INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It is well known that synaptic transmission from muscle afferents in the spinal cord can be depressed by presynaptic inhibition (Eccles et al. 1961, 1962a; Frank and Fuortes 1957; Rudomin et al. 1998). The pathways mediating presynaptic inhibition have been demonstrated to be under supraspinal control in both cat (Enríquez et al. 1996; Rudomin et al. 1986, 1991) and man (Meunier 1999). This is probably the basis for the modulation of presynaptic inhibition of Ia afferents in relation to different motor behaviors, which has been observed in man (Meunier and Morin 1989; Meunier and Pierrot-Deseilligny 1989). Based on human H-reflex experiments, it is suggested that, according to the functional requirements of the motor task, the CNS may increase or decrease the level of presynaptic inhibition of Ia afferents and thereby the central gain of the monosynaptic stretch reflex (Capaday and Stein 1986; Hultborn et al. 1987; Nielsen and Kagamihara 1993).

However, data have recently appeared which suggest that the interpretation of the data from these H-reflex experiments may not be straightforward. Whereas the soleus H-reflex is depressed in the stance phase of walking compared with standing, presumably due to presynaptic inhibition, this is not the case for the monosynaptic stretch reflex (Sinkjaer et al. 1996). There is also evidence that, during co-contraction of antagonistic ankle muscles, the H-reflex is depressed through increased presynaptic inhibition (Nielsen and Kagamihara 1993), while the monosynaptic stretch reflex is not (Nielsen et al. 1994). It is thought that the same pathway mediates both reflexes. Why would the electrically evoked H-reflex be depressed by presynaptic inhibition, whereas the mechanically evoked stretch reflex is not? It has been thought for some time that the only difference between the two reflexes is their sensitivity to fusimotor drive. However, Morgan et al. (1984) observed that conditioning stimulation of gamma dynamic fibers did not increase dramatically the response in Ia afferents to a brief tendon jerk and several studies have pointed out that there are also differences in the composition of the afferent volleys evoked by the two stimuli and the way that the spinal motoneurones are activated by these volleys (Burke et al. 1983, 1984; Morita et al. 1998). Morita et al. (1998) suggested, as a possible explanation of the different sensitivity of the two reflexes to presynaptic inhibition, that a mechanical stimulus, which induces repetitive discharge of the Ia afferents, might lower the sensitivity of the synapses to presynaptic inhibition compared with the synchronous Ia discharge underlying the H-reflex. The present experiments were designed to investigate this hypothesis further in the cat.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The experiments were performed in 19 adult male cats (2.5-3.5 kg). All surgery and experimental protocols were conducted according to the National Institutes of Health Guide for the care and use of laboratory animals (National Institutes of Health publication no. 86-23, revised 1985). During surgery, animals were deeply anesthetized with halothane (2.5% in a mix of 40% oxygen in NO₂) while, during the actual experiment, anesthesia was maintained with alpha-chloralose (80 mg/kg). Gas anesthetics were replaced with a mix of 40% oxygen in air. Supplemental anesthesia, if required, was given as 5 mg/kg of pentobarbital (Mebumal, SD Denmark) every 2 h. Adequacy of anesthesia was determined by assessing that withdrawal reflexes were absent, that the pupils were constricted, and that the blood pressure was between 80 and 120 mmHg even during high-intensity electrical stimulation of cutaneous nerves (Sural, 50T, 200 Hz).

The trachea, two forelimb veins, and one carotid artery were cannulated for gas, fluid administration, and monitoring of the blood pressure. A solution of 0.5 M of glucose and 0.2 M NaHCO₃ (5 ml/h) was continuously infused throughout the experiment. If blood pressure dropped below 80 mmHg, infusion of Macrodex (300 mg/h) or Noradrenalin (starting dose 0.02 ml/h) was established. Body temperature was kept within 36-38°C with servo-controlled heating devices.

The longest possible length of the motor nerves medial gastrocnemius (MG) and lateral gastrocnemius plus soleus (LG-S) were dissected free, but left in continuity with the muscles (see Fig. 1). These nerves were mounted separately or together (GS nerves) on flexible hook electrodes (silver wire 0.29 mm diam) for electrical stimulation. In addition, the following nerves were dissected, cut, and mounted for stimulation to identify the motoneurones from which intracellular recordings were made from (see following text): quadriceps, semimembranosus and anterior biceps (SmAB), posterior biceps and semitendinosus (PBSt), flexor digitorum hallucis longus (FDHL), and deep peroneus (DP, tibialis anterior plus extensor digitorum longus).

View larger version (12K): [in this window] [in a new window]   Fig. 1. Diagram of the methods used. Gastrocnemius-soleus (GS) monosynaptic reflexes were evoked by stimulation of the lateral gastroc-soleus (LG-S) motor nerve or muscle stretch and recorded by wire electrodes inserted in the soleus muscle (EMG). The reflexes were conditioned by stimulation of the posterior biceps-semitendinosus motor nerve (PBSt). After paralysis, excitatory postsynaptic potentials (EPSPs, Intracellular) were evoked by LG-S or MG stimulation or muscle stretch and conditioned by PBSt stimulation. Afferent activity was recorded at the dorsum of the spinal cord (CDP). Further details in the text.

