A novel micro-ECoG recording method for recording multisensory neural activity from the parietal to temporal cortices in mice

Device fabrication

The thickness of the film is an important factor of stiffness for proper device insertion into the epidural space between the temporal cortex and skull. Devices with different thicknesses were fabricated to understand the effect of parylene thickness on the self-supporting characteristics. The parylene thicknesses Tp of the devices prepared in this study were 7, 13, and 23 μm. Figure 2a shows photographs of the neural electrode array device with different thicknesses, including optical micrographs of the electrode array. The thin device was curled up owing to strain caused by the thermal expansion coefficient between metal and parylene. As increased thickness of the film results in increased bending stiffness, the thick device was not as curled, but resulted in more robust self-supporting characteristics relative to the thin device. The bending stiffness S of the parylene film without metal can be expressed as.

Fig. 2figure 2

Bending stiffness of the sheet-shaped μECoG device. (a) Photographs of a μECoG sheet with different parylene thicknesses of 7, 13, 23 μm and an optical microscope image of a sheet with arranged neural electrodes (Ti/Au). While the 7 μm thick sheet strongly curls, the 23 μm thick sheet has high self-supporting characteristics. (b) Calculated bending stiffness of the device with and without the metal wiring layer as a function of the parylene thickness. (c) Frequency dependence of impedance of the representative neural electrode. The impedance at 1 kHz was approximately 0.5 MΩ. (d) The distribution of impedances of μECoG electrodes (52/64 channels). The remaining channels were removed due to high impedance (> 5 MΩ)

where Ep and b are the Young’s modulus (Ep = 2.78 GPa) and the width (b = 3.1 mm) of parylene film, respectively. It can be seen that S increases rapidly with increasing the Tp. The actual device contains 64 channel Ti/Au metal wiring layer with a high Young’s modulus, which affects S. The structured-corrected S was theoretically calculated based on a previous report [22]. Figure 2b shows the calculated bending stiffness of the device with and without the metal wiring layer as a function of the parylene thickness. The S of the device with the metal layer is ~ 1.3–1.4 times higher than that of the device without the metal layer. To determine the optimal device design, the thickness and width for parylene, and also the thickness, width, and number of metal wire channels were key factors. Figure 2c shows the frequency dependence of impedance from a representative neural electrode. Impedances at 1 kHz were approximately 0.5 MΩ. The distribution of channel impedance from the electrode used in the recording experiment in an awake mouse (Fig. 4) is shown in Fig. 2d (0.46 ± 0.07 MΩ). All the devices used (Fig. 5) showed similar impedance characteristics.

Fig. 3figure 3

Histological validation of the μECoG sheet position. (a) A schematic of the position of the μECoG sheet placed on the brain surface (upper left). Superimposed microscopic images of bright field and DAPI staining (right and lower left). White arrows indicate the μECoG sheet on the brain surface. The insets show enlargements of the dashed rectangle. Red arrows indicate the tip of the μECoG sheet. Scale bars: 1 mm. Ctx, cortex; Hipp, hippocampus; D, dorsal; M, medial; V, ventral; L, lateral. (b-d) Comparison of cortical structure between the μECoG sheet-inserted hemisphere (ipsi-lateral side) and opposite hemisphere (contralateral side). Ipsi, ipsilateral; Contra, contralateral. (b) Schematic drawing of a coronal view of the mouse brain; black and red dashed rectangles indicate the parietal (c) and temporal cortex (d), respectively. Scale bars: 500 μm

Histological validation of μECoG sheet position

To examine whether the 23 μm flexible μECoG sheet appropriately reaches to the ventral parts of temporal cortex along the curved surface of mouse cerebral cortex, we histologically validated the position of the μECoG sheet by using adhesive transfer film to prevent the samples from separating [23] (Fig. 3a, see methods). Briefly, we prepared sections including brain, skull, and μECoG sheet and took microscopic images under conditions for bright field and DAPI (nuclear staining). The results showed that the μECoG sheet was situated along the surface of cerebral cortex (Fig. 3a, white arrows), and the tip of the μECoG sheet approached the piriform cortex, which is the most ventral part of the temporal cortex (Fig. 3a, red arrows). In addition, comparisons between ipsilateral and contralateral hemispheres showed no obvious damage in either the dorsal or ventral parts of the cerebral cortex that may have resulted from the surgical procedures required for device insertion (Fig. 3b-d).

