A Large-Scale Interface for Optogenetic Stimulation and Recording in Nonhuman Primates.

NeuroResource

A Large-Scale Interface for Optogenetic Stimulation and Recording in Nonhuman Primates Highlights d

Introducing an interface for stable optogenetic stimulation and recording in NHP

d

Convection-enhanced viral infusion yields expression across a large cortical area

d

d

mECoG array obtains large-scale recordings with minimum optical attenuation Demonstration of spatiotemporal responses across cortical areas to optical stimuli

Authors Azadeh Yazdan-Shahmorad, Camilo Diaz-Botia, Timothy L. Hanson, Viktor Kharazia, Peter Ledochowitsch, Michel M. Maharbiz, Philip N. Sabes

Correspondence [email protected] (A.Y.-S.), [email protected] (P.N.S.)

In Brief Optogenetics provides powerful capabilities for linking circuit function to the sophisticated behaviors that are commonly studied in nonhuman primates (NHPs). Yazdan-Shahmorad et al. introduce a novel optogenetic interface for NHPs that enables stable, large-scale stimulation and recording across cortical areas.

Yazdan-Shahmorad et al., 2016, Neuron 89, 1–13 March 2, 2016 ª2016 Elsevier Inc. http://dx.doi.org/10.1016/j.neuron.2016.01.013

Please cite this article in press as: Yazdan-Shahmorad et al., A Large-Scale Interface for Optogenetic Stimulation and Recording in Nonhuman Primates, Neuron (2016), http://dx.doi.org/10.1016/j.neuron.2016.01.013

Neuron

NeuroResource A Large-Scale Interface for Optogenetic Stimulation and Recording in Nonhuman Primates Azadeh Yazdan-Shahmorad,1,2,* Camilo Diaz-Botia,2,3 Timothy L. Hanson,1,2 Viktor Kharazia,1 Peter Ledochowitsch,2,3 Michel M. Maharbiz,2,3,4 and Philip N. Sabes1,2,3,* 1Department

of Physiology and Center for Integrative Neuroscience, UCSF, San Francisco, CA 94143, USA Berkeley-UCSF Center for Neural Engineering and Prostheses, San Francisco, CA 94143, USA 3UC Berkeley-UCSF Graduate Program in Bioengineering, Berkeley, CA 94720, USA 4Department of Electrical Engineering and Computer Science, University of California, Berkeley, Berkeley, CA 94720, USA *Correspondence: [email protected] (A.Y.-S.), [email protected] (P.N.S.) http://dx.doi.org/10.1016/j.neuron.2016.01.013 2UC

SUMMARY

While optogenetics offers great potential for linking brain function and behavior in nonhuman primates, taking full advantage of that potential will require stable access for optical stimulation and concurrent monitoring of neural activity. Here we present a practical, stable interface for stimulation and recording of large-scale cortical circuits. To obtain optogenetic expression across a broad region, here spanning primary somatosensory (S1) and motor (M1) cortices, we used convection-enhanced delivery of the viral vector, with online guidance from MRI. To record neural activity across this region, we used a custom micro-electrocorticographic (mECoG) array designed to minimally attenuate optical stimuli. Lastly, we demonstrated the use of this interface to measure spatiotemporal responses to optical stimulation across M1 and S1. This interface offers a powerful tool for studying circuit dynamics and connectivity across cortical areas, for long-term studies of neuromodulation and targeted cortical plasticity, and for linking these to behavior.

INTRODUCTION Optogenetics is a powerful tool for relating brain function to behavior, as it enables cell-type-specific manipulation of neurons with millisecond temporal precision (Boyden et al., 2005; Yizhar et al., 2011a) and artifact-free neural recordings. Such capabilities could be particularly powerful for linking large-scale circuit function to the sophisticated behaviors that are commonly studied in nonhuman primates (NHPs). While the use of optogenetics in NHPs has grown rapidly in recent years (Han et al., 2009, 2011; Diester et al., 2011; Cavanaugh et al., 2012; Gerits et al., 2012; Han, 2012; Jazayeri et al., 2012; Ohayon et al., 2013; Ruiz et al., 2013; Dai et al., 2014; Afraz et al., 2015; Nassi et al., 2015), addressing this goal requires the ability to perform large-scale manipulation and recording. Systems to achieve this have become widely used in other an-

imal models, in particular mice (Chan et al., 2010; Bernstein and Boyden, 2011; Anikeeva et al., 2012; Portugues et al., 2013; Wu et al., 2013; Richner et al., 2014), but they have not yet been reported for the NHPs. The key challenges for designing a large-scale optogenetic interface for NHPs are obtaining high levels of optogenetic expression across regions of cortex that can span square centimeters and performing combined optical stimulation and neurophysiological recordings across these same large areas stably over long periods of time. Here we present a system for cortical optogenetics in NHPs that addresses these challenges. The first key challenge described above is to efficiently obtain reliable and consistent transduction levels across a large cortical area. Viral delivery remains the best option for genetically intractable species (Han et al., 2009; Diester et al., 2011; Han, 2012; Ruiz et al., 2013). To cover a large region of the NHP brain, conventional viral injection techniques require multiple small-volume injections. This is a time-consuming approach, and it relies on diffusion of the viral vector from the injection site, which leads to highly variable density of viral vector over the region. Instead, we employed a convection-enhanced delivery (CED) technique developed for drug delivery and gene therapy (Bobo et al., 1994; Lieberman et al., 1995; Lonser et al., 1998b; Szerlip et al., 2007; Fiandaca et al., 2008; Kells et al., 2009). Here, convection refers to the transport of a vector through the interstitial cerebrospinal fluid through a combination of diffusion and advection (bulk fluid flow). In CED, relatively large volumes (50 ml per site) are infused with sufficient pressure to drive advection and hence achieve a wider and more even distribution of the vector. The second key challenge is to perform daily recordings from large regions of cortex while maintaining optical access for stimulation. Most current systems for combined stimulation and recording rely on the use of ‘‘optrodes,’’ penetrating electrodes combined with optical fibers, which restrict the number of simultaneous stimulation and recording sites (Gerits et al., 2012; Jazayeri et al., 2012; Ozden et al., 2013). While optrodes can be moved daily, repeated penetration of relatively large-diameter optrodes will damage tissue and degrade local circuits (Gerits et al., 2012). To allow for stable stimulation at many sites—albeit with less spatial resolution than can be achieved with an optrode—we used a transparent silicone artificial dura (Bonhoeffer et al., 1995; Arieli et al., 2002; Chen et al., 2005; Ruiz et al., 2013) in Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 1


A

E

C

M A B

F

D

P L

2 mm G

2 minutes

6 minutes

8 minutes

11 minutes 1 cm

15 minutes

17 minutes

19 minutes

21 minutes

Figure 1. CED Infusion of Viral Vector (A and B) Top (A) and side (B) views of MRI-compatible infusion cylinder with fixed grid position, used for monkey G. See Figures S1D and S1E for modified infusion cylinder used for monkey J. (C) MR image of the infusion chamber and the saline-filled cannula grid. (D) Photo of the reflux-resistant injection cannula tip. (E and F) MRI surface rendering of the cortical surface below the cylinder for monkeys G (E) and J (F). The S1 infusion locations are shown in blue, M1 locations in red. (G) Spread of the viral vector in coronal sections of monkey G for one injection site in S1 (shown with an arrow in E).

combination with micro-electrocorticographic (mECoG) arrays. The artificial dura preserves optical access to the brain following infusion, allowing the user to track opsin expression via epifluorescence imaging and to target light delivery widely across the region of expression (Ruiz et al., 2013). The mECoG array allows for large-scale cortical recording with minimal optical attenuation (Park et al., 2014; Richner et al., 2014; Ledochowitsch et al., 2015). In this paper, we show how the three existing technologies described above—CED of viral vectors, an artificial dura, and mECoG recordings—can be combined into a practical and powerful large-scale interface for stimulation and recording. We then demonstrate the potential of this system with two examples that examine circuit dynamics and large-scale connectivity in primate primary somatosensory (S1) and motor (M1) cortices. 2 Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc.