After measuring the maximal physiological length of the triceps surae muscle, the tendon was separated from the calcaneous bone and attached to a servo-controlled puller (Series 300B Lever System, Cambridge Technology) (see Enríquez et al. 1996; Lennerstrand 1968).

A laminectomy was performed to expose the lumbar spinal cord from L7 to L5. All exposed tissues were covered with paraffin oil and kept at 37°C by radiant heating.

Incoming volleys (CDP), elicited by electrical stimulation of the peripheral nerves or by stretching the muscle, were recorded with a platinum ball electrode, which was placed on the spinal cord ventral to the L7 dorsal root; the reference electrode was located in the nearby back muscles. The threshold for electrical stimulation (T) was determined as the minimum voltage intensity capable of producing an incoming volley. The CDP and the EMG signals were amplified with low-noise high-gain differential amplifiers (band-pass filters 1 Hz-10 kHz).

Reflex recording

Reflexes evoked by stretching the triceps surae or by stimulation of the LG-S nerve were recorded with wire electrodes (Teflon-coated platinum-iridium 25 µm, Cooner Wire, Chatsworth, CA) inserted directly into the soleus muscle. These reflexes were elicited by stretching the muscle  2.0 mm from 4 mm of the maximum physiological length. The stretches had a triangular wave form with a rise time of 2-250 ms. It is assumed that these stretch parameters induce activation of the majority of spindle afferents with low or negligible activation of afferent fibers from tendon organs (Enríquez et al. 1996; Lennerstrand 1968; Proske et al. 1992; Wood et al. 1994). Electrically induced reflexes were elicited by stimulation of the LG-S nerve with pulses of 0.1 ms and intensities of 1.3 T. Stretch and electrically evoked reflexes were alternated with each other and conditioned by preceding electrical stimulation of the flexor PBSt nerves (1.20-5 T; 5 shocks, 200-300 Hz). It was ensured that the electrically and stretch-induced reflexes had the same size in the control situation without conditioning stimulation. This was done by adjusting the intensity of the electrical stimulation until responses of the same size as those evoked by the stretch were obtained.

Intracellular recording

After the initial EMG reflex recordings, the preparations were immobilized (pancuronium bromide, Pavulon) and artificially ventilated. Expired PCO₂ was continuously monitored and kept at 4% by adjusting the volume of the flowing gas (1800-2000 ml/min). Bilateral pneumothorax and rigid clamping of the L4 vertebra was made to ensure mechanical stability of the preparation for intracellular recording from spinal motoneurones.

Intracellular recording was made from motoneurones identified by antidromic activation from MG or LG-S nerves with KAc-filled (2 M) microelectrodes (1.2 µm, 5 M) or with CNQX-Kac-filled microelectrodes (0.1 and 2 M, respectively) coupled to a microelectrode amplifier (Axoclamp 2A, Axon Instruments). Only motoneurones with a stable membrane potential of at least 50 mV were included in the analysis. The motoneurones were not characterized according to motoneurone types. Stretch- and electrically induced excitatory postsynaptic potentials (EPSPs) were conditioned by preceding stimulation of the PBSt nerve. The interval between conditioning stimulation and the test EPSPs was carefully adjusted throughout the experiments. The effect of the conditioning stimulation was investigated for stretch-induced EPSPs of different duration as well as electrically evoked EPSPs evoked by either single stimuli or trains of stimuli. In all recorded motoneurones it was confirmed that the PBSt stimulation had no effect on the membrane potential or a small pulse (same size as the investigated EPSPs; duration, 5 ms) injected through the microelectrode at the investigated intervals.

Data analysis

In all cases, conditioned and unconditioned responses were alternated with each other at a repetition frequency of about 1 Hz. At least 16 recordings of each alternative were obtained. All signals were digitized (10 KHz) and analyzed using the SCRC Data Capture and Analysis Software System (Detillieux G. R., The University of Manitoba, Winnipeg, Manitoba, Canada). EPSPs were measured from base to peak. The area of the EPSPs was also determined in 2- and 5-ms time intervals. Statistically significant differences between control and conditioned potentials were determined using a paired Students t-test. Pooled data from different experiments were compared using a one-way ANOVA and a test of difference in proportions.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Electrically and mechanically induced reflexes are not equally sensitive to presynaptic inhibition

