Subpopulations of commissural interneurones with monosynaptic input from either group II afferents or from MLF/LVN/group I afferents
The results described here show that the midlumbar segments of the spinal cord contain two distinct groups of lamina VIII commissural interneurones, one with monosynaptic input from group II afferents and another with monosynaptic input from MLF. Furthermore, they indicate that monosynaptic input from vestibulospinal tract fibres and group I afferents is found in the group with monosynaptic reticulospinal input but not the group with monosynaptic group II input.
While these findings are clear from the data, two issues must be considered in relation to whether the organization applies more broadly. Firstly, we have restricted our recordings to the midlumbar (L3, L4 and L5) segments. We focused on this region partly for functional reasons, in that previous studies have described neurones with group II input in these regions and more recent work has highlighted the role of this region in locomotor behaviour (Rossignol et al. 2004), but also for practical reasons, in that characteristic focal synaptic potentials evoked by both afferent and descending tract fibres provide a guide to recording locations in the deep grey matter of these segments.
Secondly, since intracellular recording from interneurones deep in the grey matter is difficult, the samples of neurones on which the analysis is based are relatively small (samples of 10 neurones with monosynaptic input from each of group I afferents, group II afferents and the LVN and a larger sample of over 50 interneurones with input from MLF). The larger sample of commissural interneurones with input from MLF may indicate that these neurones are more numerous, but also that they are larger and more easily penetrated: many of the group II activated commissural interneurones from which extracellular recordings were made could not be subsequently penetrated for intracellular recordings. However, the main conclusion that monosynaptic input from group II afferents and MLF is found in separate subpopulations of commissural interneurones is supported by the observations on extracellularly recorded neurones (14 and 24, respectively; see results) which were activated as would be predicted from the intracellular records. Although evaluation of the synaptic linkage of responses in extracellular records is less reliable, the consistency of the data increases the confidence in our conclusions.
Thirdly, in most experiments we investigated group II input only from the Q and Sart nerves, which are the main source of input to laminae VI–VIII interneurones in the midlumbar segments (Edgley & Jankowska, 1987b). Consistent with this, whenever stimulation of other nerves was tested on commissural interneurones with monosynaptic input from MLF, only input from group I afferents was found. Along the same lines, the stimuli we used to activate reticulospinal axons were applied within the ipsilateral MLF at the site from which the largest descending volleys were evoked. These stimuli would be likely to excite a considerable proportion of the reticulospinal tract fibres located in the MLF and the neighbouring part of the reticular formation, given that we routinely used stimuli of 50 or 100 μA, which would act within a radius of 0.5–1 mm (Gustafsson & Jankowska, 1976). However, axons of the reticulospinal tract neurones located in the nucleus gigantocellularis, descend ipsilaterally outside MLF (Basbaum et al. 1978; Matsuyama et al. 1988; Mitani et al. 1988), so they would most likely have escaped activation by stimuli delivered in the MLF. Our conclusions therefore relate to monosynaptic coupling between reticulospinal tract fibres in MLF and commissural interneurones.
Our conclusion that monosynaptic EPSPs from group II afferents and MLF fibres are evoked in different commissural interneurones contrasts with the finding that disynaptic or oligosynaptic actions following the same stimuli were not segregated. The validity of this conclusion depends critically on the reliability of our classification of EPSPs as evoked mono- and disynaptically. As stated in Methods, three features of EPSPs were used to differentiate disynaptic from monosynaptic EPSPs: longer latencies, failure to be evoked by single stimuli and temporal facilitation of effects of successive stimuli in a train.
With respect to EPSPs evoked from group II afferents, we relied mainly on the first two criteria; temporal facilitation tests were often inconclusive, partly because responses to successive stimuli were difficult to estimate on mixtures of EPSPs and IPSPs evoked by the first stimulus, and partly because potent presynaptic inhibition follows stimulation of group II afferents (Edgley et al. 2003) and can counteract even monosynaptic EPSPs (as illustrated in Fig. 5D). The upper limit of the latencies of the EPSPs we classified as monosynaptic (2.8 ms from group I volleys) on the basis of being evoked by single stimuli near threshold for group II afferents, was higher than the minimal latencies of IPSPs evoked by group II afferents, which were unquestionably disynaptic, and could be as low as 2 ms. Thus there was a considerable overlap in latencies of EPSPs evoked monosynaptically and disynaptically from group II afferents (within the grey box in Fig. 2B). However, the latencies of EPSPs evoked from group II afferents in commissural interneurones with monosynaptic input from MLF were all longer than 2.8 ms, i.e. always exceeded the overlapping range of mono- and disynaptic latencies. Conversely, none of the EPSPs of group II origin which we classified as monosynaptic and which had latencies within the overlapping monosynaptic/disynaptic range was accompanied by monosynaptic EPSPs from the MLF.