RESULTS Large-Scale Optogenetics in NHP Cortex Using CED cortical infusion, we obtained optogenetic expression across large areas of S1 and M1 cortices. Injections were verified with real-time MRI images, and expression was initially verified with surface epifluorescence imaging. MRI-Guided CED We performed large-scale infusions of AAV5-CamKIIa-C1V1EYFP viral vector into S1 and M1 of two rhesus macaque monkeys using CED. These infusions, and the required surgical procedures, were performed during a single session with the animal under anesthesia, using techniques developed by Bankiewicz and colleagues (Krauze et al., 2005; Kells et al., 2009). We first implanted an MR (magnetic resonance)-compatible cylinder with a cannula grid (Figures 1A–1C and S1A–S1C, available

A

B

Monkey G

A

P

Monkey J

MR based estimate of Distribution Volume, Vd (mm3)


250 200 150 100 50 0

Slope=2.7 0

10

20

30

40

50

60

Infusion Volume, Vi (mm3) 250 200 150 100

Slope=3.6

50 0

0

10

20

30

40

50

60

Infusion Volume, Vi (mm3)

Figure 2. MR-Based Estimates of Virus Distribution Volume (A) MR volume reconstruction of the spread of viral vector during CED infusion. Brain is shown in light gray; S1 and M1 Vd are shown in blue and red, respectively. (B) Plots of MR-based estimate of Vd versus Vi over time for each infusion site.

online). Guided by presurgical MR images, a 25 mm craniotomy and durotomy were made over the central sulcus, spanning M1 and S1, and the cylinder was secured over the craniotomy (see Experimental Procedures for more details). Viral infusions were then performed in the MR scanner, allowing precise targeting and online monitoring of the infusions. For monkey G a fixed cannula grid allowed access to both M1 and S1 (Figures 1A and 1B); for monkey J the cylinder was modified so that the cannula grid could rotate to access a larger fraction of the cylinder (Figure S1D and S1E and Data S1, injection_cylinder_outside_J.ipt and injection_cylinder_inside.ipt). Injections were made using a refluxresistant cannula (Figure 1D) (Krauze et al., 2005) inserted approximately 2 mm below the pia mater. The grid was filled with saline and visualized with a T2 image, and the top of the grid was marked with a vitamin E capsule. For each target site, the best grid hole and the target depth were determined by inspection of both T1 and T2 images. We infused the virus into four sites in monkey G (two sites in S1 and two sites in M1; Figure 1E) and five sites in monkey J (three sites in S1 and two sites in M1; Figure 1F) in the left hemisphere. Coinfusion of a gadolinium-based MR contrast agent allowed us to monitor the spread of the virus via real-time MR images and to plan subsequent injection sites. Due to time constraints, the M1 infusions in monkey G were performed outside the MR scanner, precluding real-time MR monitoring. Figure 1G shows flash T1weighted images taken over a 21 min infusion of 50 ml of viral vector in S1 of monkey G (site indicated with arrow in Figure 1E). The vector spread to an approximately 5 mm diameter region around the infusion site, including both gray and white matter (see also Figure S2). Images for all infusions are shown in Figure 2. Figure 2B plots the distribution volume (Vd) estimated from the MR images as a function of the running total infusion volume (Vi). These plots show a linear relationship between Vd and Vi, suggesting a constant rate of convection from the infusion sites, with a slope in range of 3. We estimate a total coverage of

317 mm3 in S1 of monkey G and 233 and 433 mm3 in M1 and S1, respectively, of monkey J. MR-guided injections also allow the user to determine when injections have missed their target. During one M1 injection in monkey J, real-time imaging showed that the infusate was not spreading as expected within cortex, so we stopped the infusion early. A post hoc 3D analysis of the MR images showed that the vector had spread along the surface of the brain (Figure S3), indicating that the cannula was not placed at a sufficient depth. A lack of expression around this site was confirmed with surface epifluorescence imaging and histological analysis, as described below (note the same vector solution was used for all infusions in that session). Artificial Dura for Maintenance of Optical Access Following infusion, we replaced the MR-compatible cylinder with a titanium chamber and then placed an artificial dura within the chamber (Figure 3). Chambers were custom designed for each animal, with the base matching the curvature of the skull (Figure 3A), and 3D printed with sintered titanium (i.materialize). Artificial dura was produced and implanted following the techniques of Ruiz et al. (2013). The artificial dura was custom molded to fit inside the cylinder, with a base flange extending beyond the cylinder and placed under the margin of the dura mater to slow dura regrowth (Figures 3B–3E and S4A). The artificial dura, when left in place and cleaned regularly (see Experimental Procedures), allows for optical access to the cortical surface for up to several months without further intervention (Figure 4A). As reported previously (Arieli et al., 2002; Chen et al., 2005; Ruiz et al., 2013), we eventually see growth of a neomembrane beneath the artificial dura, which becomes increasingly opaque over time (Figure S4B). Once the neomembrane was fully opaque and covered the craniotomy, it was removed. The interval between these procedures varied from 1 to 4 months. However, on one occasion for monkey G, Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 3


Figure 3. Cylinder and Artificial Dura (A) Design of the titanium cylinder was based on coronal and sagittal curvatures of the skull. Curvature measurements of monkey G are shown here. (B) Photograph of the artificial dura. (C) Schematic of the implanted artificial dura with flange placed between the dura mater and the arachnoid. (D) Photograph of the implanted titanium cylinder used with monkey G. (E) Photograph of cylinder with artificial dura. The artificial dura does not look transparent because it is viewed at an oblique angle.

opaque tissue growth was seen within 10 days of neomembrane removal; accelerated neomembrane growth appears to have been caused by the chronic presence of the mECoG arrays that occurred during a pilot study for testing the chronic implantation of these arrays (Yazdan-Shahmorad et al., 2015). More generally, growth appears to be accelerated by mechanical interaction with the tissue; for example, by daily implanta-

A

Following Infusion

1 month 2 months post infusion post infusion

tion of the mECoG arrays or by allowing the artificial dura to rotate during cleaning. Epifluorescence Imaging We monitored optogenetic expression by surface epifluorescence imaging of the fluorescent reporter (EYFP) through the artificial dura (Diester et al., 2011; Ruiz et al., 2013). The images reveal large areas of expression across S1 and M1 for both

3 months post infusion

B

flourescence

thresholded

Monkey G

M A

P L

Monkey J

Figure 4. Surface Imaging (A) Images of the cortical surface taken over a time period of 3 months for both monkeys. The images on the left were taken shortly after the infusion, before initial placement of the artificial dura. The other images were taken with the artificial dura in place. For monkey G, a neomembrane removal was performed between months 1 and 2 (see Figure S4C). The surface of the cortex remains healthy and clear following infusion and chronic artificial dura placement. (B) Left panels: epifluorescence images were obtained approximately 3 months postinfusion (see Experimental Procedures). Right panels: epifluorescence images were thresholded (21% of full saturation; value selected by eye) in order to estimate surface area of expression (see text) and to facilitate comparison with the MR-based estimates of the spread of viral vector (dashed lines; cf. Figure 2). White dots indicate injection sites.

4 Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc.


A

Left

Right

D

F

B

50 µm E

C

5 mm

5 mm

Figure 5. Histological Analysis (A) Baseline coronal MR image. (B) Spread of contrast agent after the infusion for the same MR coronal slice as in (A). (C) A coronal tissue section from approximately the same site as in (A) and (B); peroxidase staining reflects expression of the EYFP reporter (see Experimental Procedures). (D) Example higher-magnification image showing EYFP expression in individual cells. (E) A second example, as in (D). (F) Good alignment is observed between the area of EYFP expression measured with surface epifluorescence (dark green areas) or with histological staining (light green lines). These include the region of vector spread estimated from MR images (white line); white dots indicate injection sites.