Figure 2 demonstrates that a reflex elicited by electrical stimulation (H-reflex) of the GS motor nerve (1.2 T; filled circle) is more sensitive to presynaptic inhibition than a reflex elicited by stretching the GS muscle (0.25 mm, 125 mm/s; open circle). In Fig. 2A it is shown that preceding stimulation of the PBSt nerve (1.2 T, 5 shocks, 225 Hz, 40 ms before) depressed the electrically evoked monosynaptic reflex (MSR) to 34% of the size of the control reflex (compare dotted and continuous lines), while the mechanically evoked MSR was only depressed to 76% of control. Figure 2B shows that the mechanically evoked MSR was significantly (P < 0.05) less affected than the electrically evoked MSR for conditioning-test intervals longer than 10 ms. In addition, the electrically evoked MSR was significantly (P < 0.05) depressed for 200 ms after the conditioning stimulation, whereas the mechanically evoked MSR was significantly (P < 0.05) depressed for only 60 ms. This late long-lasting inhibition can be explained by presynaptic inhibition (Eccles et al. 1961). At short conditioning-test intervals between 2 and 5 ms, an equal depression of the H-reflex and the stretch reflex was observed. The inhibition at these short conditioning-test intervals is accepted to be postsynaptic in origin (Eccles et al. 1961).

View larger version (21K): [in this window] [in a new window]   Fig. 2. Effect of PBSt stimulation on the electrically (H-reflex) and mechanically (stretch reflex) evoked monosynaptic reflexes. A: averaged soleus-EMG illustrating the H-reflex (top trace) and the stretch reflexes, in the control condition (Test, discontinuous line) and following conditioning by PBSt stimulation (Conditioned, continuous line; 40 ms before, 5 shocks, 225 Hz, 1.2 T). Incoming volley and muscle length are illustrated for the H- and stretch reflexes, respectively. B: time course of the depression of the H- (filled circle) and the stretch reflexes (open circle) following conditioning stimulation of the PBSt (1.85 T, other parameters same as before). C: intensity curve of the depression of the reflexes by preceding stimulation to PBSt at 40-ms interval. Data points are mean ± SE expressed as a percentage and each is from 25 samples. All records from experiment 181096.

Figure 2C (from the same experiment as Fig. 2B) illustrates the relationship between the depression of the two reflexes and the intensity of the conditioning stimulation at a constant conditioning-test interval of 40 ms. The electrically evoked reflex was abolished already at an intensity of 1.3 T and was significantly (P < 0.05) more depressed than the stretch reflex up to stimulation intensities of 2.5 T. Similar observations were obtained in all six preparations in which the depressions of the stretch- and electrically induced reflexes were compared. At 1.6 T, conditioning stimulation of PBSt (conditioning-test interval 30-60 ms; 5 shocks, 300 Hz) depressed the electrically evoked reflex to 28.7 ± 33.5% (mean ± SD), whereas the mechanically evoked reflex was only depressed to 50.7 ± 26%. At 2T, a depression to 24.5 ± 32.4% of the electrically evoked reflex and to 49.2 ± 27.6% of the mechanically evoked reflex was observed. At both intensities, the depression of the electrically evoked reflex was significantly larger than the depression of the stretch-evoked reflex (P < 0.05).

Different sensitivity of the first and last components of the EPSPs to presynaptic inhibition

To investigate whether there is a similar differential depression of postsynaptic potentials evoked by mechanical and electrical activation of the monosynaptic reflex pathway, the preparations were paralyzed and intracellular recordings were made from antidromically identified triceps surae motoneurones. Figure 3 illustrates data from an LG-S motoneurone. EPSPs were elicited by electrical stimulation of the GS nerve or by stretching the triceps surae muscle with (continuous line) and without (dashed line) preceding PBSt stimulation. Conditioning stimulation of PBSt (55 ms before, 1.3 T, 5 shocks, 300 Hz) evoked a clear depression of an EPSP evoked by either a single electrical stimulus (Fig. 3A; 1.2 T) to the GS nerve or by a fast stretch (ramp of 0.11 mm; 55 mm/s) to the muscle (Fig. 3B). A depression was also seen for the first of a series of EPSPs elicited by a train of electrical stimuli (27 stimuli; 250 Hz) to the GS nerve (Fig. 3, C and C1) as well as the initial part of the complex EPSP evoked by a slow stretch (0.2 mm, 1.6 mm/s) to the muscle (Fig. 3, D and D1). In contrast, both in the electrically evoked and in the stretch-evoked EPSPs, the last part of the potentials were much less depressed by the PBSt stimulation (Fig. 3, C2 and D2, respectively). Comparing the depression of the first and last EPSPs, a difference of 17% was found for the electrically evoked EPSPs and of 18% for the stretch-induced EPSPs.

