing can be elicited by mechanical stimulation of the epipharynx, up to 300 to 400 spikes. set-’ (Nail et al., 1972). In our study, the highest frequencies observed were those during a spontaneous sigh. In one instance, a new unit was recruited during a sigh but despite a further increase of 15 torr CO, did not reappear. These particular effects on phrenic motoneurone discharge pattern may reflect differences in synaptic organization depending on the input pathway involved. Newsom Davis and Plum (1972), for example, have shown that the cough reflex in the cat is mediated through a different pathway in the anterolateral spinal cord than that subserving rhythmic ‘metabolic’ respiration. In order to attain the higher discharge frequencies associated with cough or sigh, the synapses of the descending axons in the pathway involved may be located closer to the cell bodies or even on the initial segment.

References Adrian,

E. D. and D. W. Bronk

(1928). The discharge

in single fibres of the phrenic Adrian,

E. D. and D. W. Bronk

frequency Budingen,

of discharge Pfiigers

H. P. (1970). Activity

of impulses (London)

(1929). The discharge

in reflex and voluntary

H. J. and G. M. Yasargil

des Zwerchfells. Clamann,

nerve. J. Physiol.

in motor

of impulses

contractions.

(1972). Die funktionelle

nerve tibres. Part I. Impulses

66: 81-101. in motor

J. Physiol. Organisation

nerve libres. (London)

Part II. The

67: 119-l 5 1.

der motorischen

Einheiten

Arch. 332: 218-231. of single motor

units during

isometric

tension.

Neurology

(Minneup.)

20:

254-260. Downie,

N. M. and R. W. Heath (1965). Basic Statistical

Methods,

2nd ed. New York,

Harper

and Row.

128

s. ISCOE, J. DANKOFF, R. MIGICOVSKY and c. POLOSA

Eccles, R. M., T. A. Sears and C. N. Shealy (1962). Intra-cellular recording from respiratory motoneurones of the thoracic spinal cord of the cat. Nature (London) 193: 844846. Freund, .I. E. (1962). Mathematical Statistics, 1st ed. Englewood, Prentice-Hall. Gesell, R., A. K. Atkinson and R. C. Brown (1941). The gradation of the intensity of inspiratory contractions. Am. J. Physiol. 131: 659673. Gill, P. K. (1963). The effects of end-tidal CO, on the discharge of individual phrenic motoneurones. J. Physiol. (London) 168: 239-257. Grillner, S. and M. Udo (1971). Recruitment in the tonic stretch reflex. Acta Physiol. &and. 81: 571-573. Henneman, E., G. Somjen and D. 0. Carpenter (1965). Functional significance of cell size in spinal motoneurons. J. Neurophysiol. 28: 560-580. Hilaire, G., R. Monteau and M. Dussardier (1972). Modalitts du recrutement des motoneurones phrtniques. J. Physiol. (Paris) 64: 457478. Kemell, D. (196k).Input resistance, electrical excitability and size of ventral horn cells in cat spinal cord. Science 152: 1637-1639. Mannard, A. and R. B. Stein (1973). Determination of the frequency response of isometric soleus muscle in the cat using random nerve stimulation. J. Physiol. (London) 229 : 275-296. Marshall, R. D. (1962). Relationships between stimulus and work of breathing at different lung volumes. J. Appl. Physiol. 17: 911-921. Milner-Brown, H. S., R. B. Stein and R. Yemm (1973a). The orderly recruitment of human motor units during voluntary isometric contractions. J. Physiol. (London) 230: 359-370. Milner-Brown, H. S., R. B. Stein and R. Yemm (1973b). Changes in tiring rate of human motor units during linearly changing voluntary contractions. J. Physiol. (London) 230: 371-390. Nail, B. S., G. M. Sterljng and .I. G. Widdicombe (1972). Patterns of spontaneous and reflexly-induced activity in phrenic and intercostal motoneurones. Exp. Brain Res. 15: 318-332. Newsom Davis, J. and F. Plum (1972). Separation of descending spinal pathways to respiratory motoneurones. Exp. Neurol. 34: 78-94. Partridge, L. (1966). A possible source of nerve signal distortion arising in pulse rate encoding of signals. J. Theor. Biol. 11: 257-281. Pitts, R. F. (1942). Excitation and inhibition of phrenic motor neurones. J. Neurophysiol. 5: 75-88. Whitelaw, W. A., J.-Ph. Derenne and J. Milic-Emili (1975). Occlusion pressure as a measure of respiratory center output in conscious man. Respir. Physiol. 23: 181-199. Yasargil, G. M. (1964). Die motorische Innervation des Zwerchfells. Helu. Physiol. Pharmacol. Acta 22: 137-147.

Yasargil, G. M. (1967). Syystematische Untersuchung der motorischen Innervation des Zwerchfells beim Kaninchen. Helv. Physiol. Pharmacol. Acta Suppl. XVIII. Younes, M., S. Iscoe and J. Milic-Emili (1975). A method for the assessment of phasic vagal influence on tidal volume. J. Appl. Physiol. 38: 335-343.