With respect to EPSPs evoked from the MLF, all of the three criteria were applied and the upper limit of latencies linked with monosynaptic actions (0.9 ms) was found consistently for EPSPs that appeared after a single stimulus and showed only minimal temporal facilitation. We consider therefore the probability of erroneous classification of these EPSPs to be minimal.
Differentiation between monosynaptic and disynaptic EPSPs evoked by LVN stimulation was less reliable since electrical stimuli in the LVN can activate vestibulospinal tract neurones both directly and indirectly (trans-synaptically). We cannot therefore exclude the possibility that some of the EPSPs classified as having been evoked disynaptically resulted from monosynaptic connections of vestibulospinal axons on commissural interneurones. However, none of the EPSPs evoked in commissural interneurones with segmental latencies of < 1 ms were found in interneurones with monosynaptic input from group II afferents.
Given the small samples of interneurones, the patterns of input we describe may represent an oversimplification and not apply to all commissural interneurones. In addition, although the probability of direct synaptic actions of both group II afferents and reticulospinal fibres on midlumbar commissural interneurones appears to be low, the possibility that commissural interneurones located more caudally might be coactivated by group I and II afferents remains an open question. Re-inspection of records from two previous samples (Harrison et al. 1986; Jankowska & Noga, 1990) revealed that six intracellularly recorded commissural interneurones had monosynaptic input from group I but not group II afferents and five commissural interneurones had monosynaptic input from group II but not group I afferents. However, in another five interneurones EPSPs that followed monosynaptic EPSPs from group I afferents, and were not tested for temporal facilitation, could have been evoked either disynaptically by group I afferents or monosynaptically by group II afferents. These EPSPs appeared at latencies of about 2 ms, which would be compatible with monosynaptic actions of group II afferents but in two neurones they were superimposed on disynaptic IPSPs of group I origin, which might have led to the latency of the EPSPs being overestimated. Furthermore, in the remaining neurones, the threshold for inducing these EPSPs was compatible with effects of either the lowest threshold group II afferents or the highest threshold group I afferents (1.6–2T) or was not defined.
Disynaptic and polysynaptic input in subpopulations with selective monosynaptic inputs
An important finding was that di- or oligosynaptic inhibition evoked by stimulation of MLF and ipsi- and contralateral group II afferents was seen in a large majority (> 70%) of the commissural interneurones with monosynaptic EPSPs from the MLF, LVN and group I afferents. Conversely, di- or oligosynaptic inhibition was evoked by MLF stimulation in many (60%) of the commissural interneurones in which monosynaptic EPSPs were evoked by stimulation of group II afferents. This organization suggests that the two groups of commissural interneurones are unlikely to be active coincidentally. In addition, since NA depresses transmission from group II afferents but facilitates transmission from MLF fibres to lamina VIII commissural interneurones (Hammar et al. 2004), local release of NA would further reduce the probability of coactivation of these two subpopulations of interneurones.
Co-activation of commissural interneurones with monosynaptic input from either group II afferents or from MLF might nevertheless be made possible via other neurones. As shown in Fig. 3B, the excitatory disynaptic input from all sources was distributed to all subpopulations of commissural interneurones. The proportions of interneurones in which disynaptic EPSPs were detected were not particularly large but might be greater in behavioural contexts where the interneurones producing the excitation are facilitated.
The actions of commissural interneurones with monosynaptic input from either group II afferents or from the MLF could also depend on postsynaptic inhibitory control of these neurones from other sources, e.g. those listed in Fig. 3C, and by presynaptic GABAergic inhibition (Edgley et al. 2003). The widespread origin of the inhibitory control of commissural interneurones should be interpreted in terms of potential for focusing their activity rather than for indiscriminate reduction of activity. When certain combinations of afferents or descending tract neurones are activated, e.g. in specific behavioural contexts, the inhibition should be much more focused.
A source of postsynaptic inhibition which we would like to single out as being particularly important is disynaptic inhibition evoked from contralateral group II afferents. IPSPs evoked from contralateral group II afferents were evoked at similar minimal latencies as IPSPs from ipsilateral group II afferents and were evoked as easily by single stimuli as by double stimuli (Figs 2C, 6D and 7C). They were found in both subpopulations of commissural interneurones but were most common in commissural interneurones monosynaptically activated from MLF/LVN and by group I afferents, as shown in Fig. 3C. Actions mediated by either group of commissural interneurones could thus be coordinated on the basis of information from muscles on both sides of the body.