View larger version (20K): [in this window] [in a new window]   Fig. 3. Effect of PBSt stimulation on EPSPs elicited by electrical nerve stimulation or muscle stretch. A, top traces: EPSPs elicited by a single shock to the GS nerve (1.2 T) in the control situation (test, broken line) and following preceding activation to PBSt nerve (continuous line; 55 ms before 5 shocks, 250 Hz, 1.3T); B: same as A, but EPSPs were elicited by a fast muscle stretch (0.11 mm, 55 mm/s). C: same as A, but postsynaptic potentials were elicited by 27 shocks (250 Hz) to the peripheral nerve. D: EPSPs produced by slow ramp stretch (0.2 mm, 1.6 mm/s) of the muscle. C1, D1 and C2, D2: 10-ms details of C and D illustrating the beginning and last part of the composite EPSPs, respectively. Arrowheads indicate timing of measurement of the potentials. Recordings of cord dorsum and muscle length are illustrated below the EPSP traces. Calibration bar in B applies also to A, the one in D also to C, while the one in D2 also applies to C1, C2, and D1. Voltage calibration in D2 applies to all intracellular records. Data from U1-160896.

The difference in depression between the first and the last EPSP was analyzed in 17 motoneurones (see Fig. 4A; left histograms). To exclude that the different latency of the first and the last EPSP evoked by the train of stimuli was responsible for the different effect of the PBSt stimulation, the interval between the conditioning stimulus and the train of pulses to the GS nerve was adjusted (in separate trials) to obtain the same interval from the conditioning stimulation to the last as to the first EPSP (see Fig. 5A). The average difference in the PBSt-induced depression of the last and the first EPSP in the train was 11 ± 15% (range 76 to 33%) when the conditioning-test interval was 25 to 100 ms (1.3 to 2 T). A similar analysis was performed for the stretch-induced EPSPs in 12 motoneurones (35 measurements; Fig. 4A; right histograms). The interval from the conditioning stimulation was adjusted in the same way as for the electrically evoked EPSPs to obtain the same conditioning-test interval for the last as for the first part of the stretch-evoked EPSPs. The average difference in the depression of the last and the first component of the composite stretch-induced EPSP was 6 ± 10% (range 25 to 11%). By the use of a paired t-test, a statistically significant difference in depression between the first and last components was found both for the electrically induced EPSPs (first component: 77 ± 10%, last component: 88 ± 16%; P < 0.001) and the stretch-induced EPSPs (first component: 75 ± 15%; last component: 81 ± 12%; P = 0.002). There was no difference in the depression of the first EPSP evoked by a train of electrically stimuli and the first part of the stretch-evoked EPSPs (compare columns 1 and 3 in Fig. 4A).

View larger version (17K): [in this window] [in a new window]   Fig. 4. Comparison of the effect of PBSt stimulation on the first and last of a series of EPSPs. A: pooled data from all recorded motoneurones showing the size of the conditioned EPSP as a percentage of the size of the control EPSP. The conditioning stimulation was applied to the PBSt motor nerve in all cases. Measurements were made for the first and last EPSP in the series of electrically evoked EPSPs or for the initial and last part of the EPSP induced by slow muscle stretch. N = number of motoneurones, from which recordings were obtained from. NM = total number of measurements. Stars indicate the statistical significance of the difference between the last and first potential. B: the area of the conditioned EPSP expressed as a percentage of the size of the control EPSP. Measurements were made in 2- to 5-ms intervals for both the electrically (closed circles) and mechanically (open circles) evoked EPSPs. ***P < 0.001, **P < 0.01, *P < 0.05.

View larger version (19K): [in this window] [in a new window]   Fig. 5. Time course of the effect of PBSt stimulation on the first and last of a series of EPSPs. A: actual traces for the comparison in the depression of the first vs. last EPSP elicited by a train (5 shocks, 1.2 T, 200 Hz) in relation to the latency from conditioning stimulation of PBSt (5 pulses, 300 Hz, 2.5 T). B: size of the conditioned EPSP as a percentage of the control EPSP at different intervals after the conditioning PBSt stimulation for the first (filled circles) and last (open triangles) EPSPs in the train. Each point is an average of 57 traces. Data from U3-021298. C: relationship between the amplitude of the conditioned EPSP and the amplitude of the alternated test EPSP, for 2 different motoneurones. Squares U1-170997 conditioned by PBSt 1.6 T, 5 times, 60 ms before; circles U1-100895 conditioned by PBSt 2 T, 5 times, 34 ms before. Amplitude of the test EPSP was changed by varying the stimulation intensity to the peripheral nerve. Each point is the average of 20 traces elicited at 0.8 Hz. Equality line is indicated.

The area of the electrically and mechanically evoked EPSPs was also measured in 2- and 5-ms intervals (Fig. 4B). For the population of EPSPs, only the initial 5 ms of the EPSPs were significantly depressed by PBSt stimulation. For the stretch-induced EPSPs, a tendency toward facilitation was even seen for the part of the EPSP from 5 to 15 ms after its onset (Fig. 4B, open circles).

