Motor, somatosensory, and executive cortical areas elicit monosynaptic and polysynaptic neuronal activity in the auditory midbrain

We recently reported that the central nucleus of the inferior colliculus (the auditory midbrain) is innervated by glutamatergic pyramidal cells originating not only in auditory cortex (AC), but also in multiple ‘non-auditory ’ regions of the cerebral cortex. Here, in anaesthetised rats, we used optogenetics and electrical stimulation, combined with recording in the inferior colliculus to determine the functional influence of these descending connections. Specifically, we determined the extent of monosynaptic excitation and the influence of these descending connections on spontaneous activity in the inferior colliculus. A retrograde virus encoding both green fluorescent protein (GFP) and channelrhodopsin (ChR2) injected into the central nucleus of the inferior colliculus (IC C ) resulted in GFP expression in discrete groups of cells in multiple areas of the cerebral cortex. Light stimulation of AC and primary motor cortex (M1) caused local activation of cortical neurones and increased the firing rate of neurones in IC C indicating a direct excitatory input from AC and M1 to IC C with a restricted distribution. In naïve animals, electrical stimulation at multiple different sites within M1, secondary motor, somatosensory, and prefrontal cortices increased firing rate in IC C . However, it was notable that stimulation at some adjacent sites failed to influence firing at the recording site in IC C . Responses in IC C comprised singular spikes of constant shape and size which occurred with a short, and fixed latency (~ 5 ms) consistent with monosynaptic excitation of individual IC C units. Increasing the stimulus current decreased the latency of these spikes, suggesting more rapid depolarization of cortical neurones, and increased the number of (usually adjacent) channels on which a monosynaptic spike was seen, suggesting recruitment of increasing numbers of cortical neurons. Electrical stimulation of cortical regions also evoked longer latency, longer duration increases in firing activity, comprising multiple units with spikes occurring with significant temporal jitter, consistent with polysynaptic excitation. Increasing the stimulus current increased the number of spikes in these polysynaptic responses and increased the number of channels on which the responses were observed, although the magnitude of the responses always diminished away from the most activated channels. Together our findings indicate descending connections from motor, somatosensory and executive cortical regions directly activate small numbers of IC C neurones and that this in turn leads to extensive polysynaptic activation of local circuits within the IC C .


Introduction
It is becoming increasingly clear that identifying and responding to objects and events in the natural world involves interactions between the different senses, and the integration of sensory and motor processing in the brain (Bizley and Dai, 2020;Bizley and King, 2008;Brooks and Cullen, 2019;Kayser and Logothetis, 2007;Lohse et al., 2022;Schneider, 2020).
In perceptual processing, it is well established that behaviourally congruent information in one sensory domain can enhance perception in another, as illustrated by the benefit conferred by seeing the speaker's lips in perceiving speech (Grant and Seitz, 2000;Peelle and Sommers, Abbreviations: AAV, Adeno-associated virus; Cg, Cingulate cortex; ChR2, Channel rhodopsin; DP, Dorsal peduncular cortex; GFP, Green fluorescent protein; IC, Inferior colliculus; ICC, Central nucleus of the inferior colliculus; IL, Infra limbic cortex; M1, Primary motor cortex; M2, Secondary motor cortex; MC, Motor cortex; PFC, Prefrontal cortex; PrL, Prelimbic cortex; PSTH, Post-stimulus time histogram; S1FL, Somatosensory cortex forelimb area; S1HL, Somatosensory cortex hindlimb area; S1Jaw, Somatosensory cortex jaw area; SSC, Somatosensory cortex.2015; Sumby and Pollack, 1954).The corollary also holds that incongruent sensory information in one domain can disrupt perception in another, as classically demonstrated by the McGurk effect (McGurk and Macdonald, 1976).For the most part, sensory-motor interactions are considered to occur at the level of the cerebral cortex and multi-sensory interactions involve the cerebral cortex as well as the superior colliculus.Numerous structural and functional studies in animals and humans have derived evidence for sensory-motor and multi-sensory interactions at primary and higher cortical levels (e.g.(Bizley and King, 2008;Cappe et al., 2009;Di Marco et al., 2021;Jorge et al., 2018;Martuzzi et al., 2006;Porada et al., 2019).Thus, widespread cortico-cortical projections likely mediate many cross-modal interactions.But in the case of the auditory system, the situation might be different for, unique amongst the sensory modalities, the auditory system features extensive pre-cortical (brainstem) processing networks.These are a potential substrate for multi-sensory and motor-auditory integration much earlier in the pathway, as well as providing a mechanism by which executive function might influence auditory processing.
The inferior colliculus (IC, the auditory midbrain) represents the culmination of brainstem auditory processing and the source of almost all ascending auditory input to the thalamus and beyond (Rees, 2020;Ito and Malmierca, 2018).The central nucleus of the IC (ICC) is the main recipient of afferent inputs from parallel pathways established in the brainstem.It is well established that there are also descending connections from the auditory cortex (AC) to the IC (Andersen et al., 1980;Coleman and Clerici, 1987;Feliciano and Potashner, 1995;Saldana et al., 1996;Bajo and Moore, 2005;Blackwell et al., 2020;Syka et al., 1988).Although projections from auditory cortex to the dorsal and lateral cortices are the most intensively studied, AC projections also terminate in the ICC (Bajo and Moore, 2005;Saldana et al., 1996).Intriguingly, we have recently presented evidence that the ICC also receives non-auditory cortical inputs.Using both anterograde and retrograde tracing, we demonstrated the existence of inputs to the ICC from the visual (VC), somatosensory (SSC), motor (MC) and prefrontal cortices (PFC) (Olthof et al., 2019).These inputs are glutamatergic, and they terminate on both GABAergic and non-GABAergic (putative glutamatergic) neurons in the ICC (Olthof et al., 2019).Here we used a retrograde virus encoding GFP to label direct descending connections to ICC in the rat.We then used both optogenetic and electrical stimulation to selectively activate areas of MC, SSC, and PFC and combined this with neuronal recording across the ICC.By determining the distribution of stimulus and response sites , the temporal characteristics of responses, and their dependence on stimulus intensity, we determined the extent and the nature of the influence of non-auditory cortical regions on spontaneous activity in ICC.

Animals
Male Sprague Dawley rats (Charles River) were housed in groups of 3 in two-storey individually ventilated cages under standard conditions (12 h light/dark cycle lights on 7am; humidity 45-55 %, temperature 21-23 • C).The environment was enriched with cardboard tubes and wooden blocks/balls.Animals were allowed free access to food (rat chow, SDS) and tap water.Animals were allowed to acclimatize for at least 3 days before experiments began.We used only male animals to reduce housing costs.

