Stay or Go? Neuronal activity in medial frontal cortex during a voluntary tactile preference task in head-fixed mice

Abstract

The decision to move is influenced by sensory, attentional, and motivational cues. One such cue is the quality of the tactile input, with noxious or unpleasant sensations causing an animal to move away from the cue. Processing of painful and unpleasant sensation in the cortex involves multiple brain regions, although the specific role of the brain areas involved in voluntary, rather than reflexive movement away from unpleasant stimuli is not well understood. Here, we focused on the medial subdivision of secondary motor cortex, which is proposed to link sensory and contextual cues to motor action, and tested its role in controlling voluntary movement in the context of an aversive tactile cue. We designed a novel, 3D-printed tactile platform consisting of innocuous (grid) and mildly noxious (spiked) surfaces (50:50% of total area), which enabled monitoring neuronal activity in the medial frontal cortex by two-photon imaging during a sensory preference task in head-fixed mice. We found that freely moving mice spent significantly less time on a spiked-surface, and that this preference was eliminated by administration of a local anesthetic. At the neuronal level, individual neurons were differentially modulated specific to the tactile surface encountered. At the population level, the neuronal activity was analyzed in relation to the events where mice chose to “stop-on” or “go-from” a specific tactile surface and when they “switched” surfaces without stopping. Notably, each of these three scenarios showed population activity that differed significantly between the grid and spiked tactile surfaces. Collectively, these data provide evidence that tactile quality is encoded within medial frontal cortex. The task pioneered in this study provides a valuable tool to better evaluate mouse models of nociception and pain, using a voluntary task that allows simultaneous recording of preference and choice.

Graphical Abstract

1. INTRODUCTION

Voluntary locomotion is influenced by the integration of internal and external cues, including attention, motivation, and environmental sensation [1–3]. Painful stimuli can cause immediate, reflexive withdrawal, while noxious or unpleasant stimuli can cause an animal to consciously move away from the cue [4, 5]. The cortical processing of noxious and unpleasant sensations involves multiple brain regions, including the somatosensory cortex, anterior cingulate, and thalamus, but integration with centers of the brain responsible for initiation and termination of locomotion is not well understood [6, 7].

The medial subdivision of secondary motor cortex (M2) in rodents is referred to by multiple names, including medial precentral cortex, second frontal area and medial agranular cortex (AGm) [8, 9]. Medial M2 can be considered a part of Medial Frontal Cortex (MFC), which includes anterior cingulate, prelimbic and infralimbic structures [10]. As an association cortex, medial M2 receives multimodal inputs, including visual, sensory, auditory, somatomotor and thalamic, as well as bidirectional inputs with medial prefrontal structures such as anterior cingulate [9–12]. Medial M2 has multiple proposed functions, including decision-making, action planning, directing attention, motor learning and sensory perception [13, 14]. In particular, it has been argued that M2 is important for linking specific conditions, especially sensory cues, to motor action [9]. For example, in a free-choice paradigm, neuronal signals appear in the rodent AGm prior to behavioural action and earlier than other brain regions, suggesting its importance in action initiation [15]. Furthermore, M2 was found to be important for goal-directed responses towards a sensorimotor behavior [13]. Disconnection of medial M2 from posterior parietal lobe leads to neglect in multiple modalities, suggesting an importance in directed attention [3]. Although studies have investigated neuronal modulation in medial M2 by sensory cues, it is unknown how this area responds to sensory tactile input with inherent emotional valence, such as a noxious pain-producing stimuli.

The majority of published reports evaluating the effects of noxious stimuli have used non-voluntary tasks typically combined with reflexive nocifensive behaviors, such as paw withdrawal threshold in response to von Frey filaments [16, 17]. Although this approach allows for repeated, consistent presentations, it is inherently artificial, preventing the animal’s natural response to the stimuli, and as such has limited clinical translatability [5, 16]. Further, variations in handling or restraint techniques can alter behavioral responses to noxious stimuli by inducing stress in the animals. Therefore, behavioral evaluations of pain-like behaviors in rodents that incorporate voluntary choice to engage with noxious stimuli offer a greater clinical translatability [5].