Time course of the depression of EPSPs by PBSt stimulation

Figures 5A and B show the time course of the effect of PBSt stimulation (5 shocks, 2.5 T, 300 Hz) on the first and last EPSP in a series of EPSPs in a MG motoneurone evoked by a train of stimuli (5 shocks, 200 Hz, 1.2 T) applied to the MG motor nerve. The PBSt stimulation depressed the first EPSP in the train to 70% of the control value, whereas a depression to only 94% was seen for the last EPSP (see also Fig. 4A). This difference in depression was also noted when the interval from the conditioning PBSt stimulation was adjusted to take into account the 22-ms interval between the two EPSPs (compare top and bottom traces in Fig. 5A). As seen from the graph in Fig. 5B, the first EPSP in the train was depressed for all intervals up to around 150 ms, consistent with presynaptic inhibition of MG Ia afferents. However, the last EPSP in the train was not depressed within these intervals after the conditioning stimulation to PBSt.

Size of the EPSP and presynaptic inhibition

One possible explanation for the difference in the effect of PBSt stimulation on the first and last EPSP in a train is that the EPSPs may differ in size and that potentials of different size may be differently sensitive to presynaptic inhibition (Zengel et al. 1983). To address this possibility, an analysis of the relationship between the amplitude of the control EPSP and the effect of the conditioning stimulation was made. The amplitudes of the test EPSPs were varied by changing the intensity of the GS stimulation. Figure 5C shows the relationship between the amplitudes of the control and conditioned EPSP for two different motoneurones (marked by squares and circles, respectively). The intensity of the conditioning PBSt stimulation was 1.6 and 2 T, whereas the conditioning-test intervals were 60 and 34 ms, respectively. For both motoneurones there was a linear relationship between the control EPSP and the conditioned EPSP [slope (m) = 0.91, r = 0.99; m = 0.67, r = 0.99, respectively]. A similar relationship was found in all tested motoneurones, two other GS motoneurones, two flexor digitorum hallucis longus motoneurones (conditioned by PBSt), one quadriceps motoneurone (conditioned by deep peroneal stimulation), and one deep peroneal motoneurone (conditioned by PBSt). In all of them, the slope varied from 0.67 to 0.89 and the linear correlation factor (r) was 0.9. This suggests that the amount of presynaptic inhibition is independent of the size of the test EPSP. The different sensitivity to presynaptic inhibition of the first and the last EPSP therefore cannot be explained by their different sizes.

Effect of proceeding group I activity on the sensitivity of EPSPs to presynaptic inhibition

To assess the effect of preceding activity in afferent fibers on the magnitude of presynaptic inhibition, intracellular recordings were performed from 16 antidromically identified motoneurones. Trains of EPSPs were induced either by stimulation of the homonymous or the heteronymous nerve and the effect of PBSt stimulation on the test EPSP was assessed at different intervals after these trains. Figure 6A shows the depression of the test EPSP evoked by the PBSt stimulation when the EPSP was preceded by a train of stimuli to the homonymous or heteronymous nerve. In this MG motoneurone, stimulation of the PBSt nerve (2.5 T, conditioning-test interval: 50 ms) depressed the test EPSP to 71.7% of its control size (3.2 mV). When the test EPSP was preceded by a train of five pulses (200 Hz; 9 ms interval) to the homonymous (MG) nerve, PBSt stimulation depressed the test EPSP to only 81.2% of its control value (3.06 mV as the train induced a reduction of the test EPSP also). Presynaptic inhibition was, in other words, reduced by 9.6% when a train of stimuli to the homonymous nerve preceded the test EPSP. When the test EPSP was applied 18 ms after the train of stimuli, an even larger reduction (by 30%) of presynaptic inhibition was observed (lower group of traces). In contrast, a preceding train of stimuli applied to the heteronymous (LG-S) nerve increased the depression evoked by the PBSt stimulation by 5.1 and 7.2% at the same intervals (right traces). Figure 6B illustrates the change in presynaptic inhibition at all tested intervals after the trains applied to the homonymous (MG, squares) and heteronymous (LG-S, circles) nerves for this motoneurone. At intervals between 8 and 25 ms from the preceding train to the homonymous nerve, a reduction in the presynaptic inhibition by more than 5% was noted, whereas a similar train applied to the heteronymous nerve increased the presynaptic inhibition by a similar amount. Of 16 tested motoneurones, a similar pattern was seen in 10 motoneurones; i.e., a train of stimuli applied to the homonymous nerve decreased presynaptic inhibition, whereas no change or an increase in presynaptic inhibition was seen following a train of stimuli applied to the heteronymous nerve. In two motoneurones, trains applied to both the homonymous and the heteronymous pathway reduced presynaptic inhibition. In the remaining four motoneurones neither stimulation of the homonymous nor the heteronymous nerve had any effect on the amount of presynaptic inhibition.

