Wireless closed-loop optogenetics across the entire dorsoventral spinal cord in mice

a, Main steps of the microfabrication of the micro-LED array: Step 1: Schematic illustration of the micro-LED array microfabrication process. A Ti/Au/Ti film is sputtered on a polyimide substrate and subsequently patterned by photolithography and wet etching. Next, the device interconnects are covered by a 2^nd layer of polyimide, and the whole polyimide stack is patterned by photolithography and reactive ion etching (RIE). The preparation is covered with a thin layer of PDMS. The silicone superstrate is patterned to the device layout by photolitography and RIE, exposing the micro-LED integration sites. Then, micro-LEDs are precisely interfaced with the device interconnects. Finally, the micro-LED array is encapsulated with PDMS and released from the silicon carrier. Step 2: Photograph of the 4-inch wafer following the micro-LED array microfabrication process. Step 3: Colorized scanning electron micrograph (45° tilted view) of the micro-LED array surface, highlitghling the fine patterning of the PDMS superstrate and integration of the bare dies. Schematic cross-section of the device, showing three interconnects encapsulated in PDMS, top left inset. Step 4: Photograph of the optoelectronic device laminated on a fingertip. The array hosts 2 independent micro-LED channels, connected via serpentine interconnects that accommodate physiological motion. Step 5: Stress-strain curves for bulk PI, a fully functional implant and bulk PDMS measured under a displacement rate of 100μm/s. In the panel referring to the fully functional implant, the y-axis on the right (blue) also reports the relative resistance measured at the current input of 1 mA. The grey box highlights the strain range under in vivo conditions. b, Downconversion of light to desired wavelength. Step 1: Schematic illustration of the downconversion process using a phosphor-silicone matrix. Blue photons are converted to the desired wavelength (1), transmitted through the matrix (2) or back-scattered (3). Step 2: Photographs of the optoelectronic devices with the introduction of the phosphor-silicone matrix. The phosphor peak emission wavelengths are indicated below the corresponding photographs. Step 3: Optical characterization following downconversion of blue light. Emission spectra of the micro-LED arrays depending on their respective phosphor-silicone matrix implementation. Note the leakage of blue light at 𝜆=470 nm (left). Total optical power produced by one micro-LED channel covered with phosphor-silicone matrices with emission peaks at 𝜆=590 nm or 𝜆=620 nm. The respective optical power of the leaked blue light is depicted at 𝜆=470 nm. For reference, the optical power of bare blue LEDs is plotted. Step 4: Characterisation of different wavelength implants in Thy1-ChR2 and vGlut 2 ChrimsonR mice. Only 470 nm wavelength light results in muscle responses in Thy1-ChR2 mice. Only 590 nm and 650 nm wavelengths result in muscle responses in vGlut 2 ChrimsonR mice.

Source data

Step 1: System overview: the implants (9 stainless steel wires for ground and 4 differential EMG recording channels, micro-LEDs connected by 3 copper wires) are connected by a 16 pin Omnetics connector to the PCB assembly of the wireless headstage platform. Step 2: Illustration of the antenna design as integrated on the PCB of the head-stage. Step 3: Head-stage wireless link reliability assessment in a typical laboratory/office environment based on received signal strength indication (RSSI) measurements versus line-of-sight distance. A sufficient link is maintained for over 10 m of distance. Step 4: Verification of the LED pulsed current driver performance: LED current can be controlled in 1 mA steps and is monitored on-chip for each issued pulse to verify micro-LED-implant condition and ensure experiment validity. Step 5: Tablet user interface for pulse(-train) stimulation and configuration ranges for all parameters. Step 6: Measurement of closed-loop performance: a EMG signal, simulated as a short sine-wave burst, the triggering of the software algorithm (SW trig), and the issued LED current pulse (I LED) have been acquired by an oscilloscope to measure the delay between the beginning of EMG activity and the closed-loop response. A delay of 11.3 ms is caused by the signal processing, which includes characteristics preventing spiking caused by noise (low-pass). A delay of 0.9 ms is caused by the construction of signal to drive the LED activation. Step 7: Tablet user interface for closed-loop experiments, consisting of an experiment parameter configuration interface and a live preview of all four acquired EMG traces.

Source data

Step 1: Stepwise surgical insertion of the micro-LED array into the epidural space. Micro-LED array overlaying a spinal cord surrogate, bottom right inset. Step 2: Post-mortem evaluation of foreign body responses within the dorsal horn of spinal segments located below the micro-LED array. Coronal spinal cord sections of mice implanted for 1, 4 or 6 weeks were stained against Iba1 and GFAP proteins, and compared with the spinal cord of non-implanted mice. Histogram plots report the fluorescent staining intensity quantified within the dorsal horns. The photograph shows the staining from a window within the dorsal horn, as shown in the scheme (n=6 healthy mice, n=4 for each timepoint post-implantation one-way ANOVA, Iba1 p=0.64, GFAP p=0.36, mean±s.e.m.). Step 3: Schematic illustrating the post-mortem evaluation of spinal cord circularity. The bar graph reports the mean circularity (n=6 mice, one-way ANOVA, p=0.95, mean±s.e.m.). Step 4: Photographs show examples of subcutaneous wires taken in mice that were implanted for 4 weeks.