Injections of retrograde virus encoding GFP and channelrhodopsin
Rats were anaesthetised with isoflurane (5 % in O 2 , 0.1 l/min) and ketamine/medetomidine (15/0.2mg/kg, i.p.).The head was shaved, and the animal was moved to a sound-attenuated room and placed in a stereotaxic frame.The head was fixed using a custom-made tooth/palate bar and zygoma bars (WPI).The animal's temperature was maintained with a homeothermic blanket set to 37 • C controlled by a rectal probe (Harvard).Anaesthesia was maintained with isoflurane in O 2 delivered via a modified paediatric nasogastric cannula inserted into the nostril.The concentration of isoflurane was adjusted to maintain a surgical plane of anaesthesia.A small-animal pulse oximeter connected to a MouseVent G500 (Kent Scientific), clipped to one of the hind paws, was used to monitor arterial PO 2 and heart rate.
The scalp was cut in the midline and the periosteum scraped to reveal the skull bones and sutures.Using a dental burr, a craniotomy was drilled (AP: − 8.8 mm, ML: 2.5 mm from bregma) to allow access to the right inferior colliculus.A multichannel electrode was implanted and used to record sound related activity in IC (see below).The electrode position was adjusted to optimise the number of channels located in the ICC and these stereotaxic coordinates were used to guide the virus   A glass capillary pulled to a fine tip was pre-loaded with a retrograde virus expressing both GFP and channelrhodopsin (ChR2) (pAAV-Syn-ChR2(H134R)-GFP (Addgene, 7 × 10 12 vg/ml in PBS/NaCl/pluronic acid F-68) and fitted to an injection device (Nanoject, Drummond Scientific).The recording electrode was removed from the ICC and the glass pipette was implanted in its place.Virus was injected into the ICC (200-300 nl at three or four positions along the dorsoventral axis).To avoid leakage of virus out of the ICC, the pipette was left in place for 5 min after the final injection before being withdrawn.The animal was given postoperative analgesia (meloxicam, 1 mg/kg, s.c.).The scalp was sutured (Vicryl 4.0) and the animal was removed from the stereotaxic frame and allowed to recover from the anaesthetic.Post-operatively, animals were housed in groups of 2 or 3 and were given soft diet for 1-2 days.They received an additional dose of meloxicam (1 mg/kg, s.c.) on the day after surgery.

Placement of recording electrodes in ICC
A multi-channel silicone electrode (10 mm single shank linear 32 channel electrode, 100 µm pitch of recording sites, A1×65-10mm-100-177-A32 50, Neuronexus) was implanted into the IC.To avoid the superior and transverse sagittal sinuses, the electrode was implanted at an angle of 12 • to the vertical and 45 • from the sagittal plane (i.e., such that the electrode passed dorsoventrally from a rostral/lateral position to a caudal/medial position).The electrode was connected via a preamplifier (PZ2-32) to an RZ2 BioAmp (Tucker Davis Technologies, TDT) to amplify and record the neural signals.Voltages from 32 electrode channels were digitized at a sampling rate of 24,414 Hz.
Pure tones at varying frequencies and amplitudes (75 ms duration, 512-16,384 Hz, 5-100 dB SPL), generated by an RZ6 Multi I/O Processor (Tucker Davis Technologies), were delivered to the animal's ear canals via 10 cm lengths of Zygon tubing attached to MF-1 Multi-Field Magnetic Speakers (TDT).Frequency response area (FRA) plots of sound-evoked activity were constructed on-line (see Data analysis).Channels were considered to be in the ICC if they showed frequencytuned, sound-driven, activity characterised by excitatory FRAs with a clear tonotopic progression (channels responding best to low frequencies located dorsally and channels responding best to high frequencies located more ventrally, see Fig. 1).The electrode position was adjusted to optimise the number of channels located in the ICC.

In vivo electrophysiological recording: optogenetic stimulation in virus injected rats
Electrophysiological recordings were conducted 12-40 days post virus injection.Animals were anaesthetized with urethane (1 g/kg i.p.) and a mixture of fentanyl/midazolam (Hameln) (0.3/5 mg.kg -1 i.m. in the biceps femoris).A tracheotomy was performed, and a polythene tube was inserted and tied into the trachea.Animals were moved to a soundattenuated room and fixed in a stereotaxic frame (Kopf) on a homeothermic blanket, as above.To ensure good oxygen saturation throughout the experiment, the tracheal tube was connected to a pumped supply of O 2 and the pumping rate was adjusted to be marginally quicker than the animal's spontaneous breathing.Arterial PO 2 and heart rate were monitored as above.
The scalp was cut along the midline, and the periosteum scraped away to reveal the skull bones and sutures.The previous craniotomy over the IC was reopened.Further craniotomies were drilled to allow access to the AC and M1.For access to AC, the craniotomy was made medially, and the stimulating electrode was implanted at an angle of approximately 20 ˚mediolateral to the midline.The correct positioning of the electrode in AC was verified by the neural response to clicks.
To activate cortical neurones optogenetically, a glass fibre patch Fig. 4. Optogenetic stimulation activates units in cortex.a the effect of three ramped light stimuli (0-5 V, 5 s) in AC on activity in three channels in AC. b the effect of two ramped light stimuli (0-5 V, 5 s) applied to M1 on activity on three channels in M1.Note that in each case the light stimulus increases the firing rate, that the responses to repeated stimuli are similar, and that multiple channels are affected.Top line in each part shows the ramped light stimulus (0-5 V).Data were recorded on a 4 shank, 32 channel multielectrode probe.
cable (M126L01 400 µm, 0.5 NA, Thorlabs, Ely, UK) was connected to a 470 nm LED (M470F3, Thorlabs, Ely, UK) with DC variable voltage LED driver (LEDD1B, Thorlabs, Ely, UK).Light was delivered to the cortical surface via a 10 mm glass-fibre cannula (0.5 NA, CFMC54L10, Thorlabs, Ely, UK) connected to the patch cable by a mating sleeve (ADF2, Thorlabs, Ely, UK).The output at 5 V at the end of the fibre optic was 6.2 mW.Ramped stimuli (0-5 V, over 5 or 10 s) were applied using a custom script in Spike2 (CED, UK).To examine local effects on cortical neurones, recordings were made using multichannel electrodes with four shanks and 8 channels per shank (Neuronexus, A4×8-5 mm) implanted into the cortex close to the glass fibre.To examine the effects of optogenetic cortical activation on activity in the ICC, recordings were made using a linear 32-channel electrode (Neuronexus, A1×32-10 mm) implanted into the IC.The location of the electrode within the ICC was optimised using the FRA as described above.