However, postural adjustments could also be based on peripheral afferent information from muscle, skin and joint receptors as well as from vestibular and neck receptors and on ongoing central commands that are forwarded to commissural interneurones via reticulospinal neurones (Peterson & Felpel, 1971; Kasper et al. 1989; Bolton et al. 1992). The shared use of at least some commissural interneurones by reticulospinal and vestibulospinal neuronal systems appears to parallel the mutual interactions between these systems at medullary level (Peterson & Felpel, 1971; Bolton et al. 1992). The question might therefore be asked whether facilitation of reticulospinal actions on commissural interneurones by vestibular neurones occurs at the spinal level, or only in the medulla. Previous control experiments demonstrated that the bulk of effects of stimuli applied in MLF and LVN can be attributed to separate actions of descending reticulospinal and vestibulospinal fibres (Jankowska et al. 2003; Krutki et al. 2003; Matsuyama & Jankowska, 2004). In addition, the effects of stimuli evoked from MLF and LVN in this study differed considerably, for example monosynaptic field potentials were evoked from LVN and MLF in different regions of the spinal grey matter (Fig. 1E, F), monosynaptic EPSPs from one, but not the other, were found in a number of commissural interneurones (Fig. 2), and there were opposite effects (EPSPs and IPSPs) evoked by these systems in some commissural interneurones. Mutual facilitation between synaptic actions evoked from MLF and LVN may therefore occur in the commissural interneurones.
Functional roles of the different subpopulations of commissural interneurones
Bilateral coordination is fundamental to locomotion and spinal commissural neurones form essential elements of locomotor networks in fish (Buchanan & McPherson, 1995; Grillner & Wallen, 2002) and tadpoles (Roberts, 2000). In mammals many commissural interneurones are rhythmically active during locomotor-like activity (Kiehn & Butt, 2003; Matsuyama et al. 2004) and have been considered to be fundamental parts of the locomotor central pattern generating network (Kiehn & Butt, 2003). On the other hand, it has been reported that alternating activation of flexors and extensors on the left and right sides may be preserved even when the spinal cord is split along almost the whole length of the lumbosacral enlargement (L2–S1 in the chronic cat (Kato, 1988); L1–the cone in the neonatal rat in vitro (Cowley & Schmidt, 1997)).
Commissural interneurones with monosynaptic input from MLF and/or LVN are likely to be involved in locomotion in view of the importance of reticulospinal neurones in the initiation of locomotion (Armstrong, 1988; Jordan, 1991; Mori et al. 2001; Deliagina et al. 2002; Noga et al. 2003) and the short latencies of postsynaptic potentials evoked in motoneurones during centrally initiated locomotion (Shefchyk & Jordan, 1985; Noga et al. 2003). Furthermore, commissural interneurones monosynaptically activated from the reticular formation are rhythmically active during fictive locomotion and some are disynaptically excited from the cuneiform nucleus (MLR; Matsuyama et al. 2004). Disynaptic excitation from MLR has also been found in two commissural interneurones described by Jankowska & Noga (1990) in which monosynaptic EPSPs were evoked from ipsilateral group I afferents and oligosynaptic IPSPs from group II afferents. The observation that the activation of these commissural interneurones is facilitated by locally applied noradrenaline (Hammar et al. 2004) is also consistent with involvement in locomotion since noradrenergic receptor antagonists delivered in the midlumbar segments block locomotion generated by the lumbosacral segments (Rossignol et al. 2001).
Ipsilaterally projecting midlumbar interneurones with monosynaptic input from group II muscle afferents may be disynaptically excited (Edgley et al. 1988) by stimulation of brainstem structures used to evoke fictive locomotion (the cuneiform nucleus, MLR), in agreement with connections from reticulospinal fibres in these ipsilaterally projecting neurones (Davies & Edgley, 1994). Two-thirds of these neurones were rhythmically active during one of the phases of the step cycle (Shefchyk et al. 1990). However, during MLR-evoked locomotion in decerebrate preparations responses to stimulation of group II afferents are generally depressed. If the ipsilaterally and contralaterally projecting interneurones behave in the same way, activity of the lamina VIII commissural interneurones with group II input could also be modulated during locomotion. Critically, during real locomotion additional excitatory drives might come from descending systems, e.g. the LVN and the rubrospinal or pyramidal tracts (Davies & Edgley, 1994) as well as from both group Ia and group II afferents.
Another major function of commissural interneurones with group II input is in determining different patterns of crossed reflexes. Marked differences in the expression of crossed actions of group II afferents are found with descending tracts intact, in which case the actions are strongly biased to inhibition, and after spinalization when the inhibition is less predominant (Arya et al. 1991). These patterns depend on the spinal actions of monoamines (Aggelopoulos et al. 1996), and it may be important that monoamines have differential effects on the groups of commissural interneurones with input from the MLF and from group II afferents (Hammar et al. 2004) indicating that these groups of interneurones have different roles in these crossed reflex actions. Interactions between different subpopulations of commissural interneurones and other interneurones remain to be assessed but, importantly, their actions should be strictly contralateral since commissural interneurones of both populations have only crossed terminal projection areas (Bannatyne et al. 2003).