monkeys at about 3 months postinjection (Figure 4B). From these images, we estimated expression across at least 130 and 140 mm2 of the cortical surface in monkeys G and J, respectively. The expression patterns are generally consistent with both the sites of injection and the MR-estimated spread of viral vector (Figure 4B), but there are exceptions for both monkeys. No local expression was observed for the more lateral injection site in M1 for monkey G. The reason for this failure is not known, since the M1 infusions for monkey G were not performed in the MRI, but it is likely the injection failed due to cannula placement that was either too shallow or too deep. Similarly, as discussed above, for one M1 injection in monkey J, the cannula was too shallow to deliver vector to cortex (Figure S3). These failed injections highlight the importance of using real-time MR images during infusion. Histological Analysis Monkey G was sacrificed 27 months postinfusion, and a histological analysis of the cortical tissue was performed (see Experimental Procedures). Using immunohistochemistry, we observed large regions of EYFP-reporter expression in both S1 and M1, covering approximately 2 cm2 of the cortical surface and extending beyond the cylinder (Figure 5). The numerous cell bodies with prominent apical dendrites (Figures 5D and 5E) indicate that expression is predominately in pyramidal cells, as expected with the use of the CamKIIa promoter (Nassi et al., 2015), although we cannot exclude expression in other cell types. We compared the extent of EYFP-reporter expression with both the region of epifluorescence and the MR-based region of vector spread (Figure 5F). The two measures of EYFP expression yielded highly overlapping results. Both of these extend beyond the MR estimate of vector spread, presumably

due to the diffusion of viral particles beyond the region of advection. The failure of the more-lateral injection in M1 was also confirmed by the histology. We next analyzed the tissue to evaluate the safety of the CED procedure. Using immunohistochemistry, we observed three marked gaps in EFYP expression, at locations matching the three successful infusion sites (see example in Figure S5B). Nissl stain at the center of the gap also shows a dense and highly localized patch of gliosis approximately 100 mm in diameter (Figure S5C) that was traced through to the white matter, providing further evidence that these correspond to the infusion sites. These gaps in reporter expression are consistent with the presence of a postinfusion lesion spanning roughly the size of the cannula, as seen previously with standard diffusion-based viral delivery (Jazayeri et al., 2012). We next examined regions of high EYFP expression outside the cannula track for tissue damage that could have been caused by either fluid pressure during the CED infusion or by high post-CED concentration of viral particles. We found no obvious pathology when comparing such regions with sites more distant from the injections that had low or no EYFP expression (see examples in Figure S5D). While the CED does not appear to cause cortical-tissue damage beyond the cannula track, the extended use of artificial dura does. Cortical sections revealed thinning and flattening of cortex across the entire area of the craniotomy (Figures 5C, S5B, and S5E). Importantly, this thinning did not correlate with density of viral EYFP expression; thinning also occurred in areas without EYFP (not shown), and no thinning was observed in adjacent cortical sulci that have high levels of EYFP expression but were not exposed to the artificial dura (Figures 5C and S5B). We conclude that thinning is due to the artificial dura and/or to repeated dura resections (10 in the 27 months postinfusion). Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 5


A

B 400µm perforations recording electrodes 600 µm

D

Monkey G

M1 S1

10 µV 50 ms

E

Monkey J

M1

S1

C

Figure 6. Electrophysiological Recording of Light-Evoked Activity (A) Microphotographs of the mECoG array. The array has 96 electrodes (40 mm diameter), as well as perforations for fiber optic and penetrating electrode insertion (shown in the right panel; not used in this study). (B) Placement of a mECoG array in the cylinder of monkey G. (C) Placement of two mECoG arrays in the cylinder of monkey J. (D and E) mECoG recordings from monkey G (D) and monkey J (E) during pulsed optical stimulation. Recording traces were from the closest electrode to the site of stimulation for examples of M1 (red) and S1 (green) stimulations shown in the upper left panels. Shaded areas around the traces represent SE across 25 trials. The blue squares on the traces show the timing of stimulation. The pseudocolor maps on the left show the spatial distribution of the high-gamma energy of evoked responses across the array(s) for the corresponding M1 stimulation. (F) The full set of stimulus-triggered waveforms, superimposed on the mean waveform, for the four sample pairs of stimulation and recording sites shown in (D) and (E).

witsch et al., 2011b). Each array covered 92 mm2 (7.7 3 12 mm), so that one or two arrays were needed to cover most of Monkey J the viral-expressing region. Long inteMonkey G F grated cables (2 cm) enabled easy placement into the chamber. During each recording session, we removed the artificial dura and placed one or two arrays in the chamber, spanning S1 and M1 (Figures 10 µV 10 µV 6B, 6C, S6A, and S6B). We then positioned 10 ms 10 ms the fiber optic on top of the array using our custom-made stimulation setup (YazdanShahmorad et al., 2015) (Figure S6C). With this setup, we observed reliable light-evoked activity in Additionally, gross examination of the postmortem brain surface showed tissue malformation at locations beneath the flanges of both areas from both monkeys, about 3 months postinfusion the cylinder (Figure S5E), likely due to tissue proliferation at the (94 days for monkey G and 77 days for monkey J), which was boundary between the intact dura mater and the artificial dura. the earliest we tested. Figures 6D and 6E show example stimBoth cortical thinning and tissue malformations have been re- ulus-evoked activity from both S1 and M1 in each monkey, in ported previously, 4 months after initial implantation of artificial response to 1 ms pulses of 20–22 mW, 488 nm laser light, applied at 10 Hz. The waveforms are high-gamma-band dura (Chen et al., 2002). (60–200 Hz) filtered, which appears to reflect underlying multiunit neural activity (Crone et al., 2006; Ray et al., 2008; YazLarge-Scale Recording during Optogenetic Stimulation We performed combined optical stimulation and recording ex- dan-Shahmorad et al., 2013). The mECoG array allows us to periments on both monkeys. Recordings were made using readily map the spatial distribution of these stimulus-triggered custom 96-channel mECoG arrays (fabricated in house; now responses across the cortical surface; examples of the distribucommercially available through Cortera Neurotechnologies, tion of high-gamma-band power across the array after M1 stimInc.) (Figures 6A–6C). The arrays consist of lithographically ulation are shown in the pseudocolor maps in the lower left patterned platinum-gold-platinum features encapsulated in Pary- panels of Figures 6D and 6E. The stimulus-triggered responses are highly reliable, as illuslene-C (Ledochowitsch et al., 2011a, 2011b). The arrays were largely transparent; the design kept more than 90% of the surface trated by the examples in Figure 6F. Such reliability was area of the array free of metal features, and Parylene-C is more observed for all of the experiments we performed over a than 90% transparent for the relevant wavelengths (Ledocho- 25 month period for monkey G and over a 1 month period for 5 µV 50 ms



Constant

3Hz

10Hz

25Hz

50Hz

A

50µV 0.5 s

B

50µV 100 ms

C

50 ms

20 ms

10 ms 43 22 1

Gamma (30−60Hz)

11 5

Vresp 2/ Vbase 2

167 ms

High gamma (60−200Hz)

D

1

2 mm

Figure 7. Spatiotemporal Analysis of Responses to Different Stimulation Parameters in M1 (A) Filtered evoked activity (60–200 Hz) during 1 s train of pulsed light stimulation. Black line shows the average across 50 trials. Stimulation pulses are shown in blue (filled). (B and C) Filtered evoked activity (60–200 Hz) during 1 pulse of stimulation for 50 trials for the corresponding stimulation parameters shown in (A). Black line shows the average across trials: 50 trials for constant stimuli, 150 trials for sinusoidal stimuli. Stimulation light pulses are shown in blue (filled). (B) shows temporal responses for all pulses, and (C) shows only responses to the first pulse in a 1 s train. (D) Distribution of high-gamma and gamma activity across the mECoG array averaged across 50 trials. The pseudocolor axis represents the power of the response in the high gamma (60–200 Hz) and gamma (30–60 Hz) bands, divided by the power in the same band during the baseline period (500 ms) before the stimulation train. The stimulation site is indicated as a black arrow, and the recording site shown in (A) is indicated as a black dot in upper left panel.

monkey J: when a stimulation-evoked response was seen, it was present for all simulation events, and the responses all had the same general shape. Example Applications for Studying Large-Scale Cortical Circuits We provide two examples of how this interface can be used to study large-scale circuit dynamics and connectivity in the NHP. Example 1: Spatiotemporal Response Depends on Stimulus Waveform, Despite Fixed Total Energy In the first example application, we used our interface to map the spatiotemporal response within a cortical area to parametric changes in the stimulus. We stimulated focally in M1 while recording from an array that spanned S1 and M1 (see Figures 6B and 6D). As shown in Figure 7, we stimulated with 1 s of constant-power illumination (left column) and 1 s of rectified sinusoidal illumination patterns, with frequencies between 3 and 50 Hz (remaining columns), all in a single session. With fixed peak po-

wer across stimuli, the sinusoidal stimuli all had the same total energy, which was factor of p smaller than for the constant stimulus. Figures 7A and 7B show the stimulus-evoked responses recorded at a single mECoG electrode near the stimulation site and band-pass filtered in the high-gamma band (60–200 Hz). While responses were observed for all stimulation patterns, they were less variable across repetitions for the high-frequency stimuli than for the constant and low-frequency stimuli (compare mean, black line, to data spread, gray, in Figures 7A and 7B). There are two reasons for this difference. First, comparison of response to the first pulse of each train (Figure 7C) shows that the steeper onset slopes of the higher-frequency stimuli drive a larger and more repeatable response than shallow slopes of the low-frequency stimuli. Second, during the relatively longlasting depolarization (Yizhar et al., 2011b) in low-frequency trials, firing of local units becomes desynchronized, leading to a smaller and more variable high-gamma response. Both of these effects are seen in the constant stimulus (Figure 7B, left plot), Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 7


Figure 8. Network Connectivity Observed in the Response to Optical Stimulation

A

(A) Light-evoked responses, filtered in the highgamma band (60–200 Hz) for two sites in S1, one near the site of stimulation (red) and located 3 mm away (green), as well as one site located in M1. The responses are averaged across 50 trials, and shaded areas represent SE. The stimulation pulse shape is shown in blue. (B) Pseudocolor maps show the spatiotemporal distribution of high-gamma power following stimulation averaged across 50 trials. The stimulation and recordings sites (same colors as in A) are shown in the first panel. Black line marks the central sulcus.

