Non-invasive, Focused Ultrasound-Facilitated Gene Delivery for Optogenetics

Optogenetics, a widely used technique in neuroscience research, is often limited by its invasive nature of application. Here, we present a noninvasive, ultrasound-based technique to introduce optogenetic channels into the brain by temporarily opening the blood-brain barrier (BBB). We demonstrate the efficiency of the method developed and evaluate the bioactivity of the non-invasively introduced channelrhodopsin channels by performing stimulation in freely behaving mice.

contrast agent. We then allowed the animals to survive for two weeks so that sufficient ChR2 protein channels were expressed in the targeted brain regions. Since one of our goals was to assess the comparison of the non-invasive FUS-based technique to the commonly used direct infusion approach, we implanted an electrode apparatus for optical stimulation and signal recording at the end of the two-week period. A full timeline of our experimental design is shown in Fig. 1c. We demonstrated the versatility of the FUS-facilitated gene delivery by targeting different structures in the brain. As shown in Fig. 1d, the hippocampus (top panel) and cortex (bottom panel) were targeted in different subjects. The BBB-opened regions are depicted as bright pixels in the horizontal and coronal MR images. The reconstructed three-dimensional images present a clear view of the volume of the BBB opening. Brain regions at various depths can be selected by adapting the ultrasound focus.
We compared the efficiency of FUS-facilitated AAV delivery with direct infusion. As shown in Fig. 2a, abundant viral transduction was observed in both the FUS-facilitated delivery and direct infusion. To ensure that the FUS-facilitated gene delivery technique does not cause any physiological damage to the brain, we performed histolopathological analysis on the animals that received unilateral sonication. Based on the Hematoxylin and Eosin (H&E) and Nissl staining results (shown in Fig. 2b and c), no structural damage was observed in either the sonicated side (right column) or the contralateral side (left column). Furthermore, microglia staining ( Fig. 2d) revealed no notable inflammatory response caused by FUS.
We then investigated the bioactivity of the delivered viral vectors and the expressed ChR2 proteins by stimulating the hippocampus in freely behaving mice with the blue light (470 nm) at 30 Hz. Four 2-s pulses were applied and the evoked neuronal activity was recorded as described in a previously published study 11 . The recorded signals were spike-sorted and the total number of spikes were binned into 0.4 s segments. Figure 3a is an example of optical stimulation at the hippocampus that was carried out in a mouse receiving FUS-facilitated viral delivery. Each red segment corresponds to a 2-s stimulation 'on' time, while the black segments represent stimulation 'off ' periods (resting state). As evidenced by the bottom panel, elevated number of spikes were observed during each stimulation pulse indicating the proper functionality of the ChR2 protein channels. For the purpose of comparison, we performed commonly used direct infusion procedures in a separate group of mice. Figure 3b demonstrates recorded spike signals and corresponding number of spikes. Similar to the FUS-facilitated delivery group, an increase in the amplitude and number of spikes per second was observed during the 2-s stimulation on periods (as indicated in blue). A control study where no FUS was applied (AAV injected intravenously) was performed with similar stimulations (Supplementary Fig. 1). No change in signal amplitude nor the number of spikes were observed in these subjects. Examples of individual spike signals during stimulation for the FUS-facilitated delivery group (red) and direct infusion group (blue) are shown in Fig. 3c. Quantitatively, we calculated the neuronal firing rate for each group (n = 4) based on the number of spikes (Fig. 3d). Compared to the baseline level, significant increases of firing rate were observed in both the FUS-facilitated delivery group (p < 0.001) and direct infusion group (p < 0.001).
Additionally, a 10-s continuous stimulation pulse was given and the corresponding spike signal is shown in Fig. 4a. In this case, the total number of spikes were binned into 0.5-s segments. At the onset of the long pulse stimulation, the number of spikes rose to 118 spikes/sec and was then followed by a gradual decrease to the baseline level. This observation can be attributed to the depletion of the ion concentration gradient upon prolonged stimulation. Finally, an enhanced level of c-Fos (marker for neuronal activity) was detected via immunofluorescence staining at the regions that express ChR2 and received optical stimulation. As shown in Fig. 4b, mcherry (red, co-encoded in the viral genome) was used to mark ChR2 proteins, while c-Fos was labelled in green. The high magnification images (Fig. 4c) are an example of the activation of a ChR2-expressing neuron indicating the successfully elicited neuronal activity.

