The pedunculopontine tegmental nucleus of the brainstem (PPTg) has extensive interconnections and neuronal-behavioural correlates. It is implicated in movement control and sensorimotor integration. We investigated whether single neuron activity in freely moving rats is correlated with components of skilled forelimb movement and whether individual neurons respond to both motor and sensory events. We found that individual PPTg neurons showed changes in firing rate at different times during the reach. This type of temporally specific modulation is like activity seen elsewhere in voluntary movement control circuits, such as the motor cortex, and suggests that PPTg neural activity is related to different specific events occurring during the reach. In particular, many neuronal modulations were time-locked to the end of the extension phase of the reach, when fine distal movements related to food grasping occur, indicating strong engagement of PPTg in this phase of skilled individual forelimb movements. In addition, some neurons showed brief periods of apparent oscillatory firing in the theta range at specific phases of the reach-to-grasp movement. When movement-related neurons were tested with tone stimuli, many also responded to this auditory input, allowing for sensorimotor integration at the cellular level. Together, these data extend the concept of the PPTg as an integrative structure in the generation of complex movements, by showing that this function extends to the highly coordinated control of the forelimb during skilled reach to grasp movement and that sensory and motor-related information converges on a single neuron, allowing for direct integration at the cellular level.

All procedures were approved by the University of Otago Committee on Ethics in Care and Use of Laboratory Animals and were in accord with the “Principles of Laboratory Animal Care” (National Institutes of Health publication number 80-23, revised 1996).

The behavioural and single-neuron recording methods have been previously described in detail (Hyland & Jordan, 1997; Parr-Brownlie & Hyland, 2005). Male Wistar rats (~ 400 g at the time of surgery) were trained to reach into a 25-mm-wide, 14-mm high rectangular opening 55 mm above the floor to retrieve chocolate flavoured breakfast cereal (Coco Pops™) from a tray. Once a clear paw preference for reaching had been established, chronic extracellular recording electrodes were implanted into the contralateral PPTg (anteroposterior -7.8 mm and lateral ± 2.0 mm, relative to bregma (Paxinos & Watson, 1997)) using sterile stereotaxic technique under full anaesthesia. 

During recording sessions, signals from electrodes were amplified and filtered (2000 -8000 x, 0.5 - 10 kHz bandpass). Extracellular action potentials from single neurons were recorded, along with marker channels for automatically or manually entered behavioural events using SciWorks software (Datawave, CO). When a suitably isolated neuron was identified, it was first recorded while animals made ~ 50 reaches. Following this, for a subset of neurons recordings were made while the rats were exposed to 50 trials of 50 ms duration 4.5 kHz tone stimuli (Med Associates Sonalert), with pseudorandom intertrial intervals.  

Unit activity was analysed off-line using Spike2 software (CED, Cambridge, UK). Spikes belonging to single neurons were discriminated based on waveform shape (Fig. 1 B). For analysis of reach- and tone responses, activity was averaged over the reaching trials or tone presentations in peri-event time histograms (PETH). For reach data, PETH was centred on the interruption of an infrared beam at the position of the food. We chose this moment in the reach sequence because we were interested in determining if PPTg activity may correlate to distal skilled components of the reach-to-grasp movement, and this marks the termination of the extension phase and onset of the complex movements of the paw that are generated for grasping (Whishaw & Pellis, 1990; Hyland & Jordan, 1997; Sacrey et al., 2009). For the reach-PETH, cell activity was averaged across trials, extending for 2 seconds before and after the light-beam interrupts generated by the preferred paw (contra-lateral to the recorded hemisphere) and normalised by converting to instantaneous frequency (spikes/s) (Fig. 1 C). The first 500 ms of each PETH was used to define the baseline firing rate. PETH for slow-firing cells used 50 ms bin width and for fast-firing cells, 25 ms was used. Peaks in the PETH were defined as a group of consecutive bins that were beyond + 2 standard deviations (SD) from the mean baseline firing rate. For fast-firing cells, three consecutive 25ms bins were required, for slow-firing cells, two 50 ms bins. For inhibitions, the same criteria were applied using a threshold at – 2 SD, except for cases in low-firing cells where this threshold value was < 0 spikes/s. In these, we used the criterion defined by Galvan et al. (2016), where troughs were defined as periods in the PETH with zero bin counts that lasted ≥ 2 bins longer than any silent period in the baseline. The time of the modulation onset was defined as the leading edge of the first bin to cross the threshold, and the duration measured to the trailing edge of the last bin remained beyond the threshold. Onset times are expressed relative to the time of the light-beam interrupt which marked both the end of the ballistic extension phase of the reach and the onset of grasping. The amplitude of peaks was quantified as the mean of the in-peak bins. For analysis of firing patterns, autocorrelation histograms were calculated from selected periods around or between reaching events. For tone response PETH, similar methods were used except that the PETH was centred on the onset of the tone and a 2 ms bin width was used to capture the onset time of the short latency responses.

Statistical analyses were performed using GraphPad Prism Version 8 (GraphPad Software LLC).

To assist in the demarcation of the timing of overall population responses separately for 1^st (+) and 1^st (-) responses we performed a boxcar analysis on the average normalized Z-score in 50 ms bins using one-sample t-tests to determine deviation from baseline (Z = 0), over the period – 2 to + 1.5 s (n = 71 bins), with overall alpha = 0.05. Bonferroni correction for multiple comparisons (p = alpha/n) yielded p = 0.0007 to reject the null hypothesis. We therefore set p = 0.0001 as a conservative threshold to define a significant population-level response. To mark the onset and offset times for such responses we used the time at which p values crossed 0.001 before and after reaching the threshold. To test whether the timing of onset of the population (+) and (-) response components might differ, we performed a two-factor repeated measures ANOVA for individual neuron’s original (unsmoothed) PETH bin data for the factors RESPONSE TYPE (+, -) and TIME (71 time points from -2 to 1.5 s). To account for variation in baseline firing rate between neurons, data were normalised by calculating Z-score values for each PETH bin as the difference between the bin firing rate and the mean baseline rate (-2 to -1.5 s), divided by the standard deviation of the baseline rate. For 1^st (-) neurons Z-score values were inverted so that the analysis focussed on each group’s deviation from 0, rather than the sign of the response. Sphericity was not assumed, with Geisser-Greenhouse correction as required.

To objectively quantify the instantaneous frequency of regular repeated peaks in PETH and autocorrelation histograms we used the Fit Data function in Spike2 v7 to fit a sinusoid function to ranges of data that were initially visually selected as potentially containing oscillatory activity. The least-square fitting function minimised the sum of squares of the errors between the data and the fitted curve, and the Spike2 algorithm calculated an estimate of the probability that a minimum sum of squares of errors of at least this size would occur (Spike2 for Windows Manual v7, 2017).