Relative contribution of Ia inhibitory interneurones to inhibition of feline contralateral motoneurones evoked via commissural interneurones

Authors


Corresponding author E. Jankowska: Department of Physiology, Medicinaregatan 11, Box 432, 405 30 Göteborg, Sweden. Email: elzbieta.jankowska@physiol.gu.se

Abstract

The aim of the study was to examine to what extent the crossed inhibition of hindlimb lumbar alpha-motoneurones is evoked via interneurones that mediate reciprocal inhibition between flexors and extensors (Ia inhibitory interneurones), and to what extent via other spinal interneurones. The crossed inhibition was evoked by reticulospinal and vestibulospinal tract fibres, stimulated in the contralateral medullary longitudinal fascicle and the lateral vestibular nucleus, respectively, or by group II muscle afferents in the contralateral quadriceps nerve. The components of the IPSPs recorded in motoneurones that were mediated by Ia inhibitory interneurones were identified by their depression following activation of Renshaw cells. Trisynaptic components of IPSPs of reticulospinal and vestibulospinal origin, and polysynaptic (but not trisynaptic) components of IPSPs from group II afferents were found to be depressed in the majority of the motoneurones, while disynaptic components, those due to direct actions of inhibitory commissural interneurones, were not depressed. These results indicate that the coordination of left and right hindlimb movements based on crossed inhibition from reticulospinal and vestibulospinal neurones, depends on the degree of activation of Ia inhibitory interneurones by muscle spindle afferents and on their inhibition by Renshaw cells. Our results also indicate that Ia inhibitory interneurones do not operate as last-order inhibitory interneurones in crossed trisynaptic pathways from group II afferents, even though they mediate inhibition evoked by interneurones in shared polysynaptic pathways of crossed flexor and extensor reflexes coactivated by group II and other high-threshold muscle, skin and joint afferents.

Coordinated movements of the left and right limbs depend on well-organized operation of a number of neuronal networks. Some of these must be quite complex to ensure that an appropriate degree of crossed extension accompanies an ipsilateral flexion, whether the crossed extension is a part of the withdrawal reflex, or of other limb movements, including locomotor movements (for reviews see, e.g. Baldissera et al. 1981; Armstrong, 1986; Grillner & Dubuc, 1988; Rossignol, 1996; Noga et al. 2003). Complex neuronal networks are also needed to replace the combination of ipsilateral flexion and crossed extension by other state-dependent movement synergies, e.g. by bilateral extension or bilateral flexion (Aggelopoulos et al. 1996; Jankowska et al. 2005). These networks may nevertheless operate by making use of simple neuronal subsystems. For instance, contralateral motoneurones may be excited or inhibited from the reticular formation and the vestibular nuclei, or from group II muscle spindle afferents via single interneurones with crossed projections (commissural interneurones) that directly act on motoneurones (Hongo et al. 1975; Harrison et al. 1986; Arya et al. 1991; Jankowska et al. 2003; Krutki et al. 2003). The operation of commissural neurones may likewise be relatively simply adjusted to the behavioural requirements. For instance monoaminergic neurones may depress or enhance their actions (Jankowska et al. 2000; Edgley et al. 2004; Hammar et al. 2004), and commissural interneurones in pathways from group II afferents may in addition be depressed by GABAergic presynaptic inhibition (Edgley et al. 2003). In this report we will present evidence for yet another kind of such adjustment. We will show that inhibition of contralateral hindlimb motoneurones evoked by reticulospinal and/or vestibulospinal neurones via direct actions of inhibitory commissural interneurones is supplemented by inhibition via interneurones that mediate reciprocal inhibition between ipsilateral flexors and extensors (‘Ia inhibitory interneurones’), which are activated by excitatory commissural interneurones (see Fig. 1A). The inhibitory actions of commissural interneurones may therefore be enhanced by group Ia muscle afferents and adjusted by Renshaw cells, as previously demonstrated for other inhibitory actions mediated by Ia inhibitory interneurones (see Jankowska, 1992). In contrast, no contribution from Ia inhibitory interneurones to the short-latency inhibition evoked from group II muscle afferents was found. Only polysynaptic inhibition, evoked via other neuronal networks activated by group II afferents (Lundberg et al. 1987) appeared to be mediated by these interneurones.

Figure 1.

Diagram of the hypothetical network of neurones mediating di- and trisynaptic IPSPs evoked from the reticular formation, the vestibular nuclei and group II afferents in contralateral motoneurones and the location of the stimulation sites
A, the neuronal network of inhibition evoked via either inhibitory commissural interneurones (C shaded) or excitatory commissural interneurones (C white) and Ia inhibitory interneurones (In) which are excited by group Ia afferents and inhibited by Renshaw cells (R). The arrowhead indicates the site of conditioning stimulation of Q motoneurones axons after transection of the L5 and L6 dorsal roots. Open circles represent excitatory interneurones. Filled circles represent inhibitory interneurones. BD, stimulation sites, marked by the electrolytic lesion made after one of the experiments with the locations of marking lesions in the remaining experiments superimposed on the same scanned sections (posterior 9–10 for MLF, 7–8.5 for LVN and 3–4 at the border of the brachium conjunctivum, BC; at one additional location) according to the Horsley–Clarke coordinates (Berman, 1968). IO, inferior olive; Pyr, pyramid; IN, nucleus interpositus; FN, nucleus fastigius; SO, superior olive; co Ex, contralateral extensors; co Fl, contralateral flexors.