15µV 3 ms

B

0 − 3ms

S1 (stim)

3 − 6ms

S1 (other)

M1

6 − 9ms

9 − 12ms

Stim

12 − 15ms

v2/s

Example 2: Stimulating in S1 Evokes a Sequential Activity Pattern in S1 and M1 In a second example, we demonstrate the potential for studying interarea circuit dynamics. For this example, we stimulated 15 − 18ms 18 − 21ms 21 − 24ms 24 − 27ms 27 − 30ms focally in S1, while again recording from an array spanning S1 and M1. We analyzed the stimulus-evoked response by the spatial and temporal patterns of its high-gamma-band power (60– 30 − 33ms 33 − 36ms 36 − 39ms 39 − 42ms 200 Hz). Stimulation in S1 drives a multi156 phasic, high-gamma response at the site 0 of stimulation (Figure 8A, red curve), as well as at distal locations in both S1 (green −156 curve) and M1 (black curve). The timing of 4mm the S1 and M1 responses is quite different. The response at the S1 site folwhere the sharp onset leads to a large, low-variance initial net lows largely the same temporal profile as at the stimulation site, with no appreciable delay and only a slight decrement in response, which then desynchronizes over the next 20–30 ms. The relationship between stimulus shape and response also magnitude. This suggests that the S1 response is not due to varies across response-frequency bands, as can be seen by transsynaptic spread from the stimulation site but rather to comparing the spatial maps of response power in the gamma direct illumination (see Figure S6D). In contrast, the response at the M1 site is delayed by 5.0 ± (30–60 Hz) and high-gamma bands (Figure 7D). While both bands show increasing power with increasing sinusoidal pulse 0.7 ms (mean ± SE), with the peak response lagging by 1.8 ± frequency, the rise is steeper in the gamma band, which also 0.5 ms, suggesting that this is a secondary response resulting shows little response to constant stimuli (left panel), suggesting from functional connectivity between S1 and M1. This concluthat the gamma-band response is more selective to stimulus sion is supported by the spatial and temporal spread of the stimulation-induced response, shown in Figure 8B as a series shape. These results show that the network response is strongly of pseudocolor activation maps representing 3 ms time bins dependent on stimulus waveform shape and not just a linear poststimulation. These plots show a spread of activity in S1 function of input power. This can be compared to Histed and about 6 ms following stimulation onset, with the M1 increase Maunsell (2014), who observed similar effects of waveform lagging behind. shape on single-unit spiking but who found that behavior depended linearly on input power, independent of waveform DISCUSSION shape. They suggested that waveform invariance could emerge through local circuitry, e.g., averaging of activity across cells, We have combined CED, artificial dura, and mECoG array techbut to the extent that high gamma is a good proxy for mean nologies to build a large, stable interface for optogentic stimulalocal activity, our results suggest this is not the case. This tion and recording in NHPs. We used an efficient CED technique example also demonstrates how our interface facilitates the for infusion of optogenetic viral vectors that yields large-scale rapid mapping of a stimulus set to large-scale spatiotemporal expression within M1 and S1. In order to interface with this large cortical region, we used a combination of artificial dura and responses. 8 Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc.


mECoG arrays, which allow for high-channel count recordings while maintaining optical access. Although the artificial dura for optogenetics (Ruiz et al., 2013) and ECoG (Rouse and Moran, 2009; Moran, 2010) have been used independently in NHPs, their combined use, together with CED infusions, provides the unique capability of optical stimulation with simultaneous recording from large regions of cortex spanning multiple brain areas. Furthermore, we maintained the capability for over 2 years in one animal, albeit with some cortical thinning over time. Mechanism and Effectiveness of CED The use of CED to infuse both drugs and viral vectors to the brain is well studied. Flow of infused agents through the interstitial space of the brain occurs both by diffusion and advection (bulk fluid flow) (Rosenberg et al., 1980; Fenstermacher and Kaye, 1988; Ohata and Marmarou, 1992; Bobo et al., 1994; Lieberman et al., 1995). While diffusion results from a concentration gradient and its rate decreases with molecular weight (Bobo et al., 1994; Lieberman et al., 1995), advection results from a pressure gradient; its flux is largely independent of molecular weight, and solutions are distributed in relatively homogeneous concentrations (Rosenberg et al., 1980; Fenstermacher and Kaye, 1988; Bobo et al., 1994; Lieberman et al., 1995). The Bankiewicz and Oldfield labs have demonstrated CED infusion of large molecules (such as viral vectors) into a variety of CNS structures, including thalamus (Kells et al., 2009), putamen (Sanftner et al., 2005), striatum (Szerlip et al., 2007), brainstem (Lonser et al., 2002; Krauze et al., 2005), corona radiata (Krauze et al., 2005), and spinal cord (Lonser et al., 1998a). These injections drive bulk flow through the interstitial space (Bobo et al., 1994; Lieberman et al., 1995), resulting in large and homogeneous distributions of vector (Bobo et al., 1994; Lieberman et al., 1995; Lonser et al., 1998a, 2002; Krauze et al., 2005; Sanftner et al., 2005; Szerlip et al., 2007). The linear relationship we observed over time between Vi and vector Vd in the tissue (Figure 2B) suggests a constant rate of convection from the infusion sites, with a slope of about 3 units of Vd for each unit of Vi. Previous CED studies have found the same linear dependence with slopes typically in the range of 1–4 (Lonser et al., 1998a, 2002, Krauze et al., 2005, 2008; Szerlip et al., 2007). Not all of the Vd was in the gray matter (see Figure S2); however, the constant ratio of Vd to Vi during CED suggests that the convection rates were similar in the gray matter and white matter. In addition, the histological analysis of viral expression (Figures 5C and S5B) shows no evidence that CED favors flow along the myelinated fiber tracts. Safety of the CED Infusions The safety of CED has previously been demonstrated both by histological analyses showing no tissue damage following CED infusions, beyond the cannula wound, and by the lack of postinfusion neurological deficits (Bobo et al., 1994; Lieberman et al., 1995; Lonser et al., 1998a, 2002; Krauze et al., 2005; Sanftner et al., 2005; Szerlip et al., 2007). These findings are consistent with reports that cerebral edema does not intrinsically cause irreversible neurological dysfunction (Rapoport and Thompson, 1973; Hossmann et al., 1980). Furthermore, Bankiewicz and col-