Discussion
We have introduced and implemented a FUS-based, non-invasive viral delivery technique for optogenetic stimulation. We report comparable volumes of transduction and bioactivity between the FUS method and the commonly used direct infusion technique. The difference lies in that direct infusion results in a more confined volume of gene expression while the FUS-based technique produced a more diffuse-transduction. If desired, a single FUS sonication may generate a substantially large axial volume of BBB opening in the coronal plane due to the associated FUS beam shape. The volume of opening, which also determines the volume of viral transduction, can be altered through appropriate selection of the ultrasound parameters. For instance, in Fig. 1d, the focus was placed 2 mm below the skull when targeting hippocampus and almost the entire hippocampus was transduced (indicated by the pink mCherry protein). In contrast, when cortex was the target, the ultrasound focus was placed 0.5 mm below the skull. Given the intravenous nature of the viral administration in the FUS-based method, approximately 20 times more viral vectors were used (~4 × 10 11 GC/animal). Although the absolute number of genome copies was not quantified in each group, the overall transduction volumes were comparable between the two groups, which assumes a 5% of the viral vectors crossing the opened BBB. It is worth noting that the intravenously injected viral vectors circulate systemically. As a result, we found viral transduction in the liver, kidney and heart tissue by directly imaging mcherry expression (data not shown). However, unintended viral transduction can be minimized by carefully selecting appropriate promotors, such as synapsin for neuron-specific transduction 12 .
Our technique might be preferable because of its versatility and entirely non-invasive nature, providing an ideal condition to study neuronal activities in vivo. It might therefore prove particularly useful for researchers who are interested in using long-wavelength light to transcranially stimulate intact brain regions. Furthermore, our technique can be combined with sonogenetics 13 where the expression of mechano-sensitive ion channels is enhanced to achieve fully ultrasound-controlled neuromodulation. The electrode implantation phase can be eliminated to achieve complete non-invasive transcranial stimulation when appropriate optical wavelength is used, such as the red light-driven neural inhibitor reported by Chuong et al. 14 and red-shifted variant of channelrhodopsin developed by Lin et al. 15 . A limitation of this technique is the potential barrier of commercialization and the requirement of basic experience with therapeutic ultrasound technology. The implementation of the protocol requires a separate set of focused ultrasound equipment and careful selection of parameters. The viral vector itself does not need to be altered for the proposed non-invasive technique, though further dosage optimization should be performed to obtain most efficient transduction.

Methods
Microbubble and viral vector. The lipid-shelled microbubbles were manufactured in-house following a previously published protocol 16 . Briefly, the 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC) and polyethylene Glycol 2000 (PEG2000) were mixed at a 9:1 ratio. Two milligrams of the mixture was dissolved in a 2 ml solution consisted of filtered PBS/glycerol (10% volume)/propylene glycol (10% volume) using a sonicator (Model 1510, Branson Ultrasonics, Danbury, CT, USA) and stored in a 5 ml vial. The remainder of the vial was filled with decafluorobutane (C 4 F 10 ) gas. The vial was then activated via mechanical agitation using VialMix TM (Lantheus Medical Imaging, N. Billerica, MA) shaker for a pre-set time of 45 s. The formed microbubbles were analyzed with a Coulter Counter Multisizer (Beckman Coulter Inc., Fullerton, CA). The distribution of these microbubbles is shown in Supplementary Fig. 1 and the mean diameter was measured to be 0.9 μ m. The viral vectors used in this study was AAV9-mcherry-ChR2 with a titer of 2.48 × 10 13 GC/ml (Penn Vector Core, Philadelphia, PA). The vectors were diluted with phosphate buffered saline (PBS) at a 1:6 ratio before administration.