Methods

Preparation

The experiments were performed on 10 deeply anaesthetized adult cats (weighing 2.4–3.5 kg), some of which were also used for the study reported by Matsuyama and Jankowska (2004). Anaesthesia was induced with sodium pentobarbital (40 mg kg−1i.p.) and maintained with intermittent doses of α-chloralose sufficient to abolish any withdrawal reflexes (up to a total dose of 55 mg kg−1i.v.). During recording, neuromuscular transmission was blocked by pancuronium bromide (Pavulon, about 0.2 mg kg−1 h−1i.v.) and the animals artificially ventilated. The depth of anaesthesia was assessed by continuously recording blood pressure and heart rate and by monitoring pupil diameter. The mean blood pressure was kept at 100–130 mmHg and the end-tidal concentration of CO2 at about 4% by adjusting the parameters of artificial ventilation and the rate of infusion of a bicarbonate buffer solution with 5% glucose (1–2 ml kg−1 h−1). The core body temperature was kept at about 38°C by servo-controlled infrared lamps. At the end of the experiment the animals were killed with an overdose of anaesthetic. The experimental procedures were approved by Göteborg ethics committee and followed NIH and EU guidelines of animal care.

Inhibitory postsynaptic potentials (IPSPs) were recorded from motoneurones located on the left side of the spinal cord. They were evoked by stimulation of muscle nerves or the descending reticulospinal or vestibulospinal tract fibres on the right side. To eliminate synaptic actions evoked via descending tract fibres ipsilateral to the motoneurones, the spinal cord was hemisected on the left side at the level of the Th12 or 13 segments. The left L4–S1 dorsal roots were transected in order to allow a selective stimulation of motor axons by stimuli applied to muscle nerves. Alternatively motor axons were stimulated within the ventral roots in preparations in which all dorsal roots were left intact but the L6 or both the L5 and L6 ventral roots were transected and mounted on a pair of stimulating electrodes.

The preliminary dissection included cannulation of the trachea, a carotid artery, and the left and right cephalic veins. The left quadriceps (Q), and sartorius, (Sart) nerves were dissected, transected and mounted in subcutaneous cuff electrodes. Laminectomies were performed at the level of the lower thoracic (Th11–13) and lumbar (L3–L7) segments. The cerebellum was exposed to allow insertion of stimulating electrodes into the right medial longitudinal fascicle (MLF) and the right lateral vestibular nucleus (LVN) contralateral to motoneurones to be recorded from, as previously described (Jankowska et al. 2003; Matsuyama & Jankowska, 2003, 2004).

Stimulation

Axons of reticulospinal tract neurones in the MLF and of vestibulospinal neurones in the LVN were stimulated using an electrode made from an electrolytically etched tungsten wire (0.5 mm diameter) insulated except for its tip as a cathode, and a wire inserted into a neck muscle as an anode. Two to four stimuli of constant current 5.0 ms apart (0.2 ms, 50–100 μA) were applied in the MLF and one to five stimuli 3.3 or 2.5 ms apart (0.2 ms, 40–200 μA) in the LVN. The distribution of the stimulation sites, which were marked at the end of the experiments with an electrolytic lesion, is indicated in Fig. 1. These were within the MLF or at its lateral border at the level of the inferior olive (Fig. 1B) and within either rostral or caudal parts of LVN at the levels of the fastigial and interpositus nuclei (Fig. 1C and D). In two experiments, additional stimulation sites were rostral to the LVN, where fastigio-vestibular fibres providing input to it (Homma et al. 1995) were stimulated; one of these was at the medial border of the brachium conjunctivum (C). Peripheral nerves were stimulated with constant voltage stimuli (0.1 ms, intensity expressed in multiples of threshold (T) for the most sensitive fibres in a given nerve). The nerve stimulation was used to identify the motoneurones by their antidromic activation and to activate Renshaw cells via recurrent motor axon collaterals.

Recording and analysis

Intracellular records were obtained from 63 Sart and five Q motoneurones located in the 4th and 5th lumbar segments using glass micropipettes (1.5–2.0 μm tip diameter) filled with a 2 m potassium citrate solution. Descending volleys associated with the postsynaptic potentials (PSPs) evoked from the MLF and LVN were recorded from the lateral funiculus on the same side of the spinal cord as the motoneurones, usually within 5–10 mm of the microelectrode recording site. DC recording or low-pass filters were used when recording from motoneurones, and both the original records and averages of 10–40 single sweep records were stored (using a computer system designed by E. Eide, N. Pihlgren and T. Holmström, Göteborg University). Amplitudes and latencies of PSPs were measured from the averaged records.