leagues have shown that higher infusion rates (>10 ml/min) are likely to induce local tissue damage in the CNS (Krauze et al., 2005); these authors suggested that safe delivery rates should not exceed 5 ml/min (the limit in our study). Based on animal studies such as these, several clinical trials have been initiated to study the safety and efficacy of CED for clinical use (Gill et al., 2003; Kunwar, 2003; Patel et al., 2005), including one using an AAV vector (Eberling et al., 2008). Previous studies have observed gliosis around the cannula track following CED infusions (Bobo et al., 1994; Lieberman et al., 1995; Krauze et al., 2005; Szerlip et al., 2007). Two of these (Bankiewicz et al., 2000; Lonser et al., 2002) quantified the damage present at up to 9 weeks following penetration of 350–410 mm diameter cannulae, comparable to our inner cannula. Gliosis was observed, extending 50–100 mm from the center of the cannula track. Similarly, we find gliosis around the cannula track 27 months postinfusion (Figure S5C), as well as a decrease in EYFP-reporter expression (Figure S5B) that indicates an initial lesion comparable in size to the cannula itself. A lack of expression at the injection sites has been seen in previous studies using traditional diffusion-based injections as well (Jazayeri et al., 2012). We conclude that the extent of tissue damage could be reduced by using a smaller-diameter cannula. In contrast, there is considerable evidence that the cortical tissue beyond the cannula site has not been substantially damaged from the CED procedure: (1) the MR images document an even flow of viral vector through the tissue, and the MR-estimated Vd matches the surface regions that exhibit both epifluorescence and light-evoked activity, both of which rely on intact cells; (2) no tissue damage was observed on the postinfusion MR scans or visually on the surface of the brain; (3) no histological evidence of tissue damage was seen outside the cannula track in regions of high viral expression; (4) while cortical thinning was observed, it was even across the cylinder and is due to the use of the cylinder and artificial dura (Chen et al., 2002); (5) the electrophysiological properties of the tissue closest to the infusion sites are the same as those in surrounding regions (Figure S7); and (6) there has been no evidence of neurological deficits in these animals. Comparison of CED and Conventional Infusions Conventional diffusion-based injections of 1–2 mL yield typical regions of expression of about 2–3 mm diameter; Diester and colleagues (Diester et al., 2011) saw 2 mm diameter, and Ruiz et al. (2013) and Lerchner et al. (2014) report no more than 3 mm diameter. Jazayeri et al. (2012) report a diameter of 4 mm from injection site; however, they also report a decrement in expression levels by 1 mm from the injection. In contrast, our epifluorescence imaging and histological analysis show vector spread of up to 10 mm diameter from the injection site. In principle, CED should also deliver more homogeneous vector densities across this region (Bobo et al., 1994; Lieberman et al., 1995; Lonser et al., 1998b; Szerlip et al., 2007; Fiandaca et al., 2008; Kells et al., 2009) compared to diffusion-based injections. This difference is reflected in observed patterns of expression; traditional injections yield regions of dense expression surrounded by weaker expression (Jazayeri et al., 2012; Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 9


Lerchner et al., 2014), whereas we observe large regions with high levels of expression (Figures 5C and S5B). The CED technique offers other advantages over traditional microinfusions. We obtained as much as 2 cm2 of coverage with only three injections. Assuming an expression diameter of 3 mm following conventional injections, it would take about 30 injections to achieve the same volume coverage. The reduced number of injections and faster infusion rates make it practical to perform CED infusions in the MRI, yielding more precise targeting and online monitoring. Evidence of Large-Scale Expression We have documented optogenetic expression with both in vivo epifluorescence imaging and with postmortem histology. Surface epifluorescence imaging through an artificial dura has previously been validated (Ruiz et al., 2013). That study showed that the regions of epifluorescence agreed with both the viral injection sites and the histologically identified regions of fluorescent reporter expression. We also observed good agreement between surface epifluorescence imaging and histology (Figures 4B and 5), further validating the in vivo technique. Additional evidence for large-scale expression comes from the reliable physiological responses to optical stimulation we observed in both M1 and S1 (Figures 5, 6, and 7). This evoked activity cannot be due to simple photoelectric artifact; since the activity extends tens of milliseconds beyond stimulus (Figure 7), we see activity distal to the stimulation site, and we do not see activity when stimulation is performed outside the sites exhibiting epifluorescence (Figure S8). The Use of mECoG Arrays Our mECoG-based interface can be contrasted with optogenetic approaches that use penetrating electrodes (Gerits et al., 2012; Jazayeri et al., 2012; Ruiz et al., 2013). mECoG arrays have the principal benefits of allowing the mapping of activity across large cortical areas and long-term stability (Schalk and Leuthardt, 2011; Buzsa´ki et al., 2012). In particular, since ECoG electrodes do not penetrate the pia arachnoid, they minimize the risk of cortical damage and allow for repeated recordings from the same sites over long periods of time. The data presented here were acquired with acute daily placement of the arrays, with no evident tissue damage (Figure S4C) or signal decrement over as many as 25 months of use. It is also feasible to make a chronic version of the setup (Yazdan-Shahmorad et al., 2015), where the arrays are left in place for several months, allowing for more precise long-term recording from the same sites. For cases where long-term stability of the interface is not critical, our setup also allows for the combined use of mECoG with penetrating probes. The mECoG arrays have periodic perforations (Figure 6A) that can serve as access points for standard penetrating electrodes or optical fibers. These could be used, for example, to record single units from sharp electrodes or laminar probes, to vary the depth of optical stimulation by inserting an optical fiber into the cortex, or to obtain more focal optical stimulation. For such experiments, inclusion of the mECoG array offers the substantial benefit of large-scale spatiotemporal mapping with little overhead. 10 Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc.

Potential Applications Lastly, we showed two examples of how this interface can be used to study neural circuit dynamics. The first example demonstrates the relationship between features of stimulus pulse shape and the cortical spatiotemporal response. The second example compares the temporal dynamics of activity propagation between and across cortical areas. Our system allows measurements such as these to be readily performed on a daily basis during behavioral studies, for example, to perform causal experiments mapping circuit dynamics to cognition and behavior. The stability of the interface also makes it ideal for studying the dynamics of long-term learning. The increasing use of optogenetics in NHPs suggests the possibility of clinical applications for the technology (Busskamp et al., 2010; Doroudchi et al., 2011; Han, 2012; Chow and Boyden, 2013). We note that optogenetics can aid in the development of stimulation-based therapies, even if they are ultimately implemented with electrical stimulation. For example, our interface allows for the manipulation of neural activity while performing simultaneous large-scale, artifact-free electrophysiological recordings; this makes it an ideal test bed for developing closed-loop stimulation paradigms for the treatment of neurological disorders or for studying neural circuit changes during long-term, continuous-stimulation paradigms. EXPERIMENTAL PROCEDURES Two adult male rhesus monkeys were used in this study. All procedures were performed under the approval of the University of California, San Francisco Institutional Animal Care and Use Committee and were compliant with the Guide for the Care and Use of Laboratory Animals. Surgical Procedures Cylinder implantation, viral injections, and the initial placement of the artificial dura were all performed in a single surgical session under isoflurane anesthesia and using standard aseptic technique. The cylinder implantation site was selected to span M1 and S1, guided by presurgical MR images. A 25 mm diameter craniotomy was made, and the underlying dura was resected. An MR-compatible, 3D-printed chamber that contained a nylon, custom-made cannula grid (Figures 1A–1C) was implanted over the craniotomy. For monkey G, the infusion holes spanned M1 and S1 (Figure 1E). For monkey J, the chamber was modified to allow rotation of the cannula grid, expanding the region of potential injection sites (see Figures S1D and S1E). After securing the injection chamber, the skin was closed around the chamber, and the animal was transported to the MR facility. Viral-Infusion Technique We used anatomical MR images (T1 and T2 weighted) to plan the location and depth of injection sites (Figures 1C, 1E, and 1F) and to validate infusion in real time. Animals were placed in an MR-compatible stereotaxic frame, and the cannula grid holes were filled with sterile saline. MR images were obtained using a surface coil in a Siemens 3T MR scanner. We acquired preliminary images of the brain and the injection cylinder in order to plan the injection sites. We used a reflux-resistant infusion cannula that consisted of two sections of silica tubing (0.32 mm inner diameter [ID] and 0.43 mm outer diameter [OD] for the inner tubing and 0.45 ID and 0.76 OD for the outer tubing) glued with cyanoacrylate such that the inner tube extended 1 mm beyond the outer tube, making the reflux-resistant step (Figure 1D shows the distal end of the cannula). The cannula was inserted approximately 2 mm below the pia. To monitor the infusion via real-time MRI, the viral vector was coinfused with Gadoteridol MR contrast agent (ratio of 250:1), and fast (2 min) flash T1-weighted images were acquired. We infused AAV5CamKIIa-C1V1-EYFP viral vector (2.5 3 1012 virus molecules/ml; UPenn


