We examined the interactions between the cortical inhibition due to peripheral sensory stimulation and intracortical inhibition. We found that the inhibitory effects of LICI and MNSI were reduced when applied together, whereas SICI seemed unaffected by MNSI and their effects were additive.
Interaction between MNSI and LICI
Experiment 1 showed that both MNSI and LICI decreased with higher test MEP amplitude, although the effect was more marked for LICI than MNSI. Therefore, both MNSI and LICI had a greater effect on neurones activated at relatively low intensities than those activated at higher intensities. In Experiment 2, we studied the interaction between MNSI and LICI by applying them together. In order to produce a similar degree of corticospinal activation with and without LICI, we matched the test MEP amplitude by increasing the TS when preceded by the CS100TMS. MNSI in the presence of LICI was significantly reduced compared to MNSI whether matched for the TS intensity or test MEP amplitude. LICI also appeared to be reduced in the presence of MNSI.
Several possible mechanisms of interaction between MNSI and LICI are illustrated in Fig. 5. It should be noted that these models represent populations of neurones that are activated in certain experimental settings and they do not necessarily reflect functional populations in normal voluntary movement. In Model A, we hypothesize that MNSI and LICI are mediated via different cell populations that do not interact. Model B suggests that MNSI and LICI are both mediated via the same pathways with two possible scenarios: B1, LICI and MNSI activate the same neuronal population; B2, MNSI activates the neuronal population mediating LICI. Model C proposes that MNSI and LICI are mediated via independent cell populations, but may have excitatory or inhibitory interactions with each other.
Figure 5. Possible interactions between MNSI and LICI
‘Inter’ indicates interneurones, whereas ‘output’ stands for output neurones. A, MNSI and LICI are independently mediated and do not interact; B1, MNSI and LICI are mediated simultaneously via the same interneurones; B2, MNSI is sequentially mediated via the LICI; C, MNSI and LICI are independently mediated, but interact via excitatory or inhibitory mechanisms.
Download figure to PowerPoint
It is unlikely that the interactions between MNSI and LICI can be explained with Model A. If these two mechanisms were mediated by independent pathways that do not interact, an additive effect of LICI and MNSI could be expected. However the results of Experiment 2 indicate that the effects of both MNSI and LICI were reduced in the presence of each other, suggesting there are interactions between these mechanisms. Since both MNSI and LICI predominately affect neurones activated at low intensities (Experiment 1), one possible explanation for this result is a saturation or occlusion effect. This possibility cannot be excluded from the results of the average data. However, there are several observations from single subject data that are not consistent with this mechanism. In four subjects, the effect of MNS changed from inhibition when applied alone to facilitation in the presence of LICI and in six subjects the CS100TMS pulse changed from inhibiting the test MEP to facilitation in the presence of MNSI. These facilitatory effects of either CS200MNS or CS100TMS in the triple pulse condition cannot be explained by the occlusion model. Moreover, the occlusion model predicts that the effect will be greater with greater baseline MNSI and LICI. We found that this is true for MNSI but not for LICI (Fig. 3).
The scheme depicted in Model B1 is unlikely since there was no correlation between the strength of LICI and MNSI at any of the target test MEP amplitudes studied in Experiment 1. If MNSI sequentially activates LICI pathways (Model B2), an additive effect of MNSI on LICI may be expected, which is different from the results in Experiment 2. A saturation effect should also be considered. One inhibitory mechanism alone might already activate the pathway to a high degree, so that another inhibitory mechanism has little additional effect. However, we have argued in the preceding paragraph that it is unlikely that a saturation effect can explain our finding.
Model C probably best explains our findings. Since Experiment 2 showed a reduced inhibitory effect of MNSI and LICI when the two mechanisms are combined, the interactions between MNSI and LICI are predominately inhibitory. This can be due to MNSI inhibiting LICI or LICI inhibiting MNSI. Since the interaction between MNSI and LICI is related to the strength of baseline MNSI but not baseline LICI, the most parsimonious explanation is that MNSI inhibits LICI rather than LICI inhibiting MNSI. This scheme is illustrated in Fig. 6A.