In vivo electrophysiological recording: electrical stimulation in naïve rats
Naïve male Sprague Dawley rats (median 350 g, range 240-386 g) were anaesthetized and prepared for surgery as described above.Craniotomies were drilled over the right IC and one or more of the following ipsi-and/or contralateral cortical regions: primary and secondary motor cortex (M1 and M2); somatosensory cortices-jaw region (S1Jaw), forelimb region (S1FL), hindlimb region (S1HL), barrel cortex, and medial prefrontal cortex (PFC), as delineated in the rat brain atlas of Paxinos and Watson (1998).Note: for access to barrel cortex, the craniotomy was made close to the midline and the stimulating electrode was implanted at an angle mediolateral to the midline.
A concentric, bipolar stimulating electrode with poles separated by 500 µm (Clarke Electromedical) was implanted in the cortical region of interest targeting layer V.The two poles of the stimulating electrode were connected to a battery powered constant current stimulus isolator (Isoflex, AMPI, Israel).Electrical stimulation (triggered by the TDT system) were delivered as single cathodal square pulses of 100 µs duration at a rate of 2.5 Hz (i.e. one pulse every 400 ms) for a period of 40 s (100 sweeps, denoted as a stimulus 'block').To examine the threshold and current dependence of responses, the output of the stimulus isolator was varied between 0.3 and 8 mA for different blocks.Multiple cortical sites were stimulated in individual animals.Data were recorded from the 32 channels of the IC electrode during electrical stimulation of the cortex.

Data analysis
To construct FRAs online, the signal was bandpass filtered (300-3000 Hz) and thresholded at ± 3 x the standard deviation of the average signal.Positive or negative voltage deflections greater than threshold were counted as spikes.
For all other analysis, we used a refinement of this spike thresholding script.The signal was filtered at 500-5000 Hz.A threshold was set at ± 2.5 x the standard deviation of the voltage signal across the 40 s data 'block' (excluding a period of 10 ms before and 80 ms after each stimulus to avoid the stimulus artefact and any response).Where the voltage passed this positive or negative threshold, a spike was counted.To avoid double counting of spikes, a positive and a negative deflection that occurred within 1 ms, was counted as a single, bipolar, spike.Peristimulus time histograms (PSTHs) were constructed by binning the spike data into 1 ms bins across the 400 ms sweep and summing the spikes in each bin over the 100 sweeps.The spontaneous firing rate was defined as the average rate in the last 40 ms of the sweeps.The total spikes presented in the PSTH is the sum of positive (only), negative (only), and bipolar spikes.
Having examined the nature of the excitatory and inhibitory responses (described below in Results), we designed custom scripts in Matlab to identify channels in the IC which showed a response to

Table 1
Stimulation sites evoking responses in ICC.In most cases multiple stimulation sites within a cortical region were tested.Not all sites were tested in all animals.Y indicates that electrical stimulation in this region elicited a response in ICC.N indicates no detectable response in ICC.Numbers in brackets are number of stimulation sites from which a response was evoked/number of sites tested.
electrical stimulation of the cortex, and to quantify the parameters of these responses.To avoid contamination of the response by the electrical stimulation artefact, responses were measured starting 3 ms from the time of electrical stimulation.We used the following criteria to define responses to electrical stimulation: an excitatory response was defined as a group of at least 6, 1 ms bins (consecutive or separated by 2 or fewer bins) in which the firing rate was at least 2 spikes above the spontaneous firing rate, and which contained at least 60 spikes in total.An inhibitory response was defined as at least 60 connected bins in which the moving average over 20 ms bins was at least 2 spikes below the spontaneous rate, and the overall average in that period was at least 4 spikes below the spontaneous rate.showing that increasing current results in a progressive decrease in the latency of the spike.Note also that the size of the averaged event increases with increasing current, reflecting the fact that at higher currents, the event occurs with greater probability.The FRA for his channel is shown in Fig. 6d.b group data from 17 M1 stimulation sites in 3 animals in which between 2 and 6 different stimulus currents (0.3-8 mA) were tested.Note that there is a decrease in latency with increasing stimulus current and an increase in the number of channels (spots) on which a spike is seen.
In order to group data from equivalent positions in the ICC (i.e. with similar frequency selectivity), the data from the 32 recorded channels were aligned on the basis of the low frequency end of the tonotopic representation in the ICC (see Fig. 1).

Injection of GFP expressing retrograde virus into the ICC labels brainstem auditory nuclei
The site of injection of retrovirus was evident within the ICC with retrograde labelling of its intrinsic connections.There was minimal spread to structures adjacent to the IC such as the superior colliculus.In some cases, labelling was also visible in the visual cortex which is likely a combination of virus deposited as the pipette was withdrawn, and retrograde labelling from the IC.Following injection of retrograde virus encoding GFP into the ICC, GFP was expressed in brainstem auditory structures as expected.Cell bodies of neurones expressing GFP intensely were present in the lateral lemniscus and brainstem olivary nuclei (Fig. 2).

Injection of GFP expressing retrograde virus into the ICC labels multiple cortical regions
Following injection of retrograde virus encoding GFP into the ICC, GFP was also expressed in multiple cortical regions both ipsilateral and contralateral to the injection.In coronal sections, retrogradely labelled cells were evident in deep layers of cortex.Cell somata were intensely labelled as were dendrites extending into more superficial layers.Fine axons could also be seen exiting the cortex and entering the white matter (Fig. 3).Although we injected virus at multiple points in the ICC to maximise coverage, GFP was expressed by relatively small groups of cells in restricted areas of cortex.The highest density of GFP expressing cells was in the ipsilateral AC, but GFP expressing cells were also evident in VC, S1, M1, and PFC.

Local optogenetic activation of auditory cortex (AC)
The retrograde virus we used also expressed channelrhodopsin (ChR2).To verify expression of ChR2 we examined the effect of blue light stimulation (470 nm) on neuronal activity in the cortex using a 'ramp' of light of increasing intensity to avoid desensitization (Adesnik and Scanziani, 2010).Application of blue light to the AC (0-5 V ramped stimuli of 5 or 10 s duration), evoked increases in neuronal activity which began a few hundred milliseconds after the light onset and ended as the light stimulus was switched off.We were able to evoke activity in AC in 7 of the 8 virus injected animals tested (see Fig. 4a).Responses were seen on many channels spanning a considerable area of AC.Responses were reproducible on repeated stimulation (Fig. 4a) and the magnitude (number of channels activated or size/duration of response) was dependent on the light intensity (data not shown).

Local optogenetic activation of M1/SSC
Application of blue light to the M1 and somatosensory cortices (SSC) also evoked an increase in activity in these areas (Fig. 4b).Responses were like those evoked in the AC, indicating that the ChR2 was also functionally expressed in these regions.However, although we evoked activity in M1 and the border of M1/S1 in 4 of 9 animals tested, the regions where activation could be seen were small and we failed to evoke activity at adjacent sites.Moreover, in several other animals, we failed to find regions in M1 or S1 where the light stimulus evoked measurable increases in activity locally.

