Long-Term Two-Photon Imaging in Awake Macaque Monkey Ming Li, Fang Liu, Hongfei Jiang, Tai Sing Lee, Shiming Tang  Neuron  Volume 93, Issue 5, Pages 1049-1057.e3 (March 2017) DOI: 10.1016/j.neuron.2017.01.027 Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 1 The Optical Window for Two-Photon Imaging in Awake Monkey (A) Photograph of the optical chamber before implantation, which is composed of a titanium ring mounted with the white ring-shape GORE-membrane (20 mm in outer diameter). (B) A coverslip (8 mm in diameter and 0.17–0.2 mm in thickness) was fitted to the titanium ring and then gently pressed down onto the cortical surface. The GORE membrane was tucked under the dura to prevent dura tissue infiltration into the chamber. Dental acrylic was filled between the titanium ring and the skull to support the imaging chamber. (C) Photograph of the cortex through the optical window demonstrates the clarity of the optical access, which can be maintained for months. Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 2 Two-Photon Imaging of Cortical Neurons in Awake Macaques (A) Fluorescence image of a region in V1 in monkey A, 6 months after AAV-GCaMP5 injection. (B) An example frame of an 850 × 850 μm two-photon imaging area, in V1 superficial layer (180 μm depth), expanded from the white box in (A), under a 16× objective. About 2,000 neurons could be clearly identified and stably monitored for many months. See also Figure S1 for a zoom-in image. (C) Multi-layer two-photon images from the surface down to 600 μm depth. See also Movie S1. Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 3 Long-Term Stability of Two-Photon Images of V1 Neurons in Awake Monkeys (A) We tracked the neurons across recording sessions based on their relative spatial locations and the shapes of their cell bodies to confirm long-term stability of the imaging conditions. Two-photon images from monkey A on Day 65 (captured at the resolution of 1,024 × 1,024 pixels over the 850 μm × 850 μm imaged window) and Day 160 (captured at the resolution of 512 × 512 pixels). Correspondence between the cells imaged in the two sessions can be established for most of the individual neurons even 95 days apart. (B) The central regions (265 × 212 μm) of the two images were enlarged to show the details. The sub-image for Day 65 has been rotated clockwise (3°) to maximize match. The cell pairs with correspondence were marked by red points and those without correspondence were marked by cyan points. Roughly 90% of the individual cells (81 of 90) were identifiable in this region from the static images and tracked across 95 days. Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 4 Motion Correction of Two-Photon Image in Awake Monkeys (A) An example single frame from a T-Series images (No. 1,000). (B) An average frame used as a template, normally computed by averaging a few hundred frames in the middle of an imaging session. (C) 2D cross-correlation of the single frame (frame No. 1,000) and the template shows that the individual frame has shifts in x-y plane (dx = 2, dy = 4) relative to the template. This individual frame (frame No. 1,000) should be corrected with negative shifts of this amount to align with the template. All the frames are aligned to the template in the same manner. (D) The averaged frame of 250,000 frames in a session before correction. (E) The average of the same 250,000 individual frames after alignment correction. The realignment was accurate enough to allow dendrites of one pixel in width be discernible in the images. Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 5 Robust and Long-Term Stable Visual Responses of Imaged Neurons (A and C) The neuronal population responses to drifting gratings (30° and 120° orientations, respectively) in V1 from monkey A. (B and D) Orientation tuning curves for the neurons marked by the arrows in (A) and (C) (averaged over 8 repetitions). Error bars, mean ± SEM. (E) Spatial organizations of these orientation-tuned neurons (cells colored according to their preferred orientation; Ohki et al., 2005). (F and H) The neuronal population responses to drifting gratings in V1 from monkey B in day 127 from AAV injection. (G and I) The neuronal population responses in day 146. See also Figure S2 for raw traces of calcium signal. (J) Mean responses plots for 51 cells on day 127 and 146. See also Figure S3 for additional orientation tuning distributions of V1 neurons. (K) Correlation of responses (ΔF/F0) on day 127 with day 147 for all strong activated cells (ΔF/F0 > 3 SD, n = 213 cells). Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 6 Simultaneously Intracellular Recording and Two-Photon Imaging in Awake Monkey (A) Intracellular recording from a GCaMP5-expressing cell. Sharp electrodes filled with potassium acetate (2 M) and Alexa 594 (20 μM) were used to perform intracellular recording, current injection, and dye labeling on target cells under two-photon imaging. See also Figures S4 and S5. (B) Typical intracellular recording from an example cell (cell 1). The amplifier was set to I-clamp mode. The middle red curve indicates the injected current. The bottom green curve indicates the calcium signal from the target cell. Notably, there was large subthreshold activity above the resting membrane potential near −65 mV, but no strong calcium signal activity under resting conditions. (C and D) Action potentials evoked by injection of 400 pA and 500 pA, respectively. The 400 pA current evoked 33 spikes (66 Hz) and a 43.4% fluorescence change (ΔF/F) from the recorded cell, while the 500 pA current evoked 40 spikes (80 Hz) and a 62.8% in calcium signal. (E) Relationship between the spike rate and the calcium signal of a neuron. These two signals were linearly correlated across firing rates up to 120 Hz with a standard error of 7.8% by linear fit. The two example points in (C) and (D) are indicated by the left and right arrows, respectively. (F–I) Relationships between the spike rate and the calcium signal in four other cells (cells 2 to 5) are also over a large range of firing rates in all these cases (standard errors 4.2% [2], 11.8% [3], 13.8% [4], and 15.6% [5]). See also Figure S6 for a single-unit recording in monkey A. Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 7 Sparse Labeling and Dendritic Imaging of Cortical Neurons in Awake Macaques (A) Sparse labeling of GCaMP6s in V4 neurons in monkey D, produced by injection of low concentration of AAV1.hSyn.Cre and AAV1.CAG.Flex.GCaMP6s. (B) Two-photon imaging of the spontaneous activities on the somas and dendrite branches. (C) Sparse labeling of tdTomoto in V1 neurons in monkey C, produced by injection of low concentration of AAV1.hSyn.Cre and AAV1.CAG.Flex.tdTomato. (D) A high-resolution two-photon image of one neuron that was expressing eGFP third day after a single-cell electroporation of pCMV-eGFP plasmid shows clearly observable dendrites and spines in an awake monkey (monkey C). Neuron 2017 93, 1049-1057.e3DOI: (10.1016/j.neuron.2017.01.027) Copyright © 2017 Elsevier Inc. Terms and Conditions