Extracellular records from seven Ia inhibitory interneurones were used to evaluate effects of reticulospinal and vestibulospinal tract neurones upon them. The interneurones were identified by using criteria of Hultborn et al. (1971b), including activation by lowest threshold group I afferents, inhibition following stimulation of motor axons and the ability to be activated at high frequencies. Effects of excitatory and inhibitory stimuli on these neurones were quantified by comparing the number and the latencies of responses to 20–40 successive stimuli, from peristimulus histograms and cumulative sums of these responses, as in previous studies (e.g. Jankowska et al. 1997). Student's t test for paired normally distributed data was used for the statistical analysis.

Results

Depression of trisynaptic components of IPSPs evoked from the reticular formation by Renshaw cells

Figure 2 illustrates changes in IPSPs evoked by short trains of MLF stimuli following stimulation of motor axons and activation of Renshaw cells in preparations in which the L4–S1 dorsal roots were transected. The records in A and D show IPSPs evoked by test stimuli alone. These IPSPs were evoked at segmental latencies of 1.1–1.5 ms (measured from the MLF descending volleys), which indicates that they were evoked disynaptically, i.e. via single inhibitory interneurones interposed between the MLF fibres and the contralateral motoneurones. These interneurones are represented by the shaded circle labelled ‘C’ in Fig. 1A. The larger amplitudes and steeper rising phases of the IPSPs evoked by successive stimuli are in keeping with temporal facilitation of activation of these inhibitory interneurones. A notch in the declining phase of the last IPSP evoked by stronger stimuli (at the level of the second broken line in Fig. 2C where records from A are expanded), or a longer duration, suggested however, that the disynaptic components of the IPSPs were followed by longer latency components attributable to actions of two interposed neurones rather than one, an excitatory interneurone activated by reticulospinal tract fibres, represented by the upper commissural interneurone in Fig. 1A, and an inhibitory interneurone.

Figure 2.

Selective depression of trisynaptic IPSPs evoked from the MLF in contralateral motoneurones by Renshaw cells
A, IPSPs evoked in a Sart motoneurone by a train of MLF stimuli which were sufficiently strong to evoke not only disynaptic but also trisynaptic IPSPs. B, IPSPs evoked by the same stimuli preceded by conditioning stimulation of the ipsilateral Q nerve (i Q) to activate Renshaw cells (after the dorsal roots were transected). Lower traces in A and B are records of afferent volleys from the surface of the spinal cord. All are averages of 10 successive records. C, superimposed expanded records of the IPSPs in A and B and the difference between them. Broken vertical lines indicate the onset of the disynaptically and trisynaptically evoked IPSPs. DF, similar records from another Sart motoneurone in which only disynaptic IPSPs were evoked by weaker stimuli. Note in F that activation of Renshaw cells had no effect on the IPSP evoked by the thirdd stimuli and, if anything, increased IPSPs evoked by the second stimuli. Double arrows indicate the time windows within which the areas of the early parts of the conditioned IPSPs and of the differences between the test and conditioned IPSPs were measured (for the comparisons summarized in Table 1, column 9).

When MLF stimuli were preceded by conditioning stimulation of motor axons in the Q nerve, and a subsequent activation of Renshaw cells, rapidly declining IPSPs evoked by the first stimuli in a train remained practically unchanged (Fig. 2B and E and grey traces in the expanded records in Fig. 2C and F). The same was true for the early parts of longer duration IPSPs evoked by the last stimulus. In contrast, the later components of the longer duration IPSPs were considerably reduced. The test (black) and conditioned (grey) records in Fig. 2C, started to deviate after the level of the second broken vertical line which corresponded to the notch on the control IPSP. Subtraction of the conditioned IPSPs from the test IPSPs gave the difference trace in Fig. 2C, confirming that only the later components of the IPSPs were depressed.

A total of 36 Sart and one Q motoneurones in which amplitudes of IPSPs evoked from MLF exceeded 0.5 mV, and in which the IPSPs were not preceded by EPSPs, were investigated in the way illustrated in Fig. 2. Following conditioning stimuli the first components of IPSPs evoked by the second or third stimuli (within about 1 ms from their onset) decreased only insignificantly (Table 1 column 5). However when the areas of the test and conditioned IPSPs including both the early and the later components (within 4 ms from their onset) were compared, highly statistically significant differences were found (Table 1, column 7).