Here, we have designed a voluntary tactile preference task using a mildly noxious stimulus, which can be used in head-fixed mice that move freely on a floating air platform, while monitoring neuronal activity by two-photon microscopy. This approach allows evaluation of the mouse’s decision to engage with the stimuli, either by walking over it or resting on it, and the relative neuronal response during this decision. We found that mice demonstrated an avoidance of the mildly noxious stimuli and lost this avoidance upon subcutaneous administration of a local anesthetic. We have used individual and population-level neuronal analysis to show the activity of medial M2 neurons during initiation of movement and the decision to stop in the context of noxious versus innocuous sensory stimuli.

2. METHODS

2.1. Animals

All experiments involving mice were approved by the University of Iowa Institutional Animal Care and Use Committee and were carried out in strict accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals. Every effort was made to minimize the number of mice used and their suffering. 6 Male and female mice expressing a genetic Ca²⁺ fluorescent sensor GCaMP6s (C57BL/6J-Tg(Thy1-GCaMP6s)GP4.3Dkim/J, Jackson Lab, Stock# 024275) were injected at 6–7 months of age (181–203 days) with viral vector encoding for mCherry (AAV6-CAG-mCherry-WPRE, Vector Biolabs), which was used for image stabilization, with bilateral stereotaxic injection at 1.9 mm anterior-posterior (AP), 1.0 mm medial-lateral (ML), angled 45 degrees posterior, 45 degree medial, with injection made throughout a 1mm needle track. Naïve, non-head-fixed C57BL/6J male and female mice (5–7 months old) were also used for a subset of experiments.

2.2. Surgical Procedures

Six weeks following stereotaxic injection, mice had cranial windows placed as previously described [18]. Briefly, mice were anesthetized with inhaled isoflurane (~1.8–2% during surgery). Meloxicam analgesia (1 mg/kg) was administered prior to surgery and for 48 hours after. A craniotomy was made using a dental drill with 0.5 mm bit, from −0.5 mm bregma to 2.5 mm Bregma (AP) and +/− 1.5 mm ML, including the known injection site. Care was taken at the rostral end over the sagittal sinus near the rostral rhinal vein to avoid damage to this vasculature. The bone at this region is extremely thick and requires removal in order to image medial structures of frontal cortex. In 5 out of 6 mice in the study, the dura was removed over one side of the cranial window, in part due to adherence of dura to the skull secondary to the viral injection. Clarity at the time of imaging was superior on the side of removal, so imaging was taken from this side. The 6^th animal had dura intact on both sides. A small droplet of Kwik-Sil (biocompatible silicon, WPI) was placed over the cortex to provide thermal buffer and exert firm pressure on this uneven region of cortex. A square 3 mm coverslip was then installed and sealed with cyanoacrylate. Two skull screws (#000–120) were placed over occipital cortex, with care taken not to penetrate brain. Stainless steel headplates compatible with the magnetic Levelt clamp (Neurotar Ltd, Finland) were then installed using cyanoacrylate and dental cement. The mouse was allowed to recover for 3 weeks following surgery prior to imaging. One week following surgery mice began to undergo daily handling using a standardized protocol until animals became acclimated and comfortable with being manipulated. At the time of imaging, mice could be guided and attached magnetically to a head fix system over the Mobile HomeCage® floating platform with an associated carbon cage (Neurotar).

2.3. Generation of grid- and spiked-pattern insert for distinct sensory stimulation

For imaging during a voluntary tactile preference (noxious versus innocuous) task, a custom 3D-printed tactile insert was designed (TinkerCAD) and printed using PLA plastic (Prusa i3 MK3S 3D printer) (Figs. 1 and 2). A plate consisting of a half-spiked (noxious), half-grid (innocuous) pattern was used. The spiked tactile pattern consisted of cones (4 mm x 4 mm x 2.5 mm height) with a 4 mm distance between each point. The grid consisted of 1 mm x 1 mm stripes in a 4 mm x 4 mm grid. The grid was smooth to the touch, but provided some degree of grip, which was preferred over a completely smooth surface in preliminary trials in naïve non-head-fixed mice. The points and the grids were printed to the same height such that variation in height was not a factor in preference during head-fixed behavior. The 3D model required for printing is available and included in supplemental files (Supplemental Fig.S1). After initial validation using freely moving non-head-fixed male and female mice to determine relative preference (Fig.2), this tactile insert was installed into a Neurotar mobile air-floating platform (Fig.1D) and used for all in-vivo Ca²⁺ imaging experiments in head-fixed mice (Figs. 3–5).