Figure 6. Proposed models for MNSI-LICI and MNSI-SICI interactions
The interactions between MNSI and LICI/SICI that are consistent with our findings. A, MNSI and LICI are independently mediated, but MNSI has inhibitory influence on LICI. B, MNSI and SICI are independently mediated and do not interact.
Download figure to PowerPoint
However, LICI inhibiting MNSI can also occur if the CS100TMS stimulus simply ‘resets’ the whole system, removing any MNSI effect and leaving only the LICI effect. In this case the MEP amplitude of the CS200MNS- CS100TMS-TS pulse combination (2G) would be the same as the MEP amplitude of the CS100TMS-TS pulse combination (2E). This hypothesis is consistent with no overall effect of MNSI in the presence of LICI (2G/2E = 0.99) in the average data. While this hypothesis cannot be ruled out, several observations are more consistent with MNSI inhibiting LICI. First, in the two subjects who showed MEP facilitation with CS200MNS pulse alone, the CS200MNS pulse in the presence of LICI caused MEP inhibition in both subjects (2G/2E = 0.74 and 0.75). This can be explained by CS200MNS increasing LICI in these two subjects. If the CS100TMS stimulus simply resets the system by wiping out any effects of MNSI, the MEP amplitude for 2G and 2E should be the same. Second, one subject had no LICI (2E/2D = 1.06) but had strong MNSI (2F/2D = 0.19). If CS100TMS abolishes MNSI regardless of its inhibitory effect, the MEP in the triple pulse stimulation (2G) should be about the same size as the CS100TMS-TS MEP (2E) and considerably larger than the CS200MNS-TS MEP (2F). However, in this subject the MEP in 2G was essentially the same size as 2F (2G/2F = 0.97) and was much smaller than 2E (2G/2E = 0.17), indicating that the CS100TMS did not abolish MNSI in this subject. Third, in two subjects CS200MNS alone lead to MEP inhibition but CS200MNS in the presence of LICI caused significant facilitation (2G > 2E). Finally, the interactions between MNSI and LICI were better predicted by multiplying the inhibitory effects of MNSI and LICI than MNSI alone. This result can be expected if MNSI inhibits LICI, but if LICI abolishes MNSI then the interaction between MNSI and LICI would be best predicted by the strength of MNSI alone.
Another issue to consider is whether the inhibition of LICI by MNSI is related to the inhibition of the CS100TMS MEP that was observed in most subjects. The magnitudes of these effects are similar since increasing the strength of the CS100TMS pulse to compensate for the MEP inhibition in Experiment 3 abolished the effects of MNSI on LICI. However, they are probably mediated by different mechanisms since there was no correlation between the extent of LICI inhibition and CS100TMS MEP inhibition (Fig. 3D).
Interactions between SICI and MNSI
For the interactions between SICI and MNSI we will consider the same possible mechanisms as for the interactions between LICI and MNSI (see Fig. 5). It is unlikely that SICI and MNSI are mediated via the same pathways as illustrated in Model B. Experiment 1 showed that changing the TS intensity had opposite effects on MNSI and SICI. There was also no correlation between SICI and MNSI in any of the three TS intensities tested. These results are different from that of a previous study where a correlation between SICI and MNSI was found in six subjects (Trompetto et al. 2001). Model C for interactions between SICI and MNSI is also unlikely because we found no significant change in SICI in the presence of MNSI and the effects of SICI and MNSI seemed to be additive. Therefore, Model A showing that SICI and MNSI are mediated via independent mechanisms seems most consistent with our data (Fig. 6B).
A previous study showed that LICI inhibits SICI (Sanger et al. 2001). If MNSI inhibits LICI, it could potentially lead to facilitation of SICI. However, the absence of SICI facilitation by MNSI may be explained if there is little background LICI activity. LICI may be related to GABAB activity and it has been shown that GABAB receptors have little spontaneous background activity (Mott & Lewis, 1994). Although unlikely, we cannot completely exclude the possibility that MNSI inhibits SICI but the effect is counter-balanced by facilitation of SICI through inhibition of LICI.