FUS sonication.
A single element FUS transducer (focal length: 60 mm and radius: 30 mm, Imasonic, France) was used for all sonications in this study. The transducer has a center frequency of 1.5 MHz and a − 6 dB focus of 7.5 × 1 × 1 mm 3 as determined by a needle hydrophone (Supplementary Fig. 2). To monitor cavitation events, a pulse-echo transducer (radius: 11.2 mm, focal length: 60 mm, and center frequency: 10 MHz, Olympus NDT, Waltham, MA) was confocally aligned with the FUS transducer. The pulse-echo transducer was driven by a pulser-receiver (Olympus, Waltham, MA), which was connected to a digitizer (Gage Applied technologies, Inc., Lachine, QC, Canada) for data acquisition. The pulse-receiver was operated on a "receive mode" and served as an amplifier during PCD acquisition. The transducer setup was then mounted onto a three-dimensional positioning system (Velmex Inc., Lachine, QC, Canada) for accurate targeting.
Targeting of specific brain regions was achieved by first visualizing the lambda suture through the shaved scalp which was described in greater details elsewhere 17 . Briefly, a small water tank with an ultrasound transparent opening was placed on top of the shaved scalp and a metallic cross was used to locate the lambda suture. C-mode ultrasound scans were then carried out to identify the position of the cross and therefore the lambda suture. The FUS transducer was then moved via the positioner to the following coordinates: AP + 2.6 mm, ML − 2 mm and DV + 2.5 mm for hippocampus targeting; AP + 6.3 mm, ML − 1 mm and DV + 0.5 mm for motor cortex targeting. For each target, a single sonication was performed with the following acoustic parameters: a free field (i.e. in water) peak rarefactional pressure (PRP) of 0.74 MPa, pulse repetition frequency (PRF) 5 Hz, pulse length 10 ms, and a total duration of 120 s. The collected cavitation signal was analyzed using a custom written program (MATLAB R2011a, MathWorks, Inc., Natick, MA). The appearance of ultraharmonics and broadband noise was deemed the signature of cavitation events. Immediately prior to the sonication, a mixture of 100 μ l diluted AAV vectors and 5 μ l microbubbles were injected intravenously via the tail vein. All experimental procedures involving animals were approved by the Columbia University Institutional Animal Care and Use Committee and in accordance to the Office of Laboratory Animal Welfare and the Association for Assessment and Accreditation of Laboratory Care regulations. Surgery and optical stimulation. For mice that received direct virus infusion, the following procedures were carried out. Mice were anesthetized with a mixture of oxygen and 1-3% isoflurane (SurgiVet, Smiths Medical PM, Inc., WI) and placed prone with its head immobilized by a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA). The head was shaved and an incision was made to expose the skull. The hippocampus was located using the following coordinates centered at the bregma: AP − 2.7 mm, ML − 2 mm and DV + 1.5 mm. A 34 G needle (Hamilton, Reno, NV) was used to infuse a total volume of 0.8 μ l AAV vectors at approximately 0.003 μ l/s. Upon completion, the needle was left in place for an additional 5 min to minimize back flow. Mice that received either FUS-facilitated viral delivery (N = 3) or direct infusion (N = 3) were allowed to survive two weeks before the implantation surgery. An optical fiber (Thorlabs Inc. Newton, NJ) and microdrive (Axona Ltd. UK) assembly was implanted to the targeted brain region for optical stimulation and neural activity recording. About 3-4 jeweler's screws were inserted into the skull to support the implants. An additional screw connected with wire was also inserted into the skull which served as a reference electrode. Dental cement was spread across the exposed skull to secure the optical fiber and the microdrive assembly. The mice were allowed one day to recover from the (a) Ten-second long pulses were also applied (red). An increased spike amplitude and number of spikes were observed at the onset of stimulation, which was followed by a gradual decrease to baseline level. (b) Neuronal activity (ChR2 is red) was revealed by labeling c-Fos proteins with fluorescent markers (green). (c) An example of a ChR2-expressing neuron that was activated by optical stimulation. Scale bars represent 20 μ m and 5 μ m in (b,c), respectively. surgery before undergoing optical stimulation. During each session, the optical fiber was connected to a LED light source and was stimulated with four 2-second pulses and one 10-second pulse of 470 nm (blue) light at 30 Hz frequency. The mouse implanted with electrodes was plugged to the electrophysiology setup (Axona) during each stimulation and neuronal activity was recorded as described previously 11 . The recorded signals were analyzed and visualized with a custom-written program (MATLAB R2011a, MathWorks, Inc., Natick, MA). To quantify the number of spikes, a threshold was set at three standard deviations above the mean of the resting state signals. Peaks above the threshold were deemed to be a spike and was binned into either 0.4 s or 0.5 s segments. The firing rate was quantified as the total number of spikes per second.

MRI. The
Histology and imaging. One hour post optical stimulations (for maximal c-Fos expression), mice were sacrificed and transcardially perfused with 30 mL PBS and followed by 60 mL 4% paraformaldehyde. The skull was removed upon sacrifice and the brain was soaked in paraformaldehyde overnight. The brain tissue was then cryo-protected with 30% sucrose solution for 48 hours and frozen on dry ice. Brain samples were sectioned coronally at 40 μ m and mounted with ProLong ® Gold mounting solution with DAPI (ThermoFisher, Waltham, MA) to examine the distribution of ChR2 (mcherry). Neuronal activation was visualized via immunochemistry staining for c-Fos proteins. Brain sections were incubated with primary anti-c-Fos antibody (ab209794, Abcam, Cambridge, MA) at 1:100 dilution, and secondary antibody (R37116, ThermoFisher, Waltham, MA) at 1:250 dilution. All fluorescence images were taken with a Nikon confocal microscope (Nikon Instruments Inc., Melville, NY). A separate group of mice (N = 3) was used to evaluate the safety aspects of the FUS sonication. One day post the sonication, these mice were sacrificed as previously described and soaked in paraformaldehyde for two days. The brains were then paraffin fixed and serially sectioned at 7 μ m. The structural integrity of the brain was examined via Hematoxylin and Eosin (H&E) staining and Nissl staining. Potential inflammatory response was evaluated by staining for microglia. The paraffin sections first underwent antigen retrieval process followed by incubation in primary anti-Iba-1 antibody (1:200, ab5076, Abcam, Cambridge, MA) and secondary donkey anti-goat Alexa 488 (1:200, A-11055, Invitrogen, Carlsbad, CA).