Figure 1. A schematic of the experimental setup for two-photon live Ca²⁺ imaging in head-fixed, freely-moving mice.

(A) A cranial window was placed over the mid sagittal sinus in order to image medial frontal cortex. (B) Approximate imaging locations in relation to bregma and the sagittal sinus are shown for all tested mice by the blue boxes. In 5/6 mice in the study, the dura was removed for clarity (see methods). (C) Neurons were selected from a time- projected average. (D) Head-fixed mice freely moving over an air-floating platform (Neurotar) with insert made of half-grid and half-spiked surface. Mice were imaged while performing spontaneous bouts of running and resting. Neuronal activity was synchronized with mouse locomotion by using the Mobile HomeCage Locomotion Tracking Software (Neurotar). (E) Examples of Ca²⁺ traces (GCaMP6s; ΔF/F₀) from individual selected neurons shown in (C) aligned with behavior. Inset shows an example series of time-averaged GCaMP6s images of an individual cell over a 50 s time interval.

Figure 2. The noxious characteristics of the spiked surface compared to the grid surface as revealed by behavioral testing in mice.

(A) Image of the platform used in behavioral experiments and imaging. (B) Time spent on each half of the platform in non-head-fixed mice during a 10-minute session (n = 18, males and females). (C) Time spent on each half of the platform in lidocaine-treated, non- head-fixed mice during a 10-minute session (n = 18, males and females). The data shown in (B-C) were analyzed using a Wilcoxon signed-rank test (*p<0.05) and are presented as median + 95%CI. (D) Percent duration spent on spiked surface compared with grid surface during imaging (n = 6, males and females). Time spent on the border did not count towards either zone. *p<0.05, paired t-test.

Figure 3. M2 neuronal activity associated with the “Go-from” events.

(A) “Go-from” events were defined as those for which mice started to move after a rest period on either grid or spiked surface. (B) An example of recordings from a single neuron, for which ΔF/F₀ averaged over all “Go-from” events for the same cell showed a significant increase during the 1 s post-initiation of movement (post-GO) period as compared to the 1 s pre-initiation (Pre-GO) period (p<0.05, t- test). 44% of active neurons showed modulation by initiation of activity. (C) An example of averaged ΔF/F₀ from a single neuron that showed a significant difference in activity during the post-initiation of movement period, depending on the tactile surface on which the mouse started (differential modulation). Less than 1% of neurons showed differential modulation based on tactile surface. (D-H) Principal component analysis (PCA) was used to evaluate population response during “Go-From” events. The heat map (D) shows changes in activity of individual neurons during the “Go-From” events that started from a grid surface (upper panel) or spiked surface (lower panel). Color coded average ΔF/F₀ was obtained by normalizing ΔF/F₀ for each neuron and averaging over “Go-From Grid” versus “Go-From Spiked” events, respectively. Neurons were then sorted by principal component 2 score for visualization. (E) Principal component activity patterns of the first three principal components. (F) Percent of variance explained by each principal component. (G) PCA scores for the top three principal components showed differences in “Go- from spiked” vs. “Go-from grid” events. *p < 0.05, t-test, corrected by false discovery rate. (H) Change in PC space for each event type, with color representing advancing time and circles indicating time = 0 s.

Figure 5. M2 neuronal activity associated with the switch between the spiked and grid surfaces.