Optogenetic activation of AC and M1 evokes increased firing in ICC
Having verified that we could evoke firing activity in cortical neurones which project to the ICC (i.e. that contain ChR2 expressed by a retrogradely transported virus), we next determined whether stimulating these neurones could elicit activity in the ICC, as would be expected if the observed cortical labelling is due to retrograde transport of virus from the ICC.As shown in Fig. 5a, optogenetic stimulation of AC increased the firing rate of neurones in the ICC.The increased firing rate occurred as the ramp of light increased and ceased when the light was off.The effect was evident on several adjacent channels in ICC.We observed this excitatory response in 2 out of the 7 animals tested.In the 5 other animals tested, optogenetic stimulation at multiple sites in AC failed to elicit responses at the recording site in ICC.
We also examined responses in ICC to optogenetic activation of M1.In one animal, in which we had seen local optogenetic activation of multiple channels in M1, we saw clear activation of several channels in ICC during optogenetic activation of M1 with the same parameters (Fig. 5b).The increased firing rate occurred as the ramp of light increased and ceased when the light was off.Effects were seen on 6 adjacent channels in the ICC spanning 500 µm.It was notable that for individual channels in ICC, we could see that the response involved an increase in the firing rate of several different units (e.g., Fig. 5c).However, we saw ICC responses to M1 activation in only one of 6 animals tested.
Our ability to optogenetically activate units in ICC from AC and M1 indicates that descending cortical neurones have a direct excitatory influence in the ICC.Consistent with our anatomical data, we found that connections between M1 and ICC were localized.Given that the retrograde virus does not cross synapses, the optogenetic method suggests that these connections are direct.However, using this method we could not determine the temporal relationship between individual firing events in cortex and firing events in ICC.Thus, we could not be sure whether individual spikes in ICC were monosynaptic responses to cortically originating spikes.We also wanted to know whether cortical stimulation elicits subsequent local circuit activity in the ICC, and to establish the spatial and temporal extent of any such activation.To address these points, we used electrical stimulation to activate neurones Fig. 9. Increasing the stimulus current in M1 increases the magnitude of excitatory responses in ICC and the number of channels affected.Sum of the PSTH response evoked by electrical stimulation.Lines show mean response and shading indicates SEM.Data are from 51 M1 stimulation sites in 7 animals for which between 2 and 6 stimulus currents were tested.Note that at low currents, responses are small and are only evident on a few channels (representing a small area of ICC), while at higher currents, responses are greater and are seen on more channels over a wider area of ICC.Note, however that there is a general decrease in the magnitude of the excitatory response both dorsal (left) and ventral (right) of the channels affected most by the lowest currents.Greyed out portions indicate data collected on channels dorsal or ventral to ICC. in the cortex while recording multiunit activity in ICC.Monosynaptic responses would be expected to be time locked to the stimulus with a short latency and be all-or none, while changes in firing rate with longer more variable latencies would be evidence of polysynaptic activation.

Electrical stimulation of M1 modulates activity in the inferior colliculus
Electrical stimulation (0.3-8 mA, 100 µs) of multiple sites within M1 increased the firing rate of units in the ipsilateral ICC.However, responses were highly localized, such that electrode sites within 200 µm of an 'active' stimulation site, sometimes failed to elicit any response at a particular recording site in ICC (Fig. 6).In total we saw excitatory responses in the ICC in all 7 animals tested (Table 1) at 26 out of 51 M1/ICC recording/stimulation site pairings (Fig. 6).

Spikes with short, fixed, latency
To determine whether there was evidence of monosynaptic activation of units in ICC, we examined the responses to individual stimuli (sweeps) on individual channels in ICC.In many cases, we could see a spike of relatively consistent size and shape which occurred with a fixed latency around 5 ms from the stimulus (e.g., Fig. 7).
To capture this event, we averaged the response to the 100 stimuli on each channel for each stimulus/recording site pairing and examined the period from 3 to 7 ms from the stimulus.A short latency spike was present in the averaged signal in at least one channel in 3 out of the 7 animals in a total of 17 out of the 51 M1/ICC stimulation/recording site pairings.Increasing the stimulus current increased the number of Fig. 10.Evidence that electrical stimulation of M1 evokes a polysynaptic excitatory response in ICC involving multiple units with temporal jitter.ai responses to 3 stimuli recorded on a single channel in ICC .Note the periods of increased activity following the stimulus artefact (large deflection, blue arrowhead at 0 ms) which comprise multiple spikes with different shapes and sizes some of which are highlighted in different colours; aii the spikes highlighted in ai shown on an expanded scale; Inset shows the FRA plot for this channel; aiii responses to 100 stimuli overlaid showing the temporal distribution of spikes; bi responses to 6 stimuli recorded on a single channel in ICC.Note that some stimuli elicit a short, fixed, latency spike (highlighted in red) and that there is a later period of increased firing.In this example, there is one large spike (highlighted in green) which shows marked temporal jitter (green shading); inset shows the FRA for this channel; bii response to 100 stimuli overlaid showing the temporal distribution of spikes.Note that in both examples, the number of spikes elicited varies between sweeps.Note also that the response is longer in example a than in b.
We also noted that the size of the averaged spike increased with increasing current.However, in individual sweeps, where spikes were resolvable from noise, they were all-or-none.Thus, the increase in the size of the averaged spike reflected an increase in the probability of the spike occurring in response to an individual stimulus as stimulus current was increased.These features are characteristic of a monosynaptic excitatory input from M1 to ICC.

Excitation: longer latency increases in firing
In addition to short, fixed, latency single spikes, electrical stimulation of M1 evoked a longer duration period of increased firing activity.The latency of this response was usually between 5 and 10 ms after the stimulus artefact (average 6.8 ms, range (4.0-20 ms), n = 112 responses), and firing rate was elevated for 10-40 ms (average 19.1, range 9.0-35.3,n = 112).
Defining a 'response' as the sum of the PSTH for 100 stimulus presentations (see Methods), in total we identified a later excitatory response in all 7 animals tested in 26 out of 51 recording/stimulus site pairings.With increasing current, the magnitude of the response increased (Fig. 9) but the duration was unaffected), indicating that this was an increase in the firing rate of individual units or an increase in the number of units activated rather than a current related prolongation of the response.The number of channels on which a response was seen (i.e., the distance within the ICC) also increased with increasing current (Fig. 9).At the highest currents, excitatory responses were frequently seen on 13 or more channels representing an overall linear distance in the ICC of 1300 µm (average number of channel on which a response was observed: 0.3 mA: 3.3 (range 0-8); 0.5 mA 7.4 (range 0-18); 1 mA: 11.7 (range 1-26); 3 mA 13.0 (range 1-23); 5 mA 13.1 (2-24); 8 mA: (15.6) (range3-23)).It was notable that, for any given current, the magnitude of the response diminished for recording locations both dorsal and ventral to the most affected channel although this diminution was not symmetrical (Fig. 9).
In cases (stimulation/recording site pairs and channels) where electrical stimulation of M1 evoked a monosynaptic spike in ICC (see above), this was always followed by a later period of increased firing activity (e.g., Fig. 10aii).In many cases (neighbouring channels and/or other stimulation/recording site pairs), where we did not observe a monosynaptic spike, electrical stimulation of M1 nevertheless evoked a longer latency period of increased firing (e.g., Fig. 10ai).To explore the nature of these later excitatory responses, we examined responses to individual stimuli.Although we could not sort individual units from our multiunit recordings, it was clear from spike size alone, that multiple units contributed to the stimulus-elicited firing (Fig. 10a, b and c).Examination of responses to individual stimulations showed that the longer latency firing was stochastic and that individual spikes occurred with a degree of temporal jitter.In some cases, we could tentatively identify a single unit with increased activity during this period (e.g., unit highlighted in Fig. 10d) and could see that the latency at which this individual unit fired varied from sweep to sweep by several milliseconds.