Table 1.  Depression of IPSPs evoked in hindlimb motoneurones by stimuli activating Renshaw cells
IPSP
from:
Motoneurone
number
IPSP
latency
(ms)
Early componentsEarly and late componentsDifferences


Window
(ms)

Conditioned:
test area
(%)


Window
(ms)

Conditioned:
test area
(%)

Relative
latency
(ms)
Difference:
conditioned
area
(%)
  1. The IPSPs were evoked from the MLF,and LVN, by joint actions of stimuli applied in the MLF and LVN and from contralateral group II afferents as indicated in column 1. The data are for IPSPs exceeding 0.4 mV and tested at conditioning–testing intervals between 8 and 25 ms using the most effective conditioning stimulation of motor axons (usually of Q sometimes of Sart). Data for IPSPs from LVN are subdivided into those with latencies that are more compatible with the disynaptic (1.4–1.8 ms) and trisynaptic (1.9–2.4 ms) coupling, respectively. The comparison of the areas was made within time windows of 0.7–1.5 ms and of 4 or 6 ms from the onset of the IPSPs in columns 5 and 7, as indicated in columns 4 and 6. The latency of the differences between the conditioned and test IPSPs in column 8 are expressed with respect to the latencies of the test IPSPs. The ratio between areas of the differences and of the conditioned IPSPs in column 9 is for the first 3 ms from their onset. *P= 0.01–0.05, **P= 0.001–0.01, ***P < 0.0001, ns P > 0.05.

12345678 9
MLF n= 371.1–1.51.594 ns4 73***1.0 ± 0.139
LVN n= 101.4–1.81.388 ns4 77**0.9 ± 0.135
LVN n= 51.9–2.4 4 44*0.3 ± 0.145
LVN and MLF n= 291.1–1.6 4 74***0.9 ± 0.157
gr II n= 162.5–4.01.591 ns6 89***1.7 ± 0.231

The depressed components were evoked at 1 ± 0.1 ms longer latencies (Table 1, column 8) than the latencies of the first components (Table 1, column 3). Since the latencies of the early components were compatible with the disynaptic coupling, the depressed components might therefore be classified as being evoked trisynaptically.

The difference in effects of conditioning stimuli on the early and later components of the IPSPs allows the depression of the later components to be attributed to the actions of Renshaw cells at a premotoneuronal rather than motoneuronal level, since any depression secondary to the effects of recurrent IPSPs evoked in the motoneurones (e.g. shunting) should have affected the early and late components of the IPSPs in a similar way. This conclusion is also consistent with the lack of relationship between the degree of depression of trisynaptic components of IPSPs of MLF origin and the size of the recurrent IPSPs recorded in motoneurones.

In order to estimate the relative contribution of Ia inhibitory interneurones and of inhibitory commissural interneurones to the IPSPs evoked from the MLF, a comparison was made between the size of the depressed area (represented by the ‘difference’ trace in Fig. 2) and of the area that resisted the effects of the conditioning stimuli. These areas were measured within 3 ms from their respective onsets. The size of the depressed areas was on average less than half the size (Table 1, column 9) of the early components that were not depressed. Considering that under our experimental conditions Renshaw cells only weakened but did not prevent activation of Ia inhibitory interneurones, this ratio suggests that the degree of inhibition mediated via the two potential parallel inhibitory pathways (see Fig. 1A), i.e. by excitatory commissural interneurones and Ia interneurones and by direct actions of inhibitory commissural interneurones, was comparable.

Depression of trisynaptic components of IPSPs evoked from the contralateral lateral vestibular nucleus by Renshaw cells

Our previous observations revealed that a considerable number of commissural interneurones are coexcited from the MLF and from the LVN (Jankowska et al. 2005) and that their inhibitory subpopulation mediates disynaptic inhibition of contralateral motoneurones (Krutki et al. 2003). It was therefore expected that not only disynaptic but also trisynaptic inhibition, via excitatory commissural interneurones and Ia inhibitory interneurones, would be likewise evoked not only from the MLF but also from the LVN. In order to verify this, the effects of Renshaw cells were tested under two experimental conditions. They were tested, firstly in motoneurones in which stimuli applied within the contralateral LVN (Fig. 3A and B) successfully evoked IPSPs, and secondly in motoneurones in which the appearance of the IPSPs required joint actions of LVN and MLF stimuli.

Figure 3.

Depression of trisynaptic components of IPSPs evoked from the LVN in contralateral motoneurones by Renshaw cells
A, an electrode track (indicated by the broken line) at Horsley–Clarke plane P7.0. The arrow indicates the site selected for testing effects of LVN stimuli. The scale is in millimetres, according to Horsley–Clarke's coordinates. FN, fastigial nucleus; H–C, Horsley–Clarke; P, posterior. B, IPSPs recorded in a Sart motoneurone from the indicated depths along the electrode track shown in A. The lowermost trace is the cord dorsum potential evoked by stimuli applied at depth −2. CF, intracellular records from two Sart motoneurones (with cord dorsum records in bottom traces in D and F); averages of 40 and 20 records, respectively. In C and E IPSPs were evoked by trains of test stimuli applied in the LVN and by conditioning stimuli alone (black) and IPSPs evoked when LVN stimuli were preceded by stimulation of motor axons in the Q nerve (grey). In D and F the right parts of records from A and B (twice expanded) are superimposed and aligned with differences between IPSPs evoked by the test stimuli alone and by joint actions of the test and conditioned stimuli from which responses to conditioned stimuli alone were subtracted. The first broken lines indicate the first components of the descending volleys from the LVN in the cord dorsum records; the second and third broken lines indicate the onsets of the test IPSPs and of the deviations between the test and conditioned IPSPs. Other indications are as in Fig. 2. Note that only the later components of IPSPs in C and D were depressed.