(A) “Switch” events were defined as those for which mice switched tactile surfaces during continuous running (B) An example of recordings from a single neuron, for which ΔF/F₀ averaged over all “Switch” events for the same cell showed a significant change when comparing Pre-Switch (from −1 to 0 s) activity with Post-Switch (from 0 to 1 s) activity (p<0.05, t-test). 3.6% of active neurons showed significant modulation by “Switch” events, regardless of the switch direction. (C) An example of ΔF/F₀ recorded from a single neuron showing a significant difference in average ΔF/F₀ when comparing “Switch to grid” versus “Switch to spiked”. Approximately 2.5% of active neurons showed significant differential modulation based on tactile surface. (D-H) Principal component analysis (PCA) was used to evaluate population response during “Switch” events. The heat map (D) shows changes in activity of individual neurons during the “Switch” events that started from a grid surface (upper panel) or spiked surface (lower panel). Color coded average ΔF/F₀ was obtained by normalizing ΔF/F₀ for each neuron and averaging over “Switch to spiked vs. “Switch to grid” events. Neurons were then sorted by principal component 1 score for visualization. (E) Principal component activity patterns of the first three principal components. (F) Percent of variance explained by each principal component. (G) PCA scores for the top three principal components, showing differences in “Switch to spiked” vs. “Switch to grid” events in PC1. *p < 0.05, t-test, corrected by false discovery rate. (H) Change in PC space for each event type, with color representing advancing time and circles indicating time = 0 s.

2.4. Behavioral preference

Behavioral preference was determined using naïve (non-head-fixed) male and female mice using the same carbon cage associated with the Mobile HomeCage platform used for imaging. After acclimatization for 5 minutes, animals were tested for their place preference on the platform for 10 minutes, during which time spent on the grid- and spiked-pattern sides of the platform was recorded. Time spent was defined as the time period during which at least one hindpaw was on the tactile cue. Thus, periods during which the mouse had one hind-paw on each cue accrued time on both the grid and spiked surface. (Fig. 2A). After lidocaine was administered subcutaneously to both hindpaws (2%, 10 μL), animals were allowed to rest briefly, followed by a ten-minute test period on the platform in the same carbon cage (Neurotar). For preference in head-fixed animals, time spent was estimated using the neurotar locomotion tracking software, which estimates the location of the mouse’s head using tracking of the position of the plate. Time on spiked or grid surface was defined as the percent of time the mouse’s head was at least 20mm onto that surface, over total time.

2.5. Two-photon imaging

Mice with implanted cranial windows were imaged while head-fixed, but freely walking and resting on the tactile air-floating plate (Figs.1B, 2A), for 30–50 min using an upright Olympus multiphoton FVMPE-RS equipped with Mai Tai DeepSee tunable laser set at 935 nm, using a 25x water-immersion objective (N.A.=1.05; Olympus).

Neuronal activity was monitored by using the genetic Ca²⁺ sensitive fluorescent indicator GCaMP6s. [19]. The target imaging region (500 μm x 500 μm) was centered 500 μm from the midline (250 μm-750 μm), confirmed by visualization and direct measurement from the midline of the mid-sagittal sinus. Anterior-posterior was centered at 1.65 mm Bregma (1.4 mm to 1.9 mm), comprising of medial M2 (Fig.1). Cells were imaged at a target depth of 140 μm from the pial surface (layer II/III), ranging from 100– 180 μm over the imaging window because of the anterior curvature of the brain. Images were captured at 512×512 pixels using the resonant scanner at 30 Hz. A wavelength of 935nm was used to excite both GCaMP6s to measure neuronal activity and mCherry for imaging stabilization. TTL pulses sent from both the Olympus microscope and the Mobile homecage Locomotion Tracking Software (Neurotar) was used to precisely align imaging and motion tracking data. The images were then motion corrected using NoRMCorre [20], using mCherry fluorescence (red filter, λ_Ex=575–645nm) to motion correct GCaMP6s fluorescence (green filter, λ_Ex=495nm-540nm).

2.6. Data analysis

Regions of Interest (ROIs) were hand selected from the z-projected average of the time series using the following criteria: distinct nucleus, soma distinct from surrounding neuropil, overlap between adjacent neurons <1/4 cell body. Each time series was reviewed to distinguish ambiguous neurons, and two investigators independently reviewed ROIs for consensus. A total of 913 ROIs from 6 mice were identified, then mean florescence intensity of each ROI was extracted from the image files. The time-dependent baseline Ca²⁺ fluorescence, F₀, was calculated as the minimum value of the smoothed florescence during a time window (as described in [21], with the following parameters: Ƭ₀ = 0.2 s, Ƭ₁ = 0.75 s and Ƭ₂ = 10 s) and ΔF/F₀ values were smoothed with exponentially weighted moving average filtering [21].