Mechanisms of peripheral sensory stimulation
Neuro-imaging studies showed that peripheral sensory stimulation primarily activates the primary somatosensory cortex (S1), the second somatosensory area (S2) and the posterior parietal cortex (Korvenoja et al. 1999; Boakye et al. 2000). Temporal aspects of cortical activation after peripheral sensory stimulation were assessed by somatosensory evoked potentials (SEPs) and magnetoencephalographic somatosensory evoked fields. At shorter latencies (< 40 ms) the contralateral S1 (Allison et al. 1989; Forss et al. 1994) and contralateral S2 (Karhu & Tesche, 1999; Korvenoja et al. 1999) are primarily activated.
It is likely that the peripheral sensory information contributing to MNSI is mediated from S1, S2 and the posterior parietal cortex and then projects to the motor cortex. Animal studies showed an extensive network of cortical connections from the S1 (Porter & Sakamoto, 1988; Burton & Fabri, 1995), S2 and the posterior parietal cortex (Ghosh et al. 1987) to the motor cortex. It has been suggested that the major sensory inputs to the motor cortex mainly terminate in superficial cortical layers (layers II-III). Cells in the superficial layers of the motor cortex respond to S1 stimulation with a consistent, short latency EPSP (Porter et al. 1990). The majority of sensory inputs seem to terminate at interneurones, that have modulatory effects on corticofugal neurones located in layers V and VI (Porter et al. 1990). These interneurones can be inhibitory or excitatory (Kosar et al. 1985; Porter et al. 1990), suggesting that sensory cortex stimulation can have both inhibitory and excitatory influences on pyramidal tract neurones. This may explain the findings of both increased (Deuschl et al. 1991; Komori et al. 1992) and decreased (Clouston et al. 1995; Manganotti et al. 1997; Tokimura et al. 2000) test MEP amplitudes at shorter latencies (< 50 ms) between MNS and TMS pulse.
However, a direct activation of the motor cortex via sensory afferents from the periphery cannot be excluded. A recent TMS study found inhibition of the motor cortex as early as 20 ms after a MNS and proposed a direct input from peripheral afferents to the motor cortex (Tokimura et al. 2000). Furthermore, the P22 peak of the short-latency SEP peaks may originate from the precentral motor area (Desmedt & Ozaki, 1991; Babiloni et al. 2001). However, motor cortex activation after MNS was not found in a PET study (Ibanez et al. 1995). Animal studies showed that the ventral posterior complex of the thalamus, the major sensory thalamic relay, only has minor direct projections to the motor cortex (Darian-Smith & Darian-Smith, 1993; Huffman & Krubitzer, 2001) and thus a structural correlate for direct motor cortex activation after peripheral sensory stimulation has not yet been found.
MNSI at short and long ISIs is likely to be mediated by different mechanisms. Ridding & Rothwell (1999) proposed a reduction of SICI in the presence of a peripheral sensory stimulus preceding the test pulse by 40 ms. MNSI at short and long ISIs also differs in pathological conditions. In patients with focal dystonia (Abbruzzese et al. 2001) the MNSI at an ISI of 200 ms is absent, whereas the MNSI at short ISIs was normal.
Interactions of peripheral stimulation and intracortical inhibition
The two inhibitory mechanisms detected by TMS (SICI and LICI) seem to be mediated via different inhibitory properties. Animal studies showed that early inhibitory postsynaptic potentials (IPSPs) (peaking around 10-20 ms) are mediated via GABAA and late IPSPs (peaking around 150-200 ms) are mediated via GABAB receptors (Davies et al. 1990; Kang et al. 1994; Deisz, 1999). The response pattern throughout the different layers of the motor cortex seems to be different for slow and fast IPSPs (Kang et al. 1994). Inhibitory interneurones producing slow IPSPs seem to be mainly distributed in layer II. This is also the layer where the main sensory input to the motor cortex arrives. Inhibitory interneurones producing fast IPSPs are distributed throughout almost all layers. It has been suggested that SICI may be mediated by GABAA receptors (Hanajima et al. 1998) and LICI by GABAB receptors (Werhahn et al. 1999; Sanger et al. 2001). If this is correct, this may explain why MNSI modulates LICI, but not SICI.
In conclusion, cortical inhibition from peripheral sensory stimulation at long ISIs (MNSI) seems to be mediated by circuits different from those mediating SICI and LICI. The MNSI circuit may inhibit LICI while MNSI and SICI circuits seem to be independent from each other.