vector core) at four sites in monkey G (two sites in S1 and two sites in M1; Figure 1E) and five sites in monkey J (three sites in S1 and two sites in M1; Figure 1F). We infused 50 ml at each site with the infusion rate starting at 1 ml/min rate and increasing to 5 ml/min by 1 ml/min steps (Figure 1G). The time course of infusion for each site was about 21 min. The efficacy of the viral construct has been documented in four previous NHP studies (Ozden et al., 2013; Ruiz et al., 2013; Dai et al., 2014; May et al., 2014), and we tested our viral vector in cultured rat hippocampal neurons before use in NHPs. Implantation of the Artificial Dura After viral infusions, the animal was transported back to the surgery, and the MR-compatible cylinder was replaced with a titanium cylinder. To achieve good stability and a good seal between the cylinder and the bone, the cylinders were custom designed so that the base matched the coronal and sagittal skull curvatures (see Figure 3A), and the cylinders were 3D printed from titanium (alloyed with 6% aluminum and 4% vanadium, by weight; i.materialize; Data S1, TitaniumCylinder.ipt). We next inserted the artificial dura into the cylinder. The custom-made artificial dura was molded (Figure S4A) from silicone (mixture of KE1300-T and CAT-1300 with the ratio of 10:1; Shin-Etsu Chemical) in the shape of a vertical cylinder (2 cm ID) attached to a 0.4 mm thick ‘‘window’’ at the bottom with a 0.2 mm thick flange extending beyond the cylinder (Figure 3B) (Ruiz et al., 2013). The flange was inserted between the native dura and the cortex (Figures 3C and 3E), helping to prevent dural growth under the artificial dura window. After implanting the artificial dura, the chamber was sealed with a stainless steel cap (Data S1, cap.pdf). Sterile Teflon tape was applied to the outer threads of the cap to create a better seal and to prevent seizing of the threads between the titanium cylinder and the steel cap. Epifluorescence Imaging We monitored expression levels of the fluorescent EYFP reporter via surface epifluorescence imaging through the artificial dura (Diester et al., 2011; Ruiz et al., 2013). We scanned the exposed tissue with a blue laser (488 nm) while imaging the surface of the brain trough with a EYFP-emission filter with a video camera (JVC GZ-HM550BU camcorder). Image frames were averaged to estimate the expression areas. Cylinder Maintenance To avoid infection we developed a cleaning procedure that roughly followed the steps outlined in Ruiz et al. (2013). We cleaned the cylinder at least three times per week. First, we cleaned the exterior of the cylinder with aseptic solutions such as Nolvasan. After opening the cylinder, we rinsed inside the artificial dura and the areas between cylinder walls and artificial dura with sterile saline. A few drops of clear antibiotic were then added to the cylinder before the caps were replaced. Electrophysiological Recordings Neural recordings were performed using 96 channel mECoGs arrays. The arrays were fabricated at the Marvell Nanofabrication Laboratory at the University of California, Berkeley using a custom, multilayer process (Ledochowitsch et al., 2011a, 2011b). Briefly, they consisted of platinum-gold-platinum traces encapsulated in Parylene-C, a biocompatible polymer. The array’s integrated cable was long enough (2 cm) for the connectors to extend above the cylinder walls when the array was in place. Combined stimulation and recording experiments were performed 3 months postinjection, while awake animals were head fixed in a primate chair. The artificial dura was removed, and one or two mECoG arrays were placed on the arachnoid. We used a 488 nm laser (PhoxX 488-60, Omicron-Laserage) for optical stimulation, which is well inside the effective range for C1V1. Light was delivered through an optical fiber (core/cladding diameter was 62.5/125 mm, Fiber Systems), positioned with our custom-made stimulation setup so that the tip of the fiber optic touched the array (Figure S6). Power at the fiber tip was 20–22 mW, with a light spread of about 0.7 mm radius (Figure S6D). We used a Tucker-Davis Technologies system for mECoG recording and for controlling the laser stimulus. Detailed procedures for the electrophysiological recordings are listed in Table S2.

Data Analysis We estimated the spread of the viral-vector infusion by analyzing the T1-weighted MR images in Mango 3.4 (RRID: SCR_009603). We constructed a 3D volume of the brain, and injection volumes were defined by thresholding intensity in flash T1 images, which were taken sequentially during the injection. Pseudocolor maps of ECoG power were generated for gamma (30–60 Hz) and high gamma (60–200 Hz) from the power of the signal during 1 s of stimulation in the respective frequency range for each electrode in the array. These maps were then linearly interpolated and averaged across trials. Trials with high-movement artifact were excluded from the dataset. Histological Analysis Sections from monkey G were processed for EYFP immunocytochemistry using a free-floating technique. Additional adjacent sections were stained with cresyl violet (Nissl) using standard techniques to reveal cortical cytoarchitecture. Imaging was done using Nikon 90i microscope equipped with a color CCD camera. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, eight figures, and two tables and can be found with this article online at http://dx.doi.org/10.1016/j.neuron.2016.01.013. AUTHOR CONTRIBUTIONS A.Y.-S. designed interface and hardware, designed experiments, conducted experiments, analyzed data, performed histology analysis, and wrote the manuscript. C.D.-B. designed hardware, fabricated arrays, conducted experiments, analyzed data, and helped write the manuscript. T.L.H. designed hardware, fabricated hardware, and helped write the manuscript. V.K. performed histology analysis and helped write the manuscript. P.L. designed arrays, fabricated arrays, helped run experiments, and helped write the manuscript. M.M.M. designed arrays, fabricated arrays, and helped write the manuscript. P.N.S. designed interface and hardware, designed experiments, analyzed data, and wrote the manuscript. ACKNOWLEDGMENTS This work was supported by an American Heart Association Western States Affiliate Postdoctoral Fellowship (A.Y.-S.), Defense Advanced Research Projects Agency (DARPA) Reorganization and Plasticity to Accelerate Injury Recovery (REPAIR; N66001-10-C-2010), and by the UCSF Neuroscience Imaging Center. We thank the following: the staff of the Berkeley Sensor and Actuator Center (BSAC) and the Marvell Nanofabrication Laboratory (MNL) for assistance in array fabrication; Quynh Anh Nguyen, Jon Levy, and Alexander Jackson for testing the virus in cell culture; Jonathan Nassi and John Reynolds from the Salk Institute for advice on the artificial dura; John Bringas, Adrian Kells, and Krystof Bankiewicz from UCSF for extensive technical assistance with the CED procedure, as well as providing the cannula grid, material for the cannula assembly, and use of equipment; Karen J. MacLeod, Evelyn Escobar, and Blakely Andrews for their help with animal care and imaging; and Daniel Silversmith, Joseph O’Doherty, Jiwei He, Nan Tian, and Josh Chartier for their help with the experiments. P.L. and M.M.M. are cofounders and board members of Cortera Neurotechnologies, Inc. Received: November 24, 2014 Revised: July 28, 2015 Accepted: January 5, 2016 Published: February 11, 2016 REFERENCES Afraz, A., Boyden, E.S., and DiCarlo, J.J. (2015). Optogenetic and pharmacological suppression of spatial clusters of face neurons reveal their causal role in face gender discrimination. Proc. Natl. Acad. Sci. USA 112, 6730–6735.

Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 11


Anikeeva, P., Andalman, A.S., Witten, I., Warden, M., Goshen, I., Grosenick, L., Gunaydin, L.A., Frank, L.M., and Deisseroth, K. (2012). Optetrode: a multichannel readout for optogenetic control in freely moving mice. Nat. Neurosci. 15, 163–170. Arieli, A., Grinvald, A., and Slovin, H. (2002). Dural substitute for long-term imaging of cortical activity in behaving monkeys and its clinical implications. J. Neurosci. Methods 114, 119–133. Bankiewicz, K.S., Eberling, J.L., Kohutnicka, M., Jagust, W., Pivirotto, P., Bringas, J., Cunningham, J., Budinger, T.F., and Harvey-White, J. (2000). Convection-enhanced delivery of AAV vector in parkinsonian monkeys; in vivo detection of gene expression and restoration of dopaminergic function using pro-drug approach. Exp. Neurol. 164, 2–14. Bernstein, J.G., and Boyden, E.S. (2011). Optogenetic tools for analyzing the neural circuits of behavior. Trends Cogn. Sci. 15, 592–600. Bobo, R.H., Laske, D.W., Akbasak, A., Morrison, P.F., Dedrick, R.L., and Oldfield, E.H. (1994). Convection-enhanced delivery of macromolecules in the brain. Proc. Natl. Acad. Sci. USA 91, 2076–2080. Bonhoeffer, T., Kim, D.S., Malonek, D., Shoham, D., and Grinvald, A. (1995). Optical imaging of the layout of functional domains in area 17 and across the area 17/18 border in cat visual cortex. Eur. J. Neurosci. 7, 1973–1988. Boyden, E.S., Zhang, F., Bamberg, E., Nagel, G., and Deisseroth, K. (2005). Millisecond-timescale, genetically targeted optical control of neural activity. Nat. Neurosci. 8, 1263–1268. Busskamp, V., Duebel, J., Balya, D., Fradot, M., Viney, T.J., Siegert, S., Groner, A.C., Cabuy, E., Forster, V., Seeliger, M., et al. (2010). Genetic reactivation of cone photoreceptors restores visual responses in retinitis pigmentosa. Science 329, 413–417.

