The response of a representative pallidal neuron to the application of standard DBS is shown in Figure 4B. When compared with the result of application of the closed-loop GPtrain|M1 stimulation (Figure 3), this neuron demonstrated only a moderate reduction in its discharge rate. Similarly, the neuron exhibited less pronounced changes in its discharge pattern, which remained bursty and oscillatory during the application of standard DBS (Figure 4D), as previously

described (Johnson et al., 2009 and McCairn and Turner, 2009). Also in line with previous reports (Boraud et al., 1996 and Johnson et al., 2009), the primates’ akinesia was alleviated during the application of standard DBS (Figure 4E), albeit to a lesser extent than during the application of GPtrain|M1 closed-loop stimulation (Figure 3E see more and Figure S2). Overall, the mean discharge rate of the GPi neurons Galunisertib solubility dmso (Figure 6B) and the M1 and GPi oscillatory activity at the double-tremor frequency band (Figure 7D) were reduced during the application of standard DBS compared with spontaneous activity, coinciding with an increase in the mean kinesis estimate (Figure 5A). Once again, the effects of stimulus application on the outcome parameters were reproducible between trials (Figure S5). As expected, the stimulus frequency delivered during the application of GPtrain|M1 closed-loop

DBS was significantly lower than that during standard DBS (30.185 ± 2.41 versus 130.007 ± 0.0004 Hz, Figure 1C, one-way ANOVA, p < 0.01). Furthermore, stimulus irregularity significantly increased (coefficient of variation of the interstimulus interval duration 5.0605 ± 0.067 versus 0.0003 ± 1.6∗10−5, one-way ANOVA, p < 0.01, Figure 1C). However, despite the reduction in the stimulus frequency in the GPtrain|M1 mode, the GPi discharge rate was significantly lower during this closed-loop stimulation than below during the standard 130 Hz open-loop GPi DBS (Figure 6B, red versus dark-red bars; one-way ANOVA, p < 0.05). When comparing the normalized oscillatory activity at tremor and double-tremor

frequencies between the two paradigms, the closed-loop strategy resulted in greater reduction of power in both frequency bands. This was true in both the cortical and the pallidal neuronal populations (one-way ANOVA, p < 0.01 for tremor frequency band and p < 0.05 for double-tremor frequency band; Figures 7C and 7D, respectively). We next set out to ensure that the apparent success of the closed-loop stimulation method was indeed due to its adaptive properties. Since setting the stimulus interval to 80 ms from trigger detection could induce a double-tremor frequency rhythm in ongoing activity, we controlled for the effect of application of such a rhythm using open-loop paradigms.