View larger version (26K): [in this window] [in a new window]   Fig. 6. Effect of preceding activation of homonymous and heteronymous muscle afferents on the sensitivity of EPSPs to presynaptic inhibition. A: intracellular recordings from an MG motoneurone. EPSP was evoked by stimulation of the MG nerve and conditioned by PBSt stimulation (Control: broken lines; conditioned: continuous lines). In the left set of traces the test EPSP was preceded by a train of stimuli applied to the MG motor nerve, whereas in the right set of traces the train was applied to the heteronymous LG-S motor nerve. Two intervals between the last EPSP in the train and the test EPSPs are illustrated (9 and 18 ms, respectively). B: time course of the difference in the effect of PBSt stimulation on the test EPSP with and without a preceding train of stimuli applied to either the MG (squares) or the LG-S motor nerve (circles). C: box chart of the difference in PBSt-induced inhibition of the test EPSP with and without a preceding train of stimuli applied to the homonymous or heteronymous nerve for the whole sample of 16 motoneurones. Data are grouped at the indicated time bins from the last synergistic homonymous or heteronymous EPSP. Box represents the 25th and 75th percentiles, and whiskers the 5th and 95th percentiles of the sample. Stars indicate the range of the data in each bin, while filled squares are the average of the sample.

Quantitatively, an average reduction of presynaptic inhibition of 7.7 ± 9.2% was found when the test EPSP was evoked 5-10 ms after a train of stimuli applied to the homonymous nerve for the 16 recorded motoneurones (12 MG and 4 LG-S; 31 measurements; P < 0.001; Fig. 6C). When the test EPSP was evoked 10-20 ms after the train, a reduction of 9.8 ± 11.5% in presynaptic inhibition was observed (19 measurements; 15 motoneurones; P < 0.05). At intervals of 20-30 ms a nonsignificant reduction of 3.8 ± 13.4% (10 measurements, 9 motoneurones) was observed, and at intervals longer than 30 ms the reduction was 1.1 ± 14.3% (11 measurements, 4 motoneurones; nonsignificant). In contrast, there were no significant changes in the amount of presynaptic inhibition when the train of stimuli were applied to the heteronymous nerve regardless of the interval between the train and the test EPSP (Fig. 6C). This suggests that the depression of presynaptic inhibition by preceding afferent activity is restricted to the afferents from which the test EPSP is evoked and that the time course of the effect is on the order of 20-30 ms.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the present study we have confirmed in the alpha-chloralose-anesthetized cat that soleus H-reflexes are less depressed by presynaptic inhibition evoked by flexor nerve stimulation than soleus stretch reflexes as has been demonstrated previously in man (Morita et al. 1998). In the sample of intracellular recordings from lumbar motoneurones, we found that the initial part of the electrically and mechanically evoked EPSPs were equally depressed by presynaptic inhibition, whereas later parts were significantly less depressed. We finally provided evidence that peripheral afferents, which have been previously activated (within 20 ms), are less sensitive to presynaptic inhibition. We believe that this observation may partly explain the different sensitivity of H-reflexes and stretch reflexes to presynaptic inhibition.

Previous activation of peripheral afferents and differential sensitivity of H-reflexes and stretch reflexes to presynaptic inhibition

The afferent volleys evoked by muscle stretch and by electrical stimulation of peripheral afferents differ significantly (Burke et al. 1983, 1984). Following electrical stimulation of peripheral nerves, each afferent discharge only once and the afferent volley is relatively synchronized with little temporal dispersion. Following muscle stretch, each afferent may discharge several times at short intervals (5 ms) and the temporal dispersion of the volley is large (Burke et al. 1983). When the intensity of electrical stimulation and muscle stretch are adjusted so as to evoke H-reflexes and stretch reflexes of similar size and shape (cf. Fig. 2), the underlying afferent volleys evoked by the two stimuli will thus be very different. Due to the temporal dispersion of the afferent volley underlying the stretch reflex, a combination of spatial and temporal summation of EPSPs contributes to its size, whereas spatial summation is mainly involved in the generation of the H-reflex. Short-lasting composite EPSPs evoked by brief muscle stretches were equally sensitive to presynaptic inhibition as electrically evoked EPSPs (cf. Fig. 3, A and B). This was also true for the initial part of the long-lasting slowly rising EPSPs evoked by similar slow stretches as those used to evoke the stretch reflexes in this study. However, the last part of these EPSPs, as well as the last part of similarly long-lasting EPSPs evoked by a train of electrical stimuli, were less depressed by presynaptic inhibition (cf. Fig. 3, C and D). In these latter cases, temporal and spatial summation of discharges in the same afferents must have contributed to the size of the later part of the EPSPs, whereas this is not the case, or at least to a much lower extent, for the initial part of the long-lasting EPSPs and for the short-lasting EPSPs evoked by very brief stretches and single electrical stimuli. We therefore believe that previous activity in the afferents and temporal summation may decrease the effect of presynaptic inhibition and explain the differential effect of presynaptic inhibition on H-reflexes and stretch reflexes. Two points are of importance in this relation. First, it has been shown that reflexes are evoked toward the end of the rising phase of the composite EPSP (Burke et al. 1984). The stretch reflexes in this study and in the study by Morita et al. (1998) were in other words evoked at the point of the EPSPs at which presynaptic inhibition was shown to have little effect. Second, the seemingly small difference in the effect of presynaptic inhibition on the electrically and mechanically evoked EPSPs (Fig. 3) is not in conflict with the very large difference in the effect of presynaptic inhibition on H-reflexes and stretch reflexes (Fig. 2). Even small differences in the size of EPSPs may lead to very pronounced differences in the size of monosynaptic reflexes, since a small decrease in the EPSP may bring several motoneurones below their firing threshold and thereby produce a large decrease in the size of the reflex (Eccles et al. 1961).