Fiandaca, M.S., Forsayeth, J.R., Dickinson, P.J., and Bankiewicz, K.S. (2008). Image-guided convection-enhanced delivery platform in the treatment of neurological diseases. Neurotherapeutics 5, 123–127. Gerits, A., Farivar, R., Rosen, B.R., Wald, L.L., Boyden, E.S., and Vanduffel, W. (2012). Optogenetically induced behavioral and functional network changes in primates. Curr. Biol. 22, 1722–1726. Gill, S.S., Patel, N.K., Hotton, G.R., O’Sullivan, K., McCarter, R., Bunnage, M., Brooks, D.J., Svendsen, C.N., and Heywood, P. (2003). Direct brain infusion of glial cell line-derived neurotrophic factor in Parkinson disease. Nat. Med. 9, 589–595. Han, X. (2012). Optogenetics in the nonhuman primate. Prog. Brain Res. 196, 215–233. Han, X., Qian, X., Bernstein, J.G., Zhou, H.H., Franzesi, G.T., Stern, P., Bronson, R.T., Graybiel, A.M., Desimone, R., and Boyden, E.S. (2009). Millisecond-timescale optical control of neural dynamics in the nonhuman primate brain. Neuron 62, 191–198. Han, X., Chow, B.Y., Zhou, H., Klapoetke, N.C., Chuong, A., Rajimehr, R., Yang, A., Baratta, M.V., Winkle, J., Desimone, R., and Boyden, E.S. (2011). A high-light sensitivity optical neural silencer: development and application to optogenetic control of non-human primate cortex. Front. Syst. Neurosci. 5, 18. Histed, M.H., and Maunsell, J.H. (2014). Cortical neural populations can guide behavior by integrating inputs linearly, independent of synchrony. Proc. Natl. Acad. Sci. USA 111, E178–E187. Hossmann, K.A., Blo¨ink, M., Wilmes, F., and Wechsler, W. (1980). Experimental peritumoral edema of the cat brain. Adv. Neurol. 28, 323–340.

Buzsa´ki, G., Anastassiou, C.A., and Koch, C. (2012). The origin of extracellular fields and currents–EEG, ECoG, LFP and spikes. Nat. Rev. Neurosci. 13, 407–420.

Jazayeri, M., Lindbloom-Brown, Z., and Horwitz, G.D. (2012). Saccadic eye movements evoked by optogenetic activation of primate V1. Nat. Neurosci. 15, 1368–1370.

Cavanaugh, J., Monosov, I.E., McAlonan, K., Berman, R., Smith, M.K., Cao, V., Wang, K.H., Boyden, E.S., and Wurtz, R.H. (2012). Optogenetic inactivation modifies monkey visuomotor behavior. Neuron 76, 901–907.

Kells, A.P., Hadaczek, P., Yin, D., Bringas, J., Varenika, V., Forsayeth, J., and Bankiewicz, K.S. (2009). Efficient gene therapy-based method for the delivery of therapeutics to primate cortex. Proc. Natl. Acad. Sci. USA 106, 2407–2411.

Chan, S., Bernstein, J., and Boyden, E. (2010). Scalable fluidic injector arrays for viral targeting of intact 3-D brain circuits. J. Vis. Exp. (35), 1489, http://dx. doi.org/10.3791/1489.

Krauze, M.T., Saito, R., Noble, C., Tamas, M., Bringas, J., Park, J.W., Berger, M.S., and Bankiewicz, K. (2005). Reflux-free cannula for convection-enhanced high-speed delivery of therapeutic agents. J. Neurosurg. 103, 923–929.

Chen, L.M., Heider, B., Williams, G.V., Healy, F.L., Ramsden, B.M., and Roe, A.W. (2002). A chamber and artificial dura method for long-term optical imaging in the monkey. J. Neurosci. Methods 113, 41–49.

Krauze, M.T., Vandenberg, S.R., Yamashita, Y., Saito, R., Forsayeth, J., Noble, C., Park, J., and Bankiewicz, K.S. (2008). Safety of real-time convectionenhanced delivery of liposomes to primate brain: a long-term retrospective. Exp. Neurol. 210, 638–644.

Chen, L.M., Friedman, R.M., and Roe, A.W. (2005). Optical imaging of SI topography in anesthetized and awake squirrel monkeys. J. Neurosci. 25, 7648–7659. Chow, B.Y., and Boyden, E.S. (2013). Optogenetics and translational medicine. Sci. Transl. Med. 5, 177ps5. Crone, N.E., Sinai, A., and Korzeniewska, A. (2006). High-frequency gamma oscillations and human brain mapping with electrocorticography. Prog. Brain Res. 159, 275–295. Dai, J., Brooks, D.I., and Sheinberg, D.L. (2014). Optogenetic and electrical microstimulation systematically bias visuospatial choice in primates. Curr. Biol. 24, 63–69. Diester, I., Kaufman, M.T., Mogri, M., Pashaie, R., Goo, W., Yizhar, O., Ramakrishnan, C., Deisseroth, K., and Shenoy, K.V. (2011). An optogenetic toolbox designed for primates. Nat. Neurosci. 14, 387–397. Doroudchi, M.M., Greenberg, K.P., Liu, J., Silka, K.A., Boyden, E.S., Lockridge, J.A., Arman, A.C., Janani, R., Boye, S.E., Boye, S.L., et al. (2011). Virally delivered channelrhodopsin-2 safely and effectively restores visual function in multiple mouse models of blindness. Mol. Ther. 19, 1220–1229. Eberling, J.L., Jagust, W.J., Christine, C.W., Starr, P., Larson, P., Bankiewicz, K.S., and Aminoff, M.J. (2008). Results from a phase I safety trial of hAADC gene therapy for Parkinson disease. Neurology 70, 1980–1983. Fenstermacher, J., and Kaye, T. (1988). Drug ‘‘diffusion’’ within the brain. Ann. N Y Acad. Sci. 531, 29–39.


Kunwar, S. (2003). Convection enhanced delivery of IL13-PE38QQR for treatment of recurrent malignant glioma: presentation of interim findings from ongoing phase 1 studies. Acta Neurochir. Suppl. (Wien) 88 (Supplement 88 ), 105–111. Ledochowitsch, P., Fe´lus, R.J., Gibboni, R.R., Miyakawa, A., Bao, S., and Maharbiz, M.M. (2011a) Fabrication and testing of a large area, high density, parylene MEMS mECoG array. IEEE International Conference on Micro Electro Mechanical Systems - MEMS 1031–1034. Ledochowitsch, P., Olivero, E., Blanche, T., and Maharbiz, M.M. (2011b). A transparent mECoG array for simultaneous recording and optogenetic stimulation. Conf. Proc. IEEE Eng. Med. Biol. Soc. 2011, 2937–2940. Ledochowitsch, P., Yazdan-Shahmorad, A., Bouchard, K.E., Diaz-Botia, C., Hanson, T.L., He, J.W., Seybold, B.A., Olivero, E., Phillips, E.A., Blanche, T.J., et al. (2015). Strategies for optical control and simultaneous electrical readout of extended cortical circuits. J. Neurosci. Methods 256, 220–231. Lerchner, W., Corgiat, B., Der Minassian, V., Saunders, R.C., and Richmond, B.J. (2014). Injection parameters and virus dependent choice of promoters to improve neuron targeting in the nonhuman primate brain. Gene Ther. 21, 233–241. Lieberman, D.M., Laske, D.W., Morrison, P.F., Bankiewicz, K.S., and Oldfield, E.H. (1995). Convection-enhanced distribution of large molecules in gray matter during interstitial drug infusion. J. Neurosurg. 82, 1021–1029.