For detection of individually modulated neurons, mean ΔF/F₀ was compared one second prior (PRE) to one second following (POST) the event of interest (GO-from, STOP-on, and SWITCH, as described in Fig. 3A, 4A, and 5A, respectively). For each neuron, PRE versus POST mean ΔF/F₀ was compared via t-test. For differential modulation, mean ΔF/F₀ for one second POST was compared between events that differed by tactile surface. A false discovery rate (FDR) correction was then applied to correct for multiple t-tests per animal, based on the number of neurons tested per animal. Neurons that met the criteria of this analysis were considered significantly modulated by the event of interest.

Figure 4. M2 neuronal activity associated with the “Stop-on” events.

(A) “Stop-On” events were defined as those for which mice stopped moving for >1 s on either the grid or spiked surface. (B) An example of recording from a single neuron, for which ΔF/F₀ averaged over all “Stop-On” events for the same cell, showed a significant difference when comparing the Pre-STOP (from −1 to 0 s) with the Post-STOP (from 0 to 1 s) values (p<0.05, t- test). Approximately 8.4% of active neurons showed significant modulation based on termination of activity. (C) An example of a single neuron ΔF/F₀, showing significant differential modulation for “Stop-on Grid” vs. Stop-ON Spiked” events. Less than 1% of active neurons showed significant differential modulation dependent on the tactile surface. (D-H) Principal component analysis (PCA) was used to evaluate the population response during “Stop-On” events. The heat map (D) shows changes in activity of individual neurons during the “Stop-On” events that started from a grid surface (upper panel) or spiked surface (lower panel). Color coded average ΔF/F₀ was obtained by normalizing ΔF/F₀ for each neuron and averaging over “Stop-ON Grid vs. “Stop-ON Spiked” events, respectively. Neurons were then sorted by principal component 2 score for visualization. (E) Principal component activity patterns of the first three principal components. (F) Percent of variance explained by each principal component. (G) PCA scores for the top three principal components, showing differences in “Stop-ON spiked” vs. “Stop-ON grid” events in PC2. *p < 0.05, t-test, corrected by false discovery rate. (H) Change in PC space for each event type, with color representing advancing time and circles indicating time = 0 s.

For population analysis, we evaluated time-based neuronal patterns of activity using principal component analysis (PCA), which we and others have widely used to identify patterns of neuronal activity in an unbiased, data-driven manner [13, 22–25] using custom-written MATLAB routines [26, 27]. The PCA was constructed from z-transformed peri-event time changes in smoothed (10 sample gaussian filter) GCaMP6s fluorescence signal, ΔF over the ±2 s of behavioral events. All ROIs from 6 mice with recordings were included in the PCA. The same principal components (PCs) were projected onto grid- surface and spiked-surface trials. For each comparison group (grid-surface vs. spiked- surface) PC scores for the first three principal components were compared via t-test, with FDR correction [26, 27].

3. RESULTS

3.1. Behavioral preference

When presented with a choice between the two types of surfaces (Fig.2A), freely-moving non-head-fixed mice showed a preference for the grid surface over the spiked surface, suggesting that the spiked surface is aversive (p < 0.05; Fig. 2B). This aversion was prevented by prior administration of a local anesthetic to the hindpaws of the mice, consistent with the noxious quality of the spiked surface (p = 0.73; Fig. 2C). We also found that during Ca²⁺ imaging experiments, mice spent less time on the spiked surface compared with the grid surface (Fig. 2D, p < 0.05).

3.2. M2 neuronal activity associated with the “Go-From” events

To analyze neuronal activity when mice chose to move after resting on the grid or spiked regions, we defined “Go-from” events as a period of rest >1 s on either grid or spiked surface, followed by initiation of movement (Fig. 3A). First, we analyzed whether individual neurons showed a change in neuronal activity (mean ΔF/F₀) based on initiation of movement. We found that 44% of active neurons showed a statistically significant change (difference in PRE versus POST mean ΔF/F₀) one second before versus one second after initiation of movement (example Fig. 3B). During Go-from events <1% of individual neurons showed differences depending on whether movement was initiated on spiked versus grid surface (Fig. 3C).