Fig. 4figure 4

LFP Recordings by the µECoG sheet electrode in an awake mouse. (a) Dorsal view of the µECoG electrode array placed on the parietal cortex. M, medial; L, lateral; A, anterior; P, posterior; Ch., channel of electrode; Sss, superior sagittal sinus. (b) The position of the µECoG sheet placed on the brain surface as revealed by CT scanning. White arrows indicate the µECoG sheet placed on space between the brain and skull. The enlarged image (right) shows the tip of µECoG sheet in the brain (yellow arrow). D, dorsal; V, ventral; M, medial; L, lateral. (c) The 64-channel µECoG electrode array. Numbers are the channel IDs of the electrode. Red rectangles indicate the thirty-two successive electrodes that recorded representative LFPs along the dorsoventral axis of the cortex. D, dorsal; V, ventral; A, anterior; P, posterior. (d) Representative LFPs before the sensory presentation task. Numbers indicate channels of the electrode corresponding to (c) h.i., channel of high impedance (> 5 MΩ). (e) Schematic drawing of the sensory presentation task and recordings in a head-fixed mouse. (f) Average sensory-evoked LFPs. (g) Averaged peak LFP amplitudes in the same channels elicited by different stimuli. BL, baseline. (h) Averaged peak LFP amplitudes in different channels elicited by the same stimulus. (i) The LFP profile calculated from the averaged tactile-evoked LFPs in (f). The central lines and edges of the boxes indicates the median and quartiles, respectively, and whiskers extend to the maximum and minimum. ***P < 0.001. ††† < 0.001 vs. baseline amplitude

Fig. 5figure 5

The μECoG electrode can monitor cortical activities in response to multiple sensory stimuli. (a) Schematic drawing of the sensory presentation task and recordings in a head-fixed mouse. (b) Dorsal view of the μECoG electrode array on the parietal cortex. M, medial; L, lateral; A, anterior; P, posterior; Ch., channel of electrode; Sss, superior sagittal sinus. (c) The 64-channel μECoG electrode array and LFPs at start of the task (representative LFP) or on task (averaged LFP) in a mouse. LFPs corresponding to the tactile, visual, and odor stimuli are indicated as blue, green, and red line, respectively. Shadows indicate SEM. D, dorsal; V, ventral; A, anterior; P, posterior. (d) Averaged peak LFP amplitudes recorded by the channels in which the maximal change was observed for the tactile stimulus. BL, baseline. (e) Averaged peak LFP amplitudes recorded by the channels in which the maximal change was observed for the odor stimulus. There are no channels showing significant differences between odor stimulus-evoked and baseline LFP amplitudes in the #3 mouse. The central lines and edges of the boxes indicate the median and quartiles, respectively, and whiskers extend to the maximum and minimum with outliers excluded. *P > 0.05, ***P < 0.001 vs. each stimulus. †P > 0.05, ††† < 0.001 vs. baseline amplitude

LFP recordings in an awake mouse

First, we tested whether the μECoG electrode arrays, implanted onto the temporal cortex by our surgical methods, can monitor LFPs from a broad cortical area of an awake and head-fixed mouse, while also recording sensory stimulus-evoked potentials. Half of the electrodes on the μECoG sheet were inserted into the space between the temporal cortex and skull, and the remaining electrodes were situated on the parietal part of cortex (Fig. 4a). The location of the μECoG sheet in the temporal cortex was confirmed by a CT imaging after the experiment, and the tip of the sheet reached to the most ventral part of the temporal cortex (Fig. 4b), which is comparable with the histological assessment (Fig. 3a). Histological and CT assessments indicated that our μECoG electrodes were placed on parietal to temporal cortices, including the barrel and piriform cortices. Under this condition, we recorded LFPs with the μECoG electrode arrays (Fig. 4c) along with the dorsoventral axis of the cortex. Representative LFPs recorded by successive thirty-two electrodes are shown in Fig. 4d. The recording condition was reliably maintained for at least 3 h under the acute experimental condition.

Next, we conducted a sensory cue-presentation task to record cortical potentials responding to sensory stimuli (Fig. 4e). In this experiment, we applied a tactile stimulus to the vibrissa in the form of mild air flow because the ECoG electrode arrays easily target the barrel cortex. The visual stimulus was used as a control stimulus as no electrodes were placed on the visual cortex, which is located in a more posterior region than the barrel cortex. Sensory stimuli were pseudorandomly presented to the mouse every ~ 10–20 s. The tactile stimulus induced strong tactile-evoked LFPs in the electrodes placed over the parietal region of the cortex, whereas the visual stimulus did not induce changes in the LFPs recorded from all channels (Fig. 4f). To examine the response characteristics to sensory stimuli on the same channel or between separated channels, we compared the amplitudes of maximal responses of another sensory stimulus and different electrodes to the tactile stimulus. We confirmed that channel 58 showed the maximal change in tactile stimulus-evoked LFP (Fig. 4g, Wilcoxon signed-rank test, Z = 6.009, p < 0.001, for comparison with baseline amplitudes), which was significantly higher than that of visual stimulus-evoked LFPs (Fig. 4g, Wilcoxon signed-rank test, Z = 6.087, p < 0.001). Moreover, we confirmed that the peak amplitudes of tactile stimulus-evoked LFP at channel 58 was significantly higher than that of channel 33 (Fig. 4h, Wilcoxon rank-sum test, Z = 5.112, p < 0.001), the most ventral electrode. To exclude the effects of currents originating far from the µECoG electrodes, we calculated the LFP profiles (see Methods). The location of the largest sensory-evoked potential (Fig. 4f) and the largest LFP profile amplitude overlapped at channels surrounding channel 58, indicating that the surrounding region of channel 58 was highly active when the tactile stimulus was applied (Fig. 4i).