When induced from the LVN alone, the IPSPs were evoked by the later stimuli in a train, but they could be related to individual stimuli as indicated in Fig. 3D. They were evoked at minimal latencies of 4.4–5.8 ms from the effective stimulus. The shortest of these latencies (1.4–1.8 ms from the descending volleys) were considered as compatible with disynaptic coupling (via single interposed interneurones), and the longest ones (1.9–2.4 ms from the descending volleys) with trisynaptic coupling (via two interneurones in a series).

The depression of IPSPs evoked at latencies of 1.4–1.8 ms was generally weak (Table 1, column 5). The depression of IPSPs evoked at latencies ≥1.9 ms, was much more potent (Table 1, column 7) for IPSPs evoked by the fourth stimuli, and it was near total for IPSPs following the second and third stimuli (Fig. 3D and F).

When LVN stimuli alone did not evoke sufficiently distinct short-latency IPSPs, effects of Renshaw cells were tested on IPSPs induced by joint actions of LVN and MLF stimuli (Krutki et al. 2003). As shown in Table 1 (columns 5 and 7) the effects on these IPSPs were similar to those illustrated in Fig. 3. Since the IPSPs were due to the spatial facilitation of actions from the MLF and LVN, their depression reflects more directly effects mediated by commissural interneurones coexcited from the MLF and LVN.

In the whole sample the depressed area of IPSPs of LVN origin, constituted about one half (Table 1 column 9) of the early area that was not affected by Renshaw cells. The depressed components were evoked at segmental latencies 0.9 ± 0.1 ms longer than the latencies of the first components of the test IPSPs (Table 1, column 8). It appears thus that effects of Renshaw cells on IPSPs evoked from the LVN generally replicate the effects on IPSPs evoked from the MLF, and that the conclusions from the previous sections can be generalized to neurones mediating disynaptic and trisynaptic IPSPs of LVN origin.

Additional evidence for the mediation of IPSPs evoked from the reticular formation by Ia inhibitory interneurones

The depression of trisynaptic components of IPSPs of MLF or LVN origin by stimulation of motor axons would be compatible with these components being mediated not only by Ia inhibitory interneurones but also by Renshaw cells, in view of the mutual inhibitory interactions between Renshaw cells (Hultborn et al. 1968; Ryall, 1970; Hultborn et al. 1971a). Two additional tests were therefore made for the involvement of Ia interneurones.

The first test is illustrated in Fig. 4, which shows that mutual facilitation occurs between effects of stimuli that evoke IPSPs from group Ia afferents of an antagonist (Ia reciprocal IPSPs) and IPSPs from the MLF. When only very small IPSPs were evoked from the MLF (Fig. 4A) and from ipsilateral Q (Fig. 4B), joint application of the MLF and Q stimuli induced a much larger IPSP (Fig. 4C, black trace). The latter was much larger than the algebraic sum of effects of the two stimuli applied separately (Fig. 4C, grey trace) and it was almost as large as the IPSP evoked by stimulation of the Q nerve at an intensity near maximal for group Ia afferents (Fig. 4D). Since the two stimuli were timed so that the relayed MLF volley coincided with the Q volley (broken line in Fig. 4AC), the facilitation is fully in keeping with the excitation of the same interneurones by Q group Ia afferents and by interneurones activated by MLF stimuli. Such facilitation was found in all seven motoneurones tested (3 Sart and 4 posterior biceps-semitendinosus (PBST)).

Figure 4.

Spatial facilitation between synaptic actions of MLF fibres and of group Ia afferents
The upper records in each pair are from a Sart (flexor) motoneurone, as shown by antidromic activation in E, lower records being from the cord dorsum. AC, effects of separate and joint stimulation of the MLF and group Ia afferents in the Q (extensor) nerve. The grey trace in C is a sum of the upper traces in A and B, showing that the facilitation of the IPSP at a premotoneuronal level was considerably larger. D, IPSP evoked by near-maximal stimulation of group Ia afferents.

The second test is illustrated in Fig. 5 which shows that Ia inhibitory interneurones may be excited from MLF on the basis of direct records from these interneurones. To demonstrate this, effects of stimuli applied in the contralateral MLF were investigated on seven Ia inhibitory interneurones. All of these interneurones were excited by the lowest threshold muscle afferents, faithfully followed a train of stimuli at 400 Hz and were inhibited by conditioning stimulation of motor axons in the L6 ventral root, as illustrated in Fig. 5AC.

Figure 5.