We would also like to point out that differences in the sensitivity to presynaptic inhibition of the EPSPs following previous afferent activity is not the only possible contributing mechanism to the observed differential effect of presynaptic inhibition on stretch reflexes and H-reflexes. Although we did make sure that the H-reflexes and stretch reflexes were of similar size, we cannot be sure that the same motoneurones were activated by the two inputs. However, we find it unlikely that this should explain the different effect of presynaptic inhibition on the two reflexes. First, it has been shown that stretches and electrical peripheral nerve stimuli both recruit motoneurones according to Henneman's size principle (Harris and Henneman 1979; Lüscher et al. 1989). Second, as shown by Zengel et al. (1983), presynaptic inhibition does not seem to be differentially distributed in the motoneurone pool.

Another possibility is that the two different inputs recruit the motoneurones with a different gain (Kernell and Hultborn 1990). Given the different temporal dispersion of the volleys it would not be surprising if the relation between input and output would differ for the two reflexes. However, as shown in the study by Morita et al. (1998) and in the present study, postsynaptic inhibition was equally effective for the two reflexes (cf. Fig. 2C). This would not be the case if differences in recruitment gain were responsible for the different sensitivity to presynaptic inhibition.

A final possibility is that the stretch reflex, rather than the H reflex, is evoked to a higher extent by afferents other than Ia afferents and/or pathways other than the monosynaptic Ia pathway. Neither of the reflexes may be assumed to be purely monosynaptic (Burke et al. 1984); several other pathways may contribute to their size, including nonmonosynaptic group I pathways and group II pathways (Behrends et al. 1983a,b; Eccles and Lundberg 1959; Jankowska et al. 1981a; Jankowska and McCrea 1983; Kniffki et al. 1981; Lundberg et al. 1987). These other pathways would mostly influence the later parts of the EPSPs. It has been demonstrated that group II pathways are less sensitive to presynaptic inhibition evoked by flexor group I stimulation than the monosynaptic Ia pathway (Eccles et al. 1962a; Riddell et al. 1995). With the experiments in the present study we cannot exclude that this may explain the different sensitivity of the two reflexes. Further experiments are necessary to address this issue.

Which mechanism explains the reduced sensitivity to presynaptic inhibition after previous peripheral afferent activation?

Regardless of whether the decreased presynaptic inhibition following previous activation of peripheral afferents is responsible for the differential sensitivity of stretch reflexes and H-reflexes, the findings are of interest in relation to the mechanisms responsible for modulation of presynaptic inhibition. One possible mechanism is that the train of stimuli applied to the peripheral nerves depressed transmission in the pathway mediating presynaptic inhibition from flexor group I afferents. In the cat the presynaptic inhibitory pathway from flexors involves a series of 2 interneurones, of which the last is GABAergic (Curtis et al. 1986; Curtis and Lodge 1982; Eccles et al. 1962a, 1962b; Jankowska et al. 1981b; Rudomin et al. 1986) and makes axoaxonic synapses on the axons of the afferent fibers from the extensors (Rudomin et al. 1998; Solodkin et al. 1984). It would be a possibility that the train of stimuli applied to the triceps surae motor nerves resulted in inhibition of one of the interneurones in this pathway. However, activation of ankle extensor Ia afferents is rather ineffective in producing presynaptic inhibition of flexor Ia afferents (Eccles et al. 1962b; Enríquez-Denton et al., 2000). Furthermore, we found that only preceding stimulation of the homonymous motor nerve produced decreased presynaptic inhibition, whereas preceding stimulation of heteronymous ankle extensor nerve had no effect (Fig. 6). This would be difficult to reconcile with an effect on transmission in the presynaptic inhibitory pathway from flexors, since the different ankle extensors would be expected to produce similar effects in this relation.

