download (10.43 mb )
Post on 14-Feb-2017
238 Views
Preview:
TRANSCRIPT
Cell Reports, Volume 17
Supplemental Information
Long-Term Optical Access to an Estimated
One Million Neurons in the Live Mouse Cortex
Tony Hyun Kim, Yanping Zhang, Jérôme Lecoq, Juergen C. Jung, Jane Li, HongkuiZeng, Cristopher M. Niell, and Mark J. Schnitzer
Long-term optical access to an estimated one million
neurons in the live mouse cortex Tony Hyun Kim1,2, Yanping Zhang1-3, Jérôme Lecoq1,2,4, Juergen C. Jung1,2, Jane Li1,
Hongkui Zeng4, Cristopher M. Niell5 and Mark J. Schnitzer1-3,
1James H. Clark Center for Biomedical Engineering & Sciences, 2CNC Program, 3Howard Hughes Medical Institute, Stanford University, Stanford, CA 94305, USA
4Allen Institute for Brain Science, Seattle, WA 98109, USA 5Institute of Neuroscience, University of Oregon, Eugene, OR 97403, USA
Supplemental Information
Supplemental Experimental Procedures.
Table S1 | Surgical protocol for installing the Crystal Skull, Related to Figure 1.
Figure S1 | Maintenance of normal pressure levels on the brain is a key aspect of installing the
Crystal Skull, Related to Figure 1.
Figure S2 | A custom stainless steel head-plate surrounds the Crystal Skull and protects the
glass implant when it is not in use for imaging, Related to Figure 1.
Figure S3 | Histologic analysis reveals minuscule levels of astrocytic activation in the neocortex
of Crystal Skull mice, Related to Figure 1.
Supplemental Movie 1 | Epi-fluorescence Ca2+ imaging through the Crystal Skull across the
cortex of a GCaMP6f-tTA-dCre transgenic mouse reveals neuropil and single cell Ca2+ activity.
Supplemental Movie 2 | Epi-fluorescence Ca2+ imaging through the Crystal Skull over the
retrospenial area of a GCaMP6f-tTA-dCre transgenic mouse reveals both neuropil and single
cell Ca2+ activity.
Supplemental Movie 3 | Epi-fluorescence video of Ca2+ activity in the same mouse as in
Movie 2, but acquired three weeks later, at 11 weeks after implantation of the curved glass
window.
Supplemental Movie 4 | Two-photon Ca2+ imaging through the Crystal Skull in a tetO-
GCaMP6s/CaMK2a-tTA transgenic mouse, from the red region-of-interest marked in Figure 4C.
Access to a million mouse neurons in vivo Kim et al., 2016
p. S2
Supplemental Experimental Procedures
Mouse behavior setup. Starting 4 d prior to performing the behavioral assay of Figure 3C, we
placed mice (n = 2) on a water restriction regimen. We maintained the mice at 85–90% of their
original weight by restricting water intake to ~1 mL per day.
We accustomed the water-restricted mice to head-fixation under the objective lens of the
fluorescence macroscope while allowing them to use the running wheel. A light-emitting diode
(LED) emitted blue illumination (470 nm center wavelength) and was situated on the mouse’s
right side, at a 60° angle to the midline of the mouse’s face. Blue light passed through a band-
pass filter (Semrock; FF01-460/60-25) positioned ~5 cm away from the mouse’s right eye. A
feeding tube gated by a solenoid valve (Valcor; SV74P61T-1) delivered water rewards (5–10 µL
water per reward).
Mosaic images of neural morphology. To produce images of neural morphology across the
cortex (Figures 2B and 2E), we converted the tiled set of two-photon images-stacks into a set
of two-dimensional projection images using the image processing software environment ImageJ
(NIH). For each tile (i.e. an individual image stack) we used the Despeckle function to remove
any pixels with outlier values of fluorescence intensity. We then produced side-projection
images using the 3D_project function. We stitched together the individual projection images
using custom software written in MATLAB (Mathworks).
Computational identification of individual neurons. To extract individual cells and their
activity traces from the epi-fluorescence Ca2+ movies, we first corrected computationally for
brain motion using the TurboReg software routine (Thevanaz et al., 1998) to perform rigid image
displacements and rotations. To facilitate cell sorting, we next reduced the visual prominence of
neuropil by dividing each image frame by a spatially low-pass filtered version of itself that we
attained using the FFT_filter function in ImageJ. We selected the cutoff distance of the low-pass
filter to be several times larger than the diameter of a cell soma. Division of the original image
Access to a million mouse neurons in vivo Kim et al., 2016
p. S3
frame by its low-pass filtered version reduced image features at low spatial frequencies and
highlighted features of individual neurons. Using the resulting filtered movie we computed the
mean image, F0, by averaging each pixel across the movie duration. We subtracted F0 from
each image frame to compute a movie of the normalized changes in
fluorescence, ΔF(t)/F0 = (F(t)–F0)/F0 (Supplementary Movie 3). We then extracted individual
cells and their activity traces using a cell sorting method based on maximum likelihood
maximization (Deneux et al., 2016; Harris et al., 2016).