Next, we examined neuronal activity at the population level surrounding periods of movement initiation. We used principal component analysis (PCA) to evaluate population activity using an unbiased, data-driven approach (Fig. 3D–H). This dimensionality reduction technique identifies a series of orthogonal basis functions that capture the maximal variance across multivariate data and has been extensively used to analyze activity of neuronal populations [26, 27]. We found 70% of the variation in population firing could be explained by the first three principal components, shown in Fig. 3E, all of which showed significant differences in PC score depending on whether the mouse initiated movement from the grid or spiked surface. Principal component 1 (PC1) represents gradual changes over the time period examined and explained 37% of the variance in changes in ΔF (Fig. 3E–F, p = 0.008). PC2, which represents a peak change in ΔF at time 0 (either peak increase or peak decrease), explained 21% of the variance in changes in ΔF (Fig. 3E–F). This component also showed a significant difference (p=0.003) between trials starting from grid versus spiked surface (Fig. 3G). PC3 is similar to PC1 (increasing after the mouse begins to move), but describes a more complicated pattern, including peak changes in firing 1–2 seconds before the event. These differences can also be visualized by graphing changes in the principal components over the time period examined (Fig. 3H), showing distinct population trajectories in PC space.

3.3. M2 neuronal activity associated with the “Stop-On” events

Next, we analyzed neuronal activity in the medial M2 region during “Stop-On” events. For these events, mice were running for at least one second and made the decision to stop on a location within the grid or spiked surface (Fig. 4A). At the single neuron level, 8.4% of active neurons showed a significant difference in their activity (mean ΔF/F₀) when comparing the pre-STOP with post-STOP status. An example of a neuron modulated by “Stop-On” events is shown in Figure 4B. Less than 1% of individual neurons were differentially modulated depending on whether the mouse stopped on spiked versus grid surface (Fig. 4C).

At the population level, 41% of the variance in ΔF for “Stop-on” events could be explained by the first PC, representing a gradual change in neuronal activity over the time period during which the mouse stopped (Fig. 4E–F). An additional 19% of the variance could be explained by PC2, representing a peak change in ΔF at time 0. Interestingly, PC2 showed a significant difference (p<0.00001) depending on whether the mouse had stopped on a spiked or grid surface (Fig. 4G).

3.4. M2 neuronal activity associated with tactile switching

For “Go-From” and “Stop-On” events, changes in running activity play a significant role in modulation of neuronal activity. We therefore also compared events during which the mouse was continuously running and switched from one tactile surface to another (Fig. 5A). Importantly, because mice tended to spontaneously run for long periods of time, or rest for long periods of time, we had significantly more events to evaluate when utilizing this switch task, which improved power to detect significantly modulated neurons. At the single neuron level, we found that 3.6% of neurons showed a change in activity during a PRE-to-POST tactile switch while running from either grid to spiked surface or spiked to grid surface. An example of a neuron modulated during a PRE-to-POST switch is shown in Fig. 5B. Interestingly, when comparing the surface onto which the mouse switched, 2.5% of neurons showed significant differences in individual activity depending on whether the mouse was switching from a grid to spiked surface or a spiked to grid surface. An example of a neuron that was differentially modulated depending on the tactile quality is shown in Fig. 5C.

At the population level, the dominant pattern that explains the majority of variance (35%) differs compared to prior analysis for the “Go-from” and “Stop-on” events. Here, PC1 represented a change that began approximately at the time of a switch and increased (or decreased) from that time point. There was a significant difference (p<0.00001) in PC1 score depending on whether the switch was from spiked or grid surface (Fig. 5E–G).