Effects of MLF stimulation on a Ia inhibitory interneurone
A,B and D upper traces, single sweep extracellular records from the interneurone; lower traces are from the cord dorsum. C and E, the two upper traces are peristimulus time histograms (PSTHs) of responses of the interneurone and the bottom traces are cumulative sums of these responses. A and B, identification of the interneurone as the Ia inhibitory interneurone, showing responses to a high frequency of stimuli, responses to weak stimulation of the quadriceps (Q) nerve and inhibition of these responses following stimulation of a ventral root (VR) (grey trace in B). C, PSTHs and cumulative sums of responses to 20 consecutive stimuli applied alone, as in B (black) and when they were preceded by VR stimuli (grey). D, sample record of responses to near-threshold Q stimuli preceded by MLF stimuli. E, PSTHs and cumulative sums of responses to 20 test Q 1.1 T stimuli alone, and to test stimuli preceded by conditioning stimulation of the MLF, aligned with records in D. The number of spikes was counted within 2 ms windows indicated by broken lines.

None of the tested interneurones responded to MLF stimuli when these were applied alone, even when the stimulus intensity was supramaximal for the MLF descending volleys (150–200 μA) and the number of stimuli in the train was increased to 5–6. However, when the MLF stimuli were used as conditioning stimuli, the responses of all of the neurones were greatly facilitated. The degree of facilitation was estimated by comparing the number and the minimal latencies of spikes evoked by a series of 20–40 successive test stimuli, alone and when preceded by MLF stimuli. When optimal parameters of the stimuli were used, the interneurones increased the ratio of responses to near-threshold stimulation of group Ia afferents from 42 ± 12% to 72 ± 12% (the difference was statistically significant at P= 0.002) and the minimal latencies of responses evoked by near threshold stimuli were often reduced by 0.1–0.4 ms. These effects are illustrated in Fig. 5E with peristimulus time histograms (PSTHs) and cumulative sums of responses of one of the interneurones.

Effects of Renshaw cells on IPSPs evoked from contralateral group II afferents

The most potent inhibition of contralateral hindlimb motoneurones is evoked via polysynaptic pathways from high-threshold muscle, skin and joint afferents, including group II as well as group III afferents (Eccles & Lundberg, 1959; Lundberg et al. 1987). Of these, group II afferents evoke inhibition that is particularly distinct (Harrison & Zytnicki, 1984). In addition to polysynaptic actions via several interneuronal systems (Lundberg et al. 1987), group II afferents may evoke IPSPs via only one or two interneurones interposed between these afferents and the contralateral motoneurones, i.e. disynaptically or trisynaptically (Arya et al. 1991). Disynaptic components of IPSPs of group II origin would thus be attributable to inhibitory commissural interneurones, but the trisynaptic and polysynaptic components could be mediated by Ia inhibitory interneurones as last-order interneurones, as proposed in Fig. 1A, in the same way as trisynaptic components of IPSPs evoked from the MLF.

In order to test this possibility we investigated effects of Renshaw cells on IPSPs evoked in 16 left-side Sart and Q motoneurones by group II afferents of the right Q nerve. The IPSPs were evoked by 3–5 T stimuli, but not by weaker stimuli, at segmental latencies of 2.5–4 ms. Conditioning stimuli were applied to the left deafferented Q nerve, and the resulting depression was evaluated in the same way as the depression of IPSPs evoked from the MLF and LVN.

The main results are illustrated in Fig. 6 where the test IPSPs are shown in A and D, and IPSPs preceded by conditioning stimuli (grey traces) are superimposed on recurrent IPSPs (black traces) in B and E. When the test and conditioned IPSPs were superimposed, and the latter subtracted from the former, as in C and F, only small differences between their earliest components were found. For the whole sample the decrease was not statistically significant (Table 1, column 5). However, the later components, as those after the second broken lines in Fig. 6C and F, were often depressed. The mean depression was also small (Table 1, column 7) but highly statistically significant. When the areas of the differences between the later components of the test and conditioned IPSPs were related to the areas of the earliest components that were not, or were only weakly depressed, the mean depressed area was found to amount to about one-third of the early ones (Table 1, column 9).

Figure 6.

Effects of activation of Renshaw cells on inhibition evoked by contralateral group II muscle afferents
Left and right panels show records from two Sart motoneurones (upper traces, averages of 10 consecutive records) and from the surface of the spinal cord (lower traces). A and D, test IPSPs evoked from contralateral group II afferents. B and E, IPSPs evoked by the same test stimuli following conditioning stimulation of the ipsilateral deafferented Q nerve (grey traces) superimposed on recurrent IPSPs (black traces) evoked by conditioning stimuli. C and F, superimposed records of the test (black) and conditioned (grey) IPSPs, at higher resolution, after subtraction of the recurrent IPSPs from the conditioned responses, together with the differences between these records. Broken lines in C and F indicate onset of the IPSPs and of the components depressed by Renshaw cells. Other indications as in Fig. 2.

On average, the onset of the depression was delayed by 1.7 ± 0.2 ms with respect to the onset of the IPSPs (Table 1 column 8). These delays were thus longer than the delays of the depression of IPSPs evoked from the MLF.