Lonser, R.R., Gogate, N., Morrison, P.F., Wood, J.D., and Oldfield, E.H. (1998a). Direct convective delivery of macromolecules to the spinal cord. J. Neurosurg. 89, 616–622. Lonser, R.R., Weil, R.J., Morrison, P.F., Governale, L.S., and Oldfield, E.H. (1998b). Direct convective delivery of macromolecules to peripheral nerves. J. Neurosurg. 89, 610–615. Lonser, R.R., Walbridge, S., Garmestani, K., Butman, J.A., Walters, H.A., Vortmeyer, A.O., Morrison, P.F., Brechbiel, M.W., and Oldfield, E.H. (2002). Successful and safe perfusion of the primate brainstem: in vivo magnetic resonance imaging of macromolecular distribution during infusion. J. Neurosurg. 97, 905–913. May, T., Ozden, I., Brush, B., Borton, D., Wagner, F., Agha, N., Sheinberg, D.L., and Nurmikko, A.V. (2014). Detection of optogenetic stimulation in somatosensory cortex by non-human primates–towards artificial tactile sensation. PLoS ONE 9, e114529. Moran, D. (2010). Evolution of brain-computer interface: action potentials, local field potentials and electrocorticograms. Curr. Opin. Neurobiol. 20, 741–745. Nassi, J.J., Avery, M.C., Cetin, A.H., Roe, A.W., and Reynolds, J.H. (2015). Optogenetic activation of normalization in alert macaque visual cortex. Neuron 86, 1504–1517. Ohata, K., and Marmarou, A. (1992). Clearance of brain edema and macromolecules through the cortical extracellular space. J. Neurosurg. 77, 387–396. Ohayon, S., Grimaldi, P., Schweers, N., and Tsao, D.Y. (2013). Saccade modulation by optical and electrical stimulation in the macaque frontal eye field. J. Neurosci. 33, 16684–16697. Ozden, I., Wang, J., Lu, Y., May, T., Lee, J., Goo, W., O’Shea, D.J., Kalanithi, P., Diester, I., Diagne, M., et al. (2013). A coaxial optrode as multifunction writeread probe for optogenetic studies in non-human primates. J. Neurosci. Methods 219, 142–154. Park, D.W., Schendel, A.A., Mikael, S., Brodnick, S.K., Richner, T.J., Ness, J.P., Hayat, M.R., Atry, F., Frye, S.T., Pashaie, R., et al. (2014). Graphenebased carbon-layered electrode array technology for neural imaging and optogenetic applications. Nat. Commun. 5, 5258. Patel, N.K., Bunnage, M., Plaha, P., Svendsen, C.N., Heywood, P., and Gill, S.S. (2005). Intraputamenal infusion of glial cell line-derived neurotrophic factor in PD: a two-year outcome study. Ann. Neurol. 57, 298–302. Portugues, R., Severi, K.E., Wyart, C., and Ahrens, M.B. (2013). Optogenetics in a transparent animal: circuit function in the larval zebrafish. Curr. Opin. Neurobiol. 23, 119–126. Rapoport, S.I., and Thompson, H.K. (1973). Osmotic opening of the bloodbrain barrier in the monkey without associated neurological deficits. Science 180, 971. Ray, S., Crone, N.E., Niebur, E., Franaszczuk, P.J., and Hsiao, S.S. (2008). Neural correlates of high-gamma oscillations (60-200 Hz) in macaque local

field potentials and their potential implications in electrocorticography. J. Neurosci. 28, 11526–11536. Richner, T.J., Thongpang, S., Brodnick, S.K., Schendel, A.A., Falk, R.W., Krugner-Higby, L.A., Pashaie, R., and Williams, J.C. (2014). Optogenetic micro-electrocorticography for modulating and localizing cerebral cortex activity. J. Neural Eng. 11, 016010. Rosenberg, G.A., Kyner, W.T., and Estrada, E. (1980). Bulk flow of brain interstitial fluid under normal and hyperosmolar conditions. Am. J. Physiol. 238, F42–F49. Rouse, A.G., and Moran, D.W. (2009). Neural adaptation of epidural electrocorticographic (EECoG) signals during closed-loop brain computer interface (BCI) tasks. Conf. Proc. IEEE Eng. Med. Biol. Soc. 2009, 5514–5517. Ruiz, O., Lustig, B.R., Nassi, J.J., Cetin, A., Reynolds, J.H., Albright, T.D., Callaway, E.M., Stoner, G.R., and Roe, A.W. (2013). Optogenetics through windows on the brain in the nonhuman primate. J. Neurophysiol. 110, 1455– 1467. Sanftner, L.M., Sommer, J.M., Suzuki, B.M., Smith, P.H., Vijay, S., Vargas, J.A., Forsayeth, J.R., Cunningham, J., Bankiewicz, K.S., Kao, H., et al. (2005). AAV2-mediated gene delivery to monkey putamen: evaluation of an infusion device and delivery parameters. Exp. Neurol. 194, 476–483. Schalk, G., and Leuthardt, E.C. (2011). Brain-computer interfaces using electrocorticographic signals. IEEE Rev. Biomed. Eng. 4, 140–154. Szerlip, N.J., Walbridge, S., Yang, L., Morrison, P.F., Degen, J.W., Jarrell, S.T., Kouri, J., Kerr, P.B., Kotin, R., Oldfield, E.H., and Lonser, R.R. (2007). Real-time imaging of convection-enhanced delivery of viruses and virus-sized particles. J. Neurosurg. 107, 560–567. Wu, F., Stark, E., Im, M., Cho, I.J., Yoon, E.S., Buzsa´ki, G., Wise, K.D., and Yoon, E. (2013). An implantable neural probe with monolithically integrated dielectric waveguide and recording electrodes for optogenetics applications. J. Neural Eng. 10, 056012. Yazdan-Shahmorad, A., Kipke, D.R., and Lehmkuhle, M.J. (2013). High g power in ECoG reflects cortical electrical stimulation effects on unit activity in layers V/VI. J. Neural Eng. 10, 066002. Yazdan-Shahmorad, A., Diaz-Botia, C., Hanson, T., Ledochowitsch, P., Maharabiz, M.M., and Sabes, P.N. (2015). Demonstration of a setup for chronic optogenetic stimulation and recording across cortical areas in non-human primates. In SPIE Proceedings 9305, http://dx.doi.org/10.1117/12. 2080405. Yizhar, O., Fenno, L.E., Davidson, T.J., Mogri, M., and Deisseroth, K. (2011a). Optogenetics in neural systems. Neuron 71, 9–34. Yizhar, O., Fenno, L.E., Prigge, M., Schneider, F., Davidson, T.J., O’Shea, D.J., Sohal, V.S., Goshen, I., Finkelstein, J., Paz, J.T., et al. (2011b). Neocortical excitation/inhibition balance in information processing and social dysfunction. Nature 477, 171–178.

Neuron 89, 1–13, March 2, 2016 ª2016 Elsevier Inc. 13

Assessing anxiety in nonhuman primates.

Operant nociception in nonhuman primates.

A new quantitative rating scale for dyskinesia in nonhuman primates.

Harms and deprivation of benefits for nonhuman primates in research.

Spatiotemporal structure of intracranial electric fields induced by transcranial electric stimulation in humans and nonhuman primates.

Small intestinal transplantation in nonhuman primates.

Amyloid precursor protein in aged nonhuman primates.

Optogenetic Stimulation and Recording of Primary Cultured Neurons with Spatiotemporal Control.

Chronic wasting disease agents in nonhuman primates.

Letter: Protein-calorie deficiency in nonhuman primates.

Social inequalities in health in nonhuman primates.

Detection of optogenetic stimulation in somatosensory cortex by non-human primates--towards artificial tactile sensation.

Fabrication of High Contact-Density, Flat-Interface Nerve Electrodes for Recording and Stimulation Applications.

SFTS virus infection in nonhuman primates.

Long-term lung transplantation in nonhuman primates.

Improving genome assemblies and annotations for nonhuman primates.

Simultaneous transcranial magnetic stimulation and single-neuron recording in alert non-human primates.

Electric Field Model of Transcranial Electric Stimulation in Nonhuman Primates: Correspondence to Individual Motor Threshold.

Implantation and Establishment of Pregnancy in Human and Nonhuman Primates.

A Wireless Optogenetic Headstage with Multichannel Electrophysiological Recording Capability.

Modification of spectral features by nonhuman primates.

Transmission of human leukaemia to nonhuman primates.

Distinct Lineages of Bufavirus in Wild Shrews and Nonhuman Primates.

Early-life Social Adversity and Developmental Processes in Nonhuman Primates.