Inhibition
In some animals, at some stimulation/recording site pairings, electrical stimulation of M1 evoked inhibition of firing in the ICC (Fig. 11).In general, inhibition occurred with both a longer latency, and a longer duration than the previously discussed excitatory responses (note the time scale in Fig. 10).This type of inhibition occurred independently of whether excitation was observed on the same channel (c.f.Figs.11a and  b).Very occasionally, we observed inhibition with a short latency and duration (e.g., Fig. 11c).Note that due to the low firing rate in ICC in the absence of auditory stimulation, the inhibition was always of limited magnitude.
In summary, electrical stimulation of M1 altered firing rate in ICC.We observed both short, fixed, latency monosynaptic spikes and later, longer duration periods of increased firing activity, involving multiple units with marked temporal jitter, characteristic of polysynaptic excitation.Long latency inhibition was also seen on a few occasions.We confirmed the biological nature of these responses by verifying that when we applied electrical stimulation post-mortem, only the stimulus artefact was apparent (data not shown).

Electrical stimulation of secondary motor cortex (M2) also evokes excitatory responses in ICC
Following electrical stimulation of M2, we also saw excitatory responses in ICC in 5 out of 6 animals at 14 (out of 24) stimulation/ recording site pairings (Fig. 12).The large deflection at 0 ms is the electrical stimulation artefact.Coloured bars represent periods of excitation (cyan) and inhibition (magenta).a an example where a substantial increase in firing activity with short latency and relatively low jitter is followed by a later period of inhibition: inset is the FRA plot for this channel.b an example in which a short period of inhibition follows a short latency excitation: inset is the FRA plot for this channel.Note that this channel exhibited an unusually high spontaneous firing rate.

M2: short fixed latency spikes
Early fixed latency spikes were seen in 2 (out of 6) animals (7 (out of 24) stimulation/recording site pairings) (e.g.Fig. 13b).Increasing current decreased the latency, and increased both the probability of a spike and the number of channels on which a spike was detected (Fig. 13).

M2: longer latency increases in firing
Irrespective of whether an early event was seen, electrical stimulation of M2 evoked a later period of increased firing comprising multiple spikes with temporal jitter (5 out of 6 animals, 14 out of 24 stimulation/ recording site pairings) (e.g., Fig. 13a).Responses occurred on multiple channels in ICC with more dorsal channels affected at the lowest currents and to the greatest degree (Fig. 13d).With increasing current, the number of spikes on individual channels was increased and responses were seen on more channels-in particular channels located more ventrally (Fig. 13d).The later excitatory responses evoked by electrical stimulation of M2 were of similar latency, duration and magnitude to those evoked by stimulation of M1 (c.f.Figs. 9 and 13).Occasionally there was a period of inhibition (data not shown).

Electrical stimulation of somatosensory regions evokes increased firing in ICC
In addition to the motor areas M1 and M2, we examined the effect of electrical stimulation of several ipsilateral somatosensory cortical regions (S1Jaw, S1Forelimb (S1FL), S1Hindlimb (S1HL), barrel field) with several different cortical sites tested in each animal (see Table 1 and Fig. 14).We found excitatory responses in ICC following stimulation of some sites within all of these somatosensory regions.

Somatosensory regions: short, fixed, latency spikes
Early fixed latency spikes were seen following stimulation of the S1Jaw region (2 out of 4 animals, 4 out of 14 stimulation/recording site pairs); S1FL (3 out of 5 animals, 6 out of 12 stimulation/recording site pairings); S1HL (2 out of 2 animals, 2 out of 2 stimulation/recording site pairings); and barrel cortex (3 out of 6 animals, 5 out of 15 stimulation/ recording pairings).For all somatosensory cortical regions, the latency of the putative monosynaptic spike decreased, and the number of channels on which a spike was seen increased with increasing current (1-8 mA) (Fig. 15aii, bii, cii).

Somatosensory regions: longer latency increase in firing
As in the motor regions, irrespective of whether we saw an early event, stimulation of somatosensory regions frequently evoked a later excitatory response which took the form of an increase in firing activity of multiple units with temporal jitter.This type of response was observed in S1Jaw (all 4 animals and all 13 stimulation/recording site pairs tested); S1FL (all 5 animals and 10 out of 12 stimulation/recording site pairings); S1HL (2 animals and 2 stimulation/recording pairings); and barrel cortex (6 animals and 11 stimulus/recording site pairings) (examples shown in Fig. 15ai, bi, ci).With increasing stimulation current, the number of channels on which the long latency responses were seen increased (Fig. 15aiii, biii, ciii).The response magnitude also generally increased, although at the highest currents, responses plateaued and sometimes diminished.For all somatosensory regions, responses appeared strongest in the most dorsal parts of the ICC (Fig. 15aiii, biii, ciii).

Electrical stimulation of prefrontal cortex evokes increased firing in ICC
Responses in the ICC were also observed following electrical stimulation of the PFC.Because of the dorsoventral orientation of this region, we were able to sequentially stimulate all subregions (cingulate, Cg; prelimbic, PrL; infralimbic, IL; dorsal peduncular, DP).In all PFC subregions, electrical stimulation evoked excitatory responses in ICC.

Prefrontal cortex: short, fixed, latency spikes
A short, fixed, latency spike occurred in the ICC in 3 out of 6 animals tested at a total of 18 anterior posterior stimulus locations in PFC (2, 3, or 4 per animal).In 14 of these 18 cases, monosynaptic spikes in ICC were elicited by stimulation of all four PFC subregions.These monosynaptic spikes were evident on multiple channels in ICC and frequently, the same channels were activated from all subregions (e.g., inset in Fig 16).There was a decrease in latency of the monosynaptic spike with increasing current (Fig. 16) as well as an increase in the number of channels on which a monosynaptic spike was seen.There was no difference in the latency of the spike between subregions.

Prefrontal cortex: longer latency increase in firing activity
Electrical stimulation of PFC also evoked longer-latency increases in firing in ICC which comprised multiple different units.These longer latency responses were, in some cases, preceded by a monosynaptic spike.In a very few cases we saw a short latency event in the averaged signal without any longer latency increase in firing.For the longer latency responses, both the response magnitude and the number of channels affected was dependent on the stimulus current.In most cases, channels which showed activation from Cg also showed activation from other PFC Fig. 12. Electrical stimulation sites in M2.In 5 of the 6 animals, electrical stimulation of at least one site in M2 evoked an excitatory response in ICC.In one animal (ES08) only one site in M2 was tested and stimulation at this location failed to evoke a response at the recording site in ICC.Filled symbols represent sites from which an excitatory response was evoked.Open symbols represent sites from which no response was evoked.Different colours represent the different animals.Distances are anterior to Bregma.subregions).There were no notable differences in the response profile evoked by stimulation of the different PFC subregions (Fig. 17).