Discussion

Modulation of inhibition evoked from the reticular formation, lateral vestibular nucleus and group II afferents in contralateral motoneurones by Ia inhibitory interneurones and Renshaw cells

The results of this study show that disynaptic inhibition of contralateral motoneurones by reticulospinal and vestibulospinal tract fibres, evoked via commissural interneurones (Hongo et al. 1975; Bannatyne et al. 2003; Jankowska et al. 2003; Krutki et al. 2003), is associated with trisynaptic inhibition mediated by Ia inhibitory interneurones. The evidence rests on the demonstration that Renshaw cells selectively depress those components of IPSPs that are evoked at latencies about 1 ms longer than the latencies of the disynaptic components.

The depression is attributed to activation of Renshaw cells because typical recurrent IPSPs induced in motoneurones showed that conditioning stimuli effectively activated Renshaw cells. Furthermore, any effects of high-threshold afferents entering the spinal cord via ventral roots that might add to effects of Renshaw cells would require much stronger conditioning stimuli (> 10 T rather than 2–5 T) to excite C fibres as well as group III and IV fibres (Coggeshall et al. 1974). The depression is attributed to actions of Renshaw cells located on the same side of the spinal cord as the motoneurones, since terminal projection areas of lumbar Renshaw cells are exclusively ipsilateral (see, e.g. Scheibel & Scheibel, 1969; Lagerback & Kellerth, 1985).

IPSPs depressed by Renshaw cells are attributed to Ia inhibitory interneurones (those mediating reciprocal inhibition between flexors and extensors) in view of the previous evidence that only Ia inhibitory interneurones are both excited by group Ia afferents and inhibited by Renshaw cells (see Jankowska, 1992), and in keeping with the demonstration of the mutual facilitation between effects of stimulation of group Ia afferents and of the MLF, both on IPSPs and on responses of individual interneurones (Figs 4 and 5).

Ia inhibitory interneurones are proposed to be activated by axon collaterals of commissural interneurones which are monosynaptically excited by reticulospinal neurones, as outlined in Fig. 1A, i.e. trisynaptically, in view of only about 1 ms longer latencies of the depressed components (Table 1 column 9) than of the earliest components which have been previously shown to be evoked disynaptically (Jankowska et al. 2003). The morphologically delineated areas of location of target cells of commissural interneurones, both within and just outside the motor nuclei (Harrison et al. 1986; Bannatyne et al. 2003; Matsuyama et al. 2004) are also appropriate for this connection.

Our study thus reveals another case of the operation of Ia inhibitory interneurones as premotor interneurones – in trisynaptic pathways between contralaterally descending reticulospinal and vestibulospinal neurones and motoneurones. The previously demonstrated actions of Ia inhibitory interneurones included disynaptic inhibition from the ipsilateral vestibular nuclei in the cat (Hultborn & Udo, 1972; Bruggencate & Lundberg, 1974) and from the contralateral corticospinal tract neurones in primates (Jankowska et al. 1976), as well as polysynaptic inhibition evoked via ipsilaterally descending cortico-, rubro- and reticulospinal tract fibres in the cat (Hultborn & Udo, 1972). The present observations show thus that supraspinal inhibition of both contralateral and ipsilateral motoneurones may be modulated as a function of muscle stretch (monitored by group Ia afferents and their excitatory actions on Ia interneurones inhibiting these motoneurones) and of muscle activation (associated with activation of Renshaw cells by motor axon collaterals).

The degree to which Ia inhibitory interneurones contribute to inhibition evoked from contralateral group II muscle afferents is more difficult to estimate. With respect to the total inhibition evoked from contralateral group II afferents, the results of this study show that Ia inhibitory interneurones and Renshaw cells may modulate it to a smaller extent than inhibition of reticulospinal origin (Table 1, column 7). However, if the components of IPSPs mediated by Ia inhibitory interneurones (later components depressed by Renshaw cells) are related to the earliest (most likely disynaptic) components, the situation is different. As shown in Table 1 (column 9), Ia inhibitory interneurones may be used as last-order interneurones in pathways from group II afferents to a similar extent as the inhibitory commissural interneurones.

Differentiation of commissural interneurones mediating inhibition from the reticular formation, lateral vestibular nucleus and group II afferents

There are strong indications that interneurones mediating crossed disynaptic actions of reticulospinal fibres and of group II afferents belong to two distinct subpopulations, since no commissural interneurones recorded so far were found to be monosynaptically excited by both reticulospinal fibres and by group II afferents (Jankowska et al. 2005). For this reason separate commissural interneurones are indicated in Fig. 7 as mediating disynaptic inhibition of group II and reticulospinal origin (cells C1 and C2). Interneurones mediating disynaptic IPSPs from the MLF and LVN are represented by the same cell, labelled C2, although it is not known how large a proportion of their total population is shared and how many may have more selective actions.

Figure 7.

Diagram of lines of interneurones mediating inhibition of contralateral motoneurones from the reticular formation, lateral vestibular nucleus and group II afferents
Excitatory and inhibitory neurones are represented by open and shaded circles, respectively. In, Ia inhibitory interneurones; R, Renshaw cells; C, commissural interneurones. For other indications see Fig. 1 and the text.