LFP recordings in anesthetized mice

We tested whether the μECoG electrode array covering the parietal to temporal cortices can monitor cortical activity corresponding to multiple sensory stimuli. In this experiment, we conducted a sensory cue-presentation task that added an odor stimulus to the previous task that used tactile and visual stimuli (Fig. 5a). The neural recordings were carried out under urethane anesthesia because sniffing or chowing behaviors were frequently observed in mice not habituated to the novel odors of the task environments. As shown in Fig. 4a, approximately half of the electrode array on the μECoG sheet were inserted into the epidural space between the temporal cortex and skull, while the other electrodes were placed on the parietal cortex (Fig. 5b).

Representative LFP data were recorded by the μECoG electrode arrays from a wide area of cerebral cortex. Before the sensory cue-presentation task, we confirmed the presence of cortical LFPs from parietal to temporal cortices under anesthetized conditions (Fig. 5c, representative LFP). In the sensory cue-presentation task, tactile, odor, and visual stimuli were presented to the mouse pseudorandomly 50 times for each sensory modality (150 times total), every ~ 10–20 s. We observed clear tactile and odor stimulus-evoked LFPs from electrodes located on both parietal (Ch. 58 and 62) and temporal (Ch. 33 and 37) cortices, respectively (Fig. 5c, averaged LFP). Sensory stimulus-evoked LFPs were not observed from the electrode placed on the middle position of the recording areas along the dorsoventral axis (Ch. 15). We next measured the maximal amplitudes of the sensory stimulus-evoked LFP responses to the tactile, odor, and visual stimuli (50 trials each) in four mice. In channels placed on the parietal cortex, the tactile stimulus induced significantly higher evoked LFP amplitudes relative to the baseline amplitudes that were recorded just before the stimulus onset (Fig. 5d, Wilcoxon signed-rank test, Z = 4.609, p < 0.001 for Mouse #1, Z = 3.595, p < 0.001 for Mouse #2, Z = 6.042, p < 0.001 for Mouse #3, Z = 4.581, p < 0.001 for Mouse #4). The tactile stimulus did not result in higher evoked LFP amplitudes than visual and odor stimuli in the same channels in one mouse (Fig. 5d; Friedman test, χ2 = 1.12, p = 0.571, for Mouse #1). In another mouse, the amplitudes evoked by the tactile stimulus were significantly higher than those evoked by the visual stimulus (Fig. 5d; one-way repeated ANOVA, F(2,147) = 4.86, p < 0.01, post hoc Tukey–Kramer test, tactile vs. odor, p = 0.805, tactile vs. visual, p < 0.01, odor vs. visual, p = 0.052, Mouse #2). In the remaining two mice, the tactile stimulus evoked higher amplitudes relative to the other stimuli (Fig. 5d; Friedman test, χ2 = 39.04, p < 0.001, post hoc Tukey–Kramer test, tactile vs. odor, p < 0.001, tactile vs. visual, p < 0.001, odor vs. visual, p = 0.916 for Mouse #3; χ2 = 14.68, p < 0.001, post hoc Tukey–Kramer test, tactile vs. odor, p < 0.01, tactile vs. visual, p < 0.01, odor vs. visual, p = 0.916 for Mouse #4). The amplitudes of odor stimulus-evoked LFP recordings from channels located on the ventral parts of the temporal cortex were significantly higher than the baseline amplitude in three mice, but not in one mouse (Mouse #3) (Fig. 5e; Wilcoxon signed-rank test, Z = 3.464, p < 0.001 for Mouse #1, Z = 5.146, p < 0.001 for Mouse #2; paired Student’s t-test, t(49) = 2.471, p < 0.05 for Mouse #4). In the three mice, the odor-evoked LFP amplitudes tended to be higher than other stimuli-evoked LFP amplitudes (Fig. 5e). Two of the mice showed that the odor stimulus-evoked amplitudes were larger than either the visual or tactile-evoked LFP amplitudes (Fig. 5e; Friedman test, χ2 = 8.32, p < 0.05, post hoc Tukey–Kramer test, tactile vs. odor, p = 0.112, tactile vs. visual, p = 0.703, odor vs. visual, p < 0.05 for Mouse #1; χ2 = 6.28, p < 0.05, post hoc Tukey–Kramer test, tactile vs. odor, p < 0.05, tactile vs. visual, p = 0.514, odor vs. visual, p = 0.341 for Mouse #4). On the other hand, the remaining mouse exhibited odor stimulus-evoked amplitudes that were larger than either the visual or tactile-evoked LFP amplitudes (Fig. 5e; Friedman test, χ2 = 20.32, p < 0.001, post hoc Tukey–Kramer test, tactile vs. odor, p < 0.001, tactile vs. visual, p = 0.978, odor vs. visual, p < 0.001 for Mouse #2).

留言 (0)

沒有登入
gif