To sort individual cells and activity traces from two-photon Ca2+ movies, we first
corrected computationally for brain motion by using TurboReg to perform rigid transformations.
To extract the individual cells, we then applied the constrained non-negative matrix factorization
(CNMF) cell sorting algorithm (Pnevmatikakis et al., 2016).
For both the one- and two-photon Ca2+ imaging studies, we used custom software
written in MATLAB to visually check all candidate cells produced by the cell-sorting algorithms
and confirm each cell had a morphology and Ca2+ activity trace consistent with a neocortical
pyramidal cell. For each candidate cell, the software presented its spatial outline and activity
trace. The user evaluated the quality of the candidate cell by synchronously playing the ΔF(t)/F0
movie and the extracted Ca2+ activity trace while ensuring: (1) Strong concordance between the
Ca2+ activity seen in the outlined region of the movie and the extracted trace; (2) Lack of cross-
contamination in the trace from nearby cells; (3) The cell sorting algorithm had not split the cell
in the raw video into two candidate cells; (4) The identified source of Ca2+ activity was a soma
and not a dendrite. (We excluded all dendrites from cell number tallies).
Histological analysis. We deeply anesthetized mice with isoflurane (4–5%) and transcardially
perfused them with 20 mL of 1× Dulbecco’s phosphate-buffered saline (Fischer Scientific;
DPBS), and then 30 mL of 4% paraformaldehyde in phosphate buffer (Santa Cruz
Biotechnology; SC-281692). After perfusion, we removed the brains and post-fixed them (4%
Access to a million mouse neurons in vivo Kim et al., 2016
p. S4
paraformaldehyde in phosphate buffer) for 4 h at room temperature, and then post-fixed them
overnight at 4 ºC. Afterward, we cut the brains into 50 µm-thick coronal sections with a
vibratome (Leica; VT 1200).
We performed all staining at room temperature as follows. We washed the floating
sections in PBST (0.5% Triton X-100 in 1× DPBS buffer) four times, each for 20 min, and then
incubated them in blocking solution (5% BSA in PBST). After removing the sections from
blocking solution, we incubated them overnight with the primary antibody, mouse anti-GFAP
(Millipore, MAB3402) 1:1000-diluted in blocking solution. The next day we washed the sections
with PBST four times as before, and incubated them for 3 h with the secondary antibody, Alexa
Fluor 594 Donkey anti-mouse IgG 1:500-diluted in blocking buffer. We transferred the sections
to 1:25000-diluted DAPI (BioChemical; A1001,0010) in PBST for 20 min, and then washed them
three times with 1× PBS buffer. After washing with PBS, we mounted the sections onto
microscope slides with fluoromount-G (Southern Biotech, #0100-01). We inspected the tissue
sections using a commercial two-photon microscope (Bruker) and a 20× 0.45 NA microscope
objective lens (Olympus; LUCPlanFL N).
Access to a million mouse neurons in vivo Kim et al., 2016
p. S5
Table S1 | Surgical protocol for installing the Crystal Skull, Related to Figure 1.
Step # Description of surgical step 1 Measure and record the mass of the mouse.
2 Put the mouse into the anesthesia induction chamber and apply 4% isoflurane in O2 until the mouse’s breathing rate reduces to ~1 s-1.
3 Mount the mouse into the stereotaxic apparatus while providing 1.5–2% isoflurane in O2 to maintain the mouse’s breathing rate at ~1 s-1.
4 Apply eye ointment to both of the mouse’s eyes.
5 Inject carprofen (5–10 mg/kg body weight) and dexamethasone (2 mg/kg) subcutaneously to help prevent inflammation and edema.
6 Remove hair from the surgical area.
7 Disinfect the surgical area with betadine and swab it with 75% ethanol two times.
8 Remove an approximately circular area of skin (0.8–1 cm diameter) atop the skull.
9 Clean off the connective tissue (periosteum) on the surface of the skull, extending from approximately 4.0 mm anterior to bregma to 5.0 mm posterior to bregma.
10 Confirm that all soft tissue is removed from the dorsal surface of the skull.
11 Mark the bregma position on the skull using the stereotaxic apparatus and record the coordinates.
12 Apply mammalian Ringer’s solution (Electron Microscopy Science; # 11763-10) to