Excitatory commissural interneurones mediating trisynaptic IPSPs of MLF and LVN origin via Ia inhibitory interneurones are likewise represented by the same cell (C3 in Fig. 7), and strong mutual facilitation of trisynaptic IPSPs evoked from the MLF and LVN as well as depression of the facilitated IPSPs following activation of Renshaw cells are in keeping with their shared use. Excitatory commissural interneurones mediating IPSPs of group II origin are indicated as interneurones of a distinct subpopulation (represented by cell C4 in Fig. 7), in view of the lack of evidence for coexcitation of commissural interneurones by direct actions of group II and MLF fibres. Another reason is that an additional interneurone might be interposed in pathways from group II afferents, but not from the MLF, between the excitatory commissural interneurones and Ia inhibitory interneurones.

It should be added in this context that input from group I afferents to the excitatory commissural interneurones with input from reticulo- and/or vestibulospinal tract fibres (Jankowska et al. 2005) is in keeping with the enhancement of Ia reciprocal inhibition of motoneurones by stimulation of contralateral group I afferents (Harrison & Zytnicki, 1984). Polysynaptic input from contralateral flexor reflex afferents (including group II afferents) to commissural interneurones with monosynaptic vestibulospinal input would also explain joint inhibitory actions of these afferents and of ipsilateral vestibulospinal tract neurones via Ia inhibitory interneurones (Bruggencate & Lundberg, 1974).

The shared use of commissural interneurones (Krutki et al. 2003) or other spinal neurones (Li et al. 2001; Yeomans et al. 2002) as relay neurones by reticulospinal and vestibulospinal neuronal systems appears to parallel mutual interactions between these neuronal systems at the medullary level (Peterson & Felpel, 1971; Bolton et al. 1992). The question might therefore be asked whether the reported facilitation of reticulospinal actions on commissural interneurones by vestibular neurones does occur at the spinal, or at a medullary level. The latter possibility has to be kept in mind since effects of stimuli applied in the brain can seldom be restricted to only one particular neuronal system, even though previous control experiments demonstrated that the bulk of effects of stimuli applied in the MLF and LVN can be attributed to separate actions of reticulospinal and vestibulospinal neurones (Jankowska et al. 2003; Krutki et al. 2003; Matsuyama & Jankowska, 2004).

Functional considerations

It is still an open question whether all subpopulations of commissural interneurones, and if not which ones, play an essential role in inducing rhythmic locomotor movements, postural reactions and/or other movements evoked from the reticular formation and the vestibular nuclei (Wilson & Peterson, 1978; Grillner, 1981; Armstrong, 1986; Mori et al. 1989; Grillner et al. 1995; Li et al. 2001; Day & Cole, 2002; Noga et al. 2003). Whether the answer to this question is positive or negative, the two parallel ways of inducing inhibition of contralateral motoneurones of descending origin, either directly via inhibitory commissural interneurones (C2 interneurones in Fig. 7) or indirectly, via excitatory commissural interneurones (C3 interneurones in Fig. 7) and Ia inhibitory interneurones, will allow inhibitory effects of these interneurones to be modulated in a number of ways. The directly induced inhibition would to a greater extent depend on descending commands via reticulospinal neurones and their activation by corticospinal neurones (Jankowska et al. 2003; Edgley et al. 2004), or via vestibulospinal neurones and their activation from the cerebellum (Krutki et al. 2003; Matsuyama & Jankowska, 2004), motor cortex (Wilson et al. 1999) and the vestibular apparatus (see e.g. Carleton & Carpenter, 1983; Wilson et al. 1990). Inhibition mediated via Ia inhibitory interneurones would on the other hand be a function of the length and tension of the ipsilateral muscles, monitored by group Ia afferents and Renshaw cells, and their resulting actions, in addition to any descending actions on Ia inhibitory interneurones (see Jankowska, 1992).

The two parallel ways of inducing inhibition of contralateral motoneurones by group II afferents, directly via C1 interneurones in Fig. 7, and indirectly via C4 interneurones and Ia inhibitory interneurones, would likewise allow it to be a function of the length and tension of the ipsilateral muscles monitored by group Ia afferents and Renshaw cells, but also of the length of the contralateral muscles monitored by group II afferents themselves (see Matthews, 1972). On the top of these modulatory actions, transmission from group I and group II muscle afferents will be selectively modulated by GABAergic presynaptic inhibition (see Riddell et al. 1995) and transmission from both primary afferents and reticulospinal fibres by monoamines (see Riddell et al. 1993; Edgley et al. 2004).

Appendix

Acknowledgements

We wish to thank Drs S. A. Edgley and I. Hammar for their comments and Mrs Rauni Larsson for her invaluable assistance during the experiments and with histological verifications. The study was supported by grants from NIH (NS 40 863) and the Swedish Research Council (15393-01 A).

Authors' present addresses

P. Krutki: Department of Neurobiology, University School of Physical Education, 60-352 Poznań, Poland.K. Matsuyama: Department of Physiology, Sapporo Medical University School of Medicine, Sapporo, 060-8556 Japan.

Ancillary