We consequently find it more likely that the observed decreased sensitivity to presynaptic inhibition following previous afferent activity is due to a mechanism intrinsic to the activated afferents and synapses. Presynaptic inhibition works by activation of GABA_a and/or GABA_b receptors. Activation of GABA_a receptors produces changes in the chloride conductance, which lead to depolarization and changes in the excitability of the terminals of the afferent fibers (Rudomin and Schmidt 1999). Activation of GABA_b receptors decreases the calcium influx and thereby the release of transmitter substance from the terminals (Edwards et al. 1989; Jiménez et al. 1991; Lev-Tov et al. 1988) without affecting the potential or excitability of the afferent fibers (Quevedo et al. 1992). Following preceding activity in Ia afferents there is an accumulation of calcium in the terminals and as a consequence of this an increased transmitter release probability, which lasts for around 20-50 ms (Borst and Sakmann 1998; Curtis and Eccles 1960; Cuttle et al. 1998; Peshori et al. 1998). This corresponds rather well to the period in which we observed that the effect of presynaptic inhibition on the EPSPs was decreased (cf. Fig. 6) and suggests that there may be a causal link between the two observations. As can be seen in the following numerical example, the ability of presynaptic inhibition to depress transmitter release from the terminals is simply reduced when there is more calcium present in the terminals. Assume that a unit x = 1 of calcium is released by the first presynaptic action potential. The resulting EPSP will depend on a power such as 4 of the calcium released, so EPSP = x⁴ = 1, as well. If presynaptic inhibition reduces the calcium released by 10% (Eccles et al. 1961), then x = 0.9 and EPSP = 0.9⁴ = 0.65. With repetitive stimulation there will be homosynaptic depression of calcium release, but some residual calcium from previous pulses will be present in the terminals. Since the EPSP is about the same, the total calcium will be about the same. For example, if the residual calcium is 0.3 and the calcium released is 0.7, the EPSP will be the same. However, presynaptic inhibition will reduce the released calcium to 0.7 × 0.9 = 0.63. The total calcium will then be 0.63 released plus the residual 0.3, i.e., x = 0.93 and the EPSP = 0.93⁴ = 0.75; i.e., it will be less affected by the presynaptic inhibitory volley. Obviously, a different power than 4 or a different balance than 0.3 and 0.7 can be used and the values would change somewhat, but the qualitative result will be the same. As can be seen in this numerical example, as the residual calcium builds up with repetitive stimulation, the effect of presynaptic inhibition will decrease. This may thus be the underlying mechanism for the different effect of presynaptic inhibition on H-reflexes and stretch reflexes as outlined in the previous section.

Finally, we have demonstrated that the amount of presynaptic inhibition for a particular motoneurone is constant and independent of the size of the unconditioned EPSP (Fig. 5). Zengel et al. (1983) found that, in absolute terms, presynaptic inhibition was small in motoneurones with small EPSPs (type F), whereas it was large in motoneurones with large EPSPs (type S). Similar to us, they thus found a constant relation between the amount of presynaptic inhibition, when expressed as a percentage of EPSP size, and the size of the EPSP measured in various motoneurones. This might imply that presynaptic inhibition is also constant for individual Ia afferents acting on the same motoneurone. This is indeed what Clements et al. (1987) found for some motoneurones (their Fig. 5A), but not for all. When testing EPSPs evoked from activation of several afferents, as in our study, it would seem likely that such variations among individual Ia afferents are insufficient to be noticed. Furthermore, our observation of constant presynaptic inhibition for EPSPs of different sizes would not be inconsistent with the findings by Clements et al. if the Ia afferents investigated by them were evenly distributed in terms of electrical threshold.

Methodological and functional considerations

Comparison of the H-reflex and stretch reflex have previously been used to infer changes in fusimotor drive as only the latter reflexes are influenced by the sensitivity of the muscle spindles. However, as argued already by Burke et al. (1983, 1984), there are significant differences in the afferent volleys evoked by mechanical and electrical stimuli and the way that they activate the spinal motoneurones. The findings in the present paper stress this. As pointed out by Morita et al. (1998), the different effect of presynaptic inhibition on stretch reflexes and H-reflexes highlights that the two reflexes represent different ways of "probing" the CNS. Changes in presynaptic inhibition are easily demonstrated with the use of H-reflexes, but may go unnoticed when using stretch reflexes (Nielsen et al. 1994; Sinkjær et al. 1996). Functionally, the findings in the present study are consistent with the idea put forward by Morita et al. (1998) that the differential effect of presynaptic inhibition on H-reflexes and stretch reflexes may be seen in relation to the discharge of Ia afferents. The decreased sensitivity to presynaptic inhibition lasted around 20-30 ms following previous activation of the afferents. There will thus be no decrease of the sensitivity to presynaptic inhibition when Ia afferents discharge slower than around 30-50 Hz. When the Ia afferents discharge at higher frequencies the sensitivity to presynaptic inhibition will decrease and when they discharge at 200 Hz (i.e., at 5-ms intervals) this effect will be maximal. As already mentioned this is the case when a sudden external muscle stretch is imposed (Burke et al. 1983). As one possible interpretation of our data, presynaptic inhibition may thus effectively gate and modulate the normal peripheral feedback evoked in relation to voluntary movements but have a much less significant effect on activity evoked by sudden external perturbations. The frequency of the afferent firing may thus code the effect of presynaptic inhibition and the internally and externally generated feedback may at the same time be differentially modulated. Further experiments are clearly necessary to demonstrate whether this simple control paradigm exists.