Source data

a, Kinematic analysis of locomotor function reveals there are no gait deficits after micro-LED implantation. Step 1: The absence of influences on locomotor performance was evaluated using longitudinal recordings of whole-body kinematics du ring walking along a corridor. Markers are attached to the skin overlaying anatomical landmarks to record whole-body kinematics with the optoelectronic Vicon system. Step 2: Sequence of leg movements during walking reconstructed from the 3D coordinates of the markers. Step 3: The kinematics is morphed onto the anatomical model of the mouse and an envelope is added over the bony structure to obtain a realistic reconstruction of leg movements during walking. The resulting chronophotography-like sequences of locomotor movements illustrate the absence of differences between gait patterns recorded before and after the implantation. Step 4: A total of 78 parameters were calculated from kinematic recordings (Supplementary Table 1). Step 5: A principal component (PC) analysis was applied to the calculated parameters, and represented in the new space created by PC1 and PC2. Each dot represents the mean and SEM values of many mice (n> 5 mice) with averaged data from many gait cycles (n> 10 pre mouse). The bar plot reports the mean values of scores on PC1, which captures the more pronounced differences between gait cycles recorded at different time-points (n=6-9 mice/timepoint, one-way ANOVA, p=0.16, mean±s.e.m.). Step 6: Color-coded representation of factor loadings with the highest level of correlation of PC1. Step 7: Functional clusters of highly correlated parameters were extracted from PC1 loadings representing gait parameters that describe differences between groups the most. Bar plots reporting mean values of calculated gait parameters with high correlations with PC1. This highly-sensitive statistical analysis failed to detect measurable changes in gait patterns following the implantation of the micro-LED array (n=6-9 mice, one-way ANOVAs, Stride length p=0.42; Step height p=0.46; Amplitude of ankle elevation p=0.13; Amplitude of Ankle Speed p=0.13, mean±s.e.m.). b, Implantation of the mico-LED had no effect on exploratory behaviours or fine locomotor function. Step 1: The potential impact of the micro-LED array on exploratory behaviours was measured in the open field paradigm. Body trajectories measured during 20 min of home cage monitoring are shown before and after implantation of micro-LED array. The bar plot reports the quantification of distance covered during 20 min period of observation (n=6, two-sided unpaired t-test, p=0.58, mean±s.e.m.). Step 2: Mice were tested on the accelerating rotarod paradigm. The bar plot reports quantification of the time to fall off (n=12 mice per timepoint, n=6 mice at week 4, one-way ANOVA, p=0.44, mean±s.e.m.). Step 3: Mice walked along a horizontal ladder with unevenly spaced rungs. The bar plot reports quantification of the percentage of foot falls (n=8 mice, one-way ANOVA, p=0.08, mean±s.e.m.). Step 4: Mice climbed up a 1-meter long vertical rope. The bar plot reports quantification of the time to complete the climb (n=5 mice, one-way ANOVA, p=0.60, mean±s.e.m.).

Source data

Step 1: Timeline detailing key experiments. Step 2: Post-moterm verification of ChR2 expression. Photographs show the robust expession of ChR2 in the spinal cord and dorsal root ganglia of Thy1-ChR2 mice. ChR2 is expressed in proprioceptive neurons of the dorsal root ganglia, as illustrated with the overlap between ChR2-eYFP with parvalbumin mRNA (HCR). Step 3: Experimental set up to evaluate muscle responses and leg kinematics when delivering single pulses of photostimulation while the mouse is suspended in the air in a robotic body weight support system. Leg movements are reconstructed following one pulse of photostimulation delivered over the rostral versus caudal lumbar spinal cord. Step 4: Associated muscle responses in the iliopsoas and gastrocnemius medialis muscles are shown following one pulse of photostimulation delivered over the rostral versus caudal lumbar spinal cord. Step 5: The bar plot reports the mean range of motion at the knee and hip following photostimulation over the rostral and caudal lumbar spinal cord (n=10 mice, two-sided unpaired t-test, p=0.2445 and p=0.3153 respectively, mean±s.e.m.). Note that the iliopsoas muscle is activated by photostimulation of rostral lumbar segments, as expected based on the anatomical position of the motoneurons innervating this muscle within these spinal segments and pre-motor interneurons, as well as by photostimulation of caudal lumbar segments through the activation of primary afferent fibres embedded in the rostral posterior roots that travel along the spinal cord, in the vicinity of the caudal LED channel. The different location at which photostimulation depolarizes afferent fibers is reflected in the increase in the latency of muscle responses with photostimulation over caudal versus rostral lumbar segments and the clear singluar wave. Step 6: The plots report the area under the curve (AUC) of rectified muscle responses of the illiopsoas and grastrocnemius following photostimulation of the rostral versus caudal lumbar spinal cord with increasing intensity (n=5 mice, one-way ANOVA, , p

Source data