4. DISCUSSION

The decision to stop or move toward a noxious stimulus requires the synthesis of sensation, perception, and directed attention. The medial subdivision of M2 has been implicated in decision making that involves sensory context [8, 9], although whether this region is also involved in generating behavioral response to a noxious pain-producing stimulus, remained largely unknown. The findings from our work suggest that the medial subdivision of M2 could contribute to sensory integration that guides behavior. Indeed, by utilizing voluntary, spontaneous engagement with noxious stimuli, we showed that medial M2 contains subsets of neurons that are individually modulated depending on the sensory stimulus quality. Further, we found that at the population level, the patterns of neuronal activity differed significantly depending on the direction of the mouse transition between the noxious (spiked) and innocuous (grid) surfaces. The involvement of the medial M2 region in the response to a noxious tactile stimulus is consistent with the prominent connectivity of this area with somatosensory cortex and thalamus, both of which are key components of pain processing in the brain [6, 7]. Notably, within the medial frontal cortex, the M2 region is adjacent and bidirectionally connected with the anterior cingulate cortex (ACC) [9, 28] which is another brain region that plays important roles in acute and chronic pain processing [6, 7, 29]. Thus, further research is warranted to elucidate the functions of the M2 region and its neuronal populations in the context of noxious sensory stimuli and to establish precise details of the connectivity of this region with the ACC, specific thalamic nuclei and somatosensory cortices S1 and S2.

Our study also demonstrates the feasibility and potential for using 3D printed platforms in behavioral tasks and during in vivo, awake imaging. These platforms can be customized to produce variable levels of noxious and innocuous tactile stimuli, with specific spatial configuration (e.g., 50:50% area split; Fig. 2A). Such tactile platforms would allow examination of preference, rather than relying on a reflexive response, while simultaneously imaging neuronal activity in specific brain regions. This approach has the potential for use in various preclinical rodent models of pain, such as models of postsurgical pain and neuropathic pain, which are characterized by prominent mechanical hypersensitivity [17, 30–32]. This behavioral task also has the benefit of requiring minimal researcher interaction with the animals, reducing the animals’ stress, and does not require multiple days of training. Habituated mice can be trained to run on the platform in two days, and no training is required for non-head fixed mice. Additionally, advances in technologies such as 3D printing allows for the standardization of tools used between labs and the ability to custom fit tools used for different cage sizes.

There are potential limitations of this study. First, only two types of sensory stimuli were employed and a local anesthetic was used as a control to define the relative noxious qualities of the spiked and grid surfaces.” (Fig. 2A–C). Although local anesthetics are commonly used to block pain sensation both in clinic for treating patients, and in a laboratory setting for animal pain research [33–36], these drugs inhibit all somatosensory modalities, making it possible that other qualities beyond innocuous and noxious experience may contribute to differences in neuronal activity that we observed. Future research can utilize a similar paradigm, either by printing different patterns, or through the use of other noxious sensory modalities, such as heat. Additionally, more specific pharmacological or genetic targeting of nociception could be used to either enhance or desensitize specific sensory modalities. Combined with a lick spout, our paradigm allows for the potential to reward walking on noxious stimuli, which would enable enticement onto stronger aversive stimuli. Second, although our data analysis allows us to identify a population of neurons that are significantly modulated by tactile cues, the exact percentage of neurons identified is influenced by multiple comparison correction (using FDR), the number of interactions and the strength of the stimulus. Thus, it is possible that in the context of stronger noxious stimuli, an increased number of neurons with significant modulation would be found.

In conclusion, medial M2 represents an under-studied brain region of potential significance in understanding the cognitive processing of noxious stimuli and pain. While most recent work has focused on the anterior cingulate cortex, thalamus, and primary sensory cortex [6], medial frontal cortex has important proposed roles that are relevant for the cognitive processing of noxious pain-producing stimuli, namely, directed attention and decision making in regards to motor movements onto and away from the stimuli. The novel behavioral task presented here has broad applicability beyond simple in vivo, awake imaging, and can be adapted to multiple behavioral paradigms to study various aspects of sensation and nociception.

HIGHLIGHTS:

• We developed a tactile stimulus task utilizing preference, not reflexive responses.

• The tactile stimulus task has broad applicability for behavior and in vivo imaging.

• A subset of medial M2 neurons are modulated by changes in sensory stimuli.