Contralateral cortical regions also have functional connections with ICC
In our previous tracing study, we observed substantial innervation of ICC from the contralateral cortex.Hence, in 3 animals we examined responses to stimulation at contralateral sites.In 2 animals we tested contralateral M1, M2 and S1Jaw and found excitatory responses in ICC from all areas.There were short, fixed latency monosynaptic spikes on some channels and longer latency, longer duration increases in firing rate on more channels.The latencies of the early and later events were not different to those following stimulation of the ipsi-lateral cortex.In two animals we tested contralateral PFC stimulation and found early and later excitatory responses in ICC in one case but not the other.The latencies of these responses were similar to those seen in response to stimulation of the ipsilateral PFC.

Discussion
Our results provide functional evidence that the descending excitatory connections from multiple non-auditory cortical areas, we previously reported using anatomical methods (Olthof et al., 2019), exert an important influence on the firing activity of neurons in the ICC via direct monosynaptic excitation and the activation of local excitatory, and to a lesser extent inhibitory, circuits in ICC.
Injection of retrograde virus encoding GFP and ChR2 into the ICC resulted in expression of GFP in cells in brainstem auditory nuclei as expected.We also observed GFP labelling in substantial numbers of pyramidal cells in layer V in AC as well as in multiple non-auditory, cortical regions confirming our previous findings using conventional tracers (Olthof et al., 2019).Interestingly, we found that, despite injecting virus over a wide area of ICC, cortical labelling was somewhat discrete with small clusters of neuronal somata expressing GFP.In some cortical areas (depending on the plane of sectioning relative to the orientation of neurones), we could also see GFP in dendrites extending to superficial cortical layers.GFP could also be seen in fine axons in the white matter below the cellular layers.
Consistent with the expression of virally encoded ChR2 as well as GFP, blue light stimulation of the cortex evoked firing activity in cortical cells.A ramped increase in light intensity, which minimises desensitization (Adesnik and Scanziani, 2010), evoked substantial increases in firing activity locally in AC, M1, and S1.However, it was of note that, the light-activated regions were discrete, and light failed to increase firing in many cortical regions consistent with the 'patchy' expression of GFP.Despite the restricted nature of the cortical activation, optogenetic stimulation of AC and M1 altered firing activity in ICC.The predominant response was an increase in firing rate sometimes evident across several hundred microns in ICC.Assuming that the cells in the cortically stimulated regions were retrogradely transfected from the ICC and that the virus does not cross synapses (Tervo et al., 2016), this is evidence that excitatory inputs to the ICC from AC and from M1 can directly influence the firing of neurones in the ICC.However, we found only small numbers (caption on next column) Fig. 13.Electrical stimulation of M2 evokes short, fixed latency spikes and longer latency/longer duration increases in firing in ICC.Example a. 20 sweeps from one channel in ICC showing a fixed latency spike (red arrowhead) occurs ~5 ms after the stimulus artefact (blue arrowhead); b the short, fixed latency event averaged over 100 sweeps.Group data showing that c. the latency of the early spike decreases with increasing current (data from 2 animals, 8 stimulation sites, in which between 2 and 5 different stimulus currents were tested.Spots represent different stimulation/recording site pairs); and d the magnitude of the later excitatory response increases with increasing current.Lines represent means and shading represents SEM (data from 5 animals, 14 stimulation/ recording site pairings).
of cortical cells could be activated at any one time, probably because of a combination of restricted light spread within the cortical tissue and patchy viral transfection.This, coupled with the likelihood that cortical neurones have a restricted pattern of innervation in the ICC, the probability of activating cortical neurones and simultaneously recording from their restricted targets is low using optogenetics.Furthermore, we cannot totally exclude the possibility that viral expression in the cortical regions stimulated with light originated from outside the IC.Electrical stimulation reduces the issue of matching stimulus and target regions and, importantly, allowed us to explore the temporal aspects of how cortical regions elicit responses in the ICC, in a manner not possible with optogenetic stimulation.An additional advantage of electrical stimulation is that its interpretation is not reliant on the retrograde transport of virus to the cerebral cortex from the ICC.
Electrical stimulation of multiple non-auditory cortical regions evoked excitatory responses in the ICC.The difference in 'successful activation' between electrical stimulation and optogenetic stimulation is likely explained by the greater spread of current and larger area of cortex depolarized by electrical stimulation versus optogenetic stimulation.Nevertheless, there were areas in the cortex from which electrical stimulation failed to evoke responses at the recording sites in ICC, and the failure to see putative monosynaptic responses in cases where only the longer latency activity was observed was also presumably because the position of the recording electrodes was not sufficiently closely matched to location where the cortical input terminated.
The nature of the responses to stimulation of motor and somatosensory cortical regions was relatively consistent.We frequently saw a singular short and fixed latency firing event of consistent size and shape which occurred with increasing probability as the stimulation current was increased.The short, fixed latency of the event suggest it is monosynaptic (Qi et al., 2020;Syka and Popelar, 1984) and its invariant size suggests that it represents a single spike from a single ICC neurone.The latency of this event decreased with increasing current consistent with higher intensity currents causing more rapid neuronal depolarization.We propose that such events are the result of activation of individual ICC neurones in response to excitation by the innervating cortico-collicular connections.Although we did not perform collision tests (Lipski, 1981), it is unlikely that the short latency event is antidromic as there is no known direct projection from the ICC to the cerebral cortex (Oliver and Cant, 2018).Monosynaptic spikes were seen in around 30 % of cases where there was evidence of cortical-ICC activation, their relative rarity is most likely due to the fact that an electrode contact would need to be close to the directly cortically innervated cell to pick up a monosynaptic response.
Independent of these monosynaptic spikes, the predominant response following electrical stimulation was a longer latency, longer duration increase in firing activity.It was clear from the presence of spikes of different sizes, shapes, and polarity, that this late period of increased firing involved multiple units.We also observed considerable temporal 'jitter' in the timing of individual spikes relative to the stimulus confirming that these responses are not antidromic and suggesting that they are polysynaptic in nature.The relatively long latencies observed would also suggest the involvement of multiple neurones activated by synaptic transmission.It is unlikely that the later polysynaptic events occur secondary to activation of a cortical circuit since the cortical regions from which the ICC can be activated are discrete.We cannot rule out the possibility that circuits outside the ICC (including in IC cortices and other auditory structures) contribute to these polysynaptic responses.However, we propose that the most likely explanation is that direct stimulation of cortico-collicular projections results in short latency monosynaptic activation of individual targeted connections in the ICC which in turn activates local circuits.The increase in response magnitude seen with increasing current likely reflects more consistent monosynaptic activation and activation of greater numbers of cortical neurones innervating ICC.With increasing current, excitatory responses also occurred on an increasing number of channels, particularly in more ventral regions of ICC.Nevertheless, the magnitude of responses on the channels recruited with higher currents was always lower than on those channels affected at lower currents.It seems likely that with higher currents, the stronger response results in further spread of local circuit activation within the ICC such that neurones more distant from the direct target of the cortico-collicular neurones become activated.In many of our recordings spontaneous activity was low and excitatory responses are most easily captured.However, as has been reported with electrical stimulation of AC (Jen et al., 1998;Syka and Popelar, 1984) we sometimes saw inhibition of firing associated with electrical stimulation.In the main, this inhibition occurred at longer latencies and was long lasting suggesting a polysynaptic mechanism within ICC.There are large numbers of GABAergic cells within the ICC which modulate the firing of other ICC neurones (Merchán et al., 2005).
It is likely that post-stimulus inhibition involves these neurones.It is likely that, owing to the low or absent spontaneous activity in our recordings, the incidence of inhibition was underestimated.
We examined the effects of stimulation of multiple cortical areas on ICC firing.In many cases with the recording electrode in the same position in ICC, we found excitatory responses to stimulation of different motor and somatosensory cortical areas occurred on the same channels.This suggests that the same small regions of ICC, or even the same individual neurones, are innervated by multiple cortical areas.Consistent with this, we have previously seen terminals anterogradely labelled  Note the example in ai has an early fixed latency spike (highlighted in red) which is immediately followed by a longer period of increased firing.In the examples shown in bi and ci, there is no early fixed latency spike, but a period of increased firing begins around 12-15 ms after the stimulus.Insets show the FRA plots for the channels shown.Group data (aii, bii, cii) showing in cases from each cortical area in which there is an early fixed latency spike.Data are from multiple stimulation sites in multiple animals (see Results text), in which between 2 and 5 different stimulus currents were tested.Spots represent different stimulation/recording site pairs.Note that in general increasing stimulus current results in a decrease in the latency and an increase in the number of channels on which a monosynaptic spike was evident.Group data showing longer latency/longer duration excitatory responses in ICC following stimulation of aiii S1Jaw, biii S1FL, and ciii barrel cortex.Note that increasing stimulus current increases the magnitude and the number of channels on which a later excitatory response is seen.Data are from multiple stimulation sites in multiple animals (see Results text) in which between 2 and 6 stimulus currents were tested.Lines represent means and shading represents SEM.Data greyed out are from channels dorsal (left) and ventral (right) to the ICC.Scale represents the relative depth of recording sites from the dorsal limit of the ICC as defined earlier (see Fig. 1).from different cortical areas, surrounding individual ICC neurones (author's unpublished observations).It was also notable that stimulation of the different subregions of the PFC often evoked responses in the same channels in ICC.Responses were also of similar magnitude and latency suggesting multiple descending cortico-collicular connections from PFC to individual neurons in ICC.
While most of our experiments involved ipsilateral electrical stimulation, we also examined whether ICC neurones would respond to electrical stimulation of contralateral cortical areas.In agreement with our tracing data (Olthof et al., 2019) showing substantial contralateral cortical innervation of ICC, we saw clear excitatory responses to contralateral stimulation which were similar in their nature and timing to those evoked by ipsilateral stimulation.
The ICC is organised tonotopically with more dorsal regions tuned to low frequency sounds and more ventral parts tuned to higher frequencies (Malmierca et al., 1995;Merzenich and Reid, 1974).It was noteworthy, that for all the cortical areas studied, the greatest responsiveness was observed in low frequency tuned areas of ICC.For M1, M2, and S1FL the largest responses were in channels recording units tuned to around 4 kHz, whereas for simulation of S1Jaw the greatest responses to were in channels recording units tuned around to 1-2 kHz.
The aim of this study was to elicit functional evidence for corticocollicular connections from non-auditory cortical areas to the ICC.As such, our analysis does not demonstrate how these connections influence the responses to sounds in the ICC, and thus does not address their physiological significance.However, several functional studies have suggested that the IC could be the recipient of inputs from cortical regions other than AC.Thus, modulation of VC has been demonstrated to influence the BOLD signal in the IC (Gao et al., 2015;Leong et al., 2018), while Groh and colleagues have shown that visual stimuli and saccade-related signals can evoke direct responses in the IC as well as modulate activity evoked by auditory stimuli (Groh et al., 2001;Porter et al., 2007).It has also been reported that self-generated movement modulates the responses of neurons in the IC and that these signals might derive from efference copy of activity in MC (Yang et al., 2020).The responses we observed may provide the substrate for some of these observations and add to the evidence that influences from non-auditory modalities in the cortex operate at the brainstem level.For example, in the case of the projections from the motor and somatosensory cortices, such circuits may mediate mechanisms that distinguish self-generated sounds from external sources, enhance modality selection, or emphasise signals associated with co-activation of multiple modalities.Furthermore, direct projections from the PFC demonstrate that even regions concerned with executive function can influence midbrain auditory processing.

Conclusion
In conclusion, we present direct physiological evidence from neuronal recordings that inputs from multiple non-auditory cortical regions have a direct excitatory influence on neuronal responses in the ICC and so, through activation of local circuits within ICC, have a widespread influence on ICC function.Note that the same channels are activated from all subregions and that stimulation of all subregions evokes responses of similar magnitude of the response is similar in all.Lines show mean response and shading indicates SEM.Data are from 8 to 9 stimulation/recording site pairings in 6 animals.Data greyed out are from channels dorsal (left) and ventral (right) to the ICC.Relative depth of recording sites from the dorsal limit of the ICC as defined earlier (see Fig. 1).

Fig. 1 .
Fig. 1.Example FRA plots from 32-channel electrodes implanted in the IC.Data from the 32 channels are arranged in 4 columns of 8 channels with upper left channel most dorsal (1) and lower right channel (32) most ventral.Note that in example a channels 12-29 (highlighted with red boxes) show a distinct tonotopic pattern of frequency response and would be considered to lie within the ICC.In example b channels 4-21 would be considered to lie within the ICC.Other channels are outside the ICC.c schematic showing the approximate position of the recording electrode (green) in ICC.Note that the active part of the electrode is longer than the ICC such that activity is also recorded from regions dorsal and ventral to the ICC.

Fig. 2 .
Fig. 2. Injection of retrovirus in ICC results in GFP labelling of cells in brainstem auditory nuclei ai-vi a series of coronal sections posterior to anterior showing the site of injection of virus in the ICC; b a sagittal section showing the site of injection of virus in the ICC.This section has DAPI staining to allow visualisation of the hippocampus and other structures.Note that in ai-avi the injection is contained within the IC.In b there is labelling in the visual cortex and retrograde labelling in the brainstem.Retrograde labelling of axons and cell bodies in c lateral lemniscus (LL) and olivary nuclei; d olivary nuclei (coronal); and e olivary nuclei (sagittal). injection.

Fig. 3 .
Fig. 3. Injection of retrovirus in ICC results in GFP labelling in regions of the cerebral cortex.a pyramidal cell in AC showing GFP expression in soma, dendrites, and axon.b large numbers of GFP expressing pyramidal cells in AC layer V. c GFP expressing cells are also found in primary visual cortex, d M1, e somatosensory cortex and f PFC.g a sagittal section shows groups of GFP expressing cells in primary motor cortex, somatosensory hindlimb and trunk regions and parietal association cortex.Note that GFP-expressing cells occur in small clusters.Scale bar 100 µm.

Fig. 5 .
Fig. 5. Optogenetic stimulation of cortex increases firing in ICC. a the effect of two ramped light stimuli (0-5 V, 10 s) applied to the AC on activity on a single channel in ICC.Inset: FRA plots for the channels shown in ai (left) and bi (right)-note the frequency tuned activity characteristic of ICC.bi the effect of two ramped light stimuli (0-5 v, 5 s) applied to M1 on activity on a single channel in ICC.Note that in each case the light stimulus increases the firing rate, that the responses to repeated stimuli are similar.Top line in each part shows the ramped light stimulus.bii an expansion of part of the first light stimulation in bi showing that, in the ICC, spikes with several distinct shapes (highlighted with different colours) are activated by the light stimulus in M1.Asterisks indicate all spikes passing threshold.

Fig. 6 .
Fig. 6.Electrical stimulation sites in M1.Grey shading indicates areas denoted as M1 in the Rat Brain Atlas (Paxinos and Watson, 1998).Filled symbols indicate sites from which measurable (excitatory) responses were evoked in ICC.Open symbols show stimulation sites from which we failed to detect a response at currents up to 8 mA.The different colours denote different animals.Distances are anterior/posterior to Bregma.

Fig. 7 .
Fig. 7. Electrical stimulation of M1 evokes short, fixed, latency spikes in ICC.Example from a single channel in ICC showing that a the spike (highlighted in red and indicated with the red arrowhead) occurs in each of 20 consecutive sweeps; b an expansion of the four sweeps in the box showing that the size and shape of the spike is relatively consistent; c the response in this channel averaged over 100 sweeps.In a, b, and c the large deflection and blue arrowhead indicate the stimulus artefact, the red arrow indicates the early fixed latency spike; d the frequency response area (FRA) plot for this channel.

Fig. 8 .
Fig. 8.The latency of the early fixed latency event in ICC decreases with increasing stimulus current.a An example response in one channel (average of 100 sweeps)showing that increasing current results in a progressive decrease in the latency of the spike.Note also that the size of the averaged event increases with increasing current, reflecting the fact that at higher currents, the event occurs with greater probability.The FRA for his channel is shown in Fig.6d.b group data from 17 M1 stimulation sites in 3 animals in which between 2 and 6 different stimulus currents (0.3-8 mA) were tested.Note that there is a decrease in latency with increasing stimulus current and an increase in the number of channels (spots) on which a spike is seen.

Fig. 11 .
Fig. 11.Electrical stimulation of M1 occasionally evokes inhibitory responses in ICC.Activity on individual channels in the ICC during electrical stimulation of M1.Data are from 100 individual sweeps overlaid.Each sweep is in a different colour.The large deflection at 0 ms is the electrical stimulation artefact.Coloured bars represent periods of excitation (cyan) and inhibition (magenta).a an example where a substantial increase in firing activity with short latency and relatively low jitter is followed by a later period of inhibition: inset is the FRA plot for this channel.b an example in which a short period of inhibition follows a short latency excitation: inset is the FRA plot for this channel.Note that this channel exhibited an unusually high spontaneous firing rate.

Fig. 14 .
Fig. 14.Electrical stimulation sites in S1FL, S1HL and S1Jaw regions.Grey shading indicates areas denoted as S1Jaw, S1FL, and S1HL (darker grey) (Paxinos and Watson, 1998).Filled symbols indicate sites from which measurable (excitatory) responses were evoked in ICC (0.5-8 mA).Open symbols denote sites from which there was no measurable response in ICC at currents up to 8 mA.Different colours represent different animals.Distances are anterior/posterior relative to Bregma.

Fig. 15 .
Fig. 15.Electrical stimulation of somatosensory cortical regions evokes excitatory responses in ICC.Examples showing that an excitatory response is seen in the ICC in 20 consecutive sweeps following stimulation of ai S1Jaw, bi S1FL and ci barrel cortex at 3 mA.Data are from a single channel in each of three different animals.Note the example in ai has an early fixed latency spike (highlighted in red) which is immediately followed by a longer period of increased firing.In the examples shown in bi and ci, there is no early fixed latency spike, but a period of increased firing begins around 12-15 ms after the stimulus.Insets show the FRA plots for the channels shown.Group data (aii, bii, cii) showing in cases from each cortical area in which there is an early fixed latency spike.Data are from multiple stimulation sites in multiple animals (see Results text), in which between 2 and 5 different stimulus currents were tested.Spots represent different stimulation/recording site pairs.Note that in general increasing stimulus current results in a decrease in the latency and an increase in the number of channels on which a monosynaptic spike was evident.Group data showing longer latency/longer duration excitatory responses in ICC following stimulation of aiii S1Jaw, biii S1FL, and ciii barrel cortex.Note that increasing stimulus current increases the magnitude and the number of channels on which a later excitatory response is seen.Data are from multiple stimulation sites in multiple animals (see Results text) in which between 2 and 6 stimulus currents were tested.Lines represent means and shading represents SEM.Data greyed out are from channels dorsal (left) and ventral (right) to the ICC.Scale represents the relative depth of recording sites from the dorsal limit of the ICC as defined earlier (see Fig.1).

Fig. 16 .
Fig. 16.Electrical stimulation of PFC subregions evokes a short, fixed latency spike in ICC.Main figure shows the latency of the early spike in PFC subregions (Cg: dark blue, PrL: orange; IL: green; DP: light blue).Data are from multiple stimulation sites in 4-6 animals.Note that there is no consistent difference in latency between subregions but that the latency decreases with increasing current.Line: median; box: IQ range, whiskers: extreme data points.Inset shows example recordings on one channel in ICC: data are averaged responses to 100 stimuli applied to each PFC subregion.

Fig. 17 .
Fig. 17.Electrical stimulation of all subregions of the prefrontal cortex evokes long latency, long duration excitatory responses in ICC.Excitatory responses (summed PSTH) to stimulation at a 1 mA, b3 mA, c5 mA and d 8 mA.Subregions are coloured in accordance with the inset figure.Note that the same channels are activated from all subregions and that stimulation of all subregions evokes responses of similar magnitude of the response is similar in all.Lines show mean response and shading indicates SEM.Data are from 8 to 9 stimulation/recording site pairings in 6 animals.Data greyed out are from channels dorsal (left) and ventral (right) to the ICC.Relative depth of recording sites from the dorsal limit of the ICC as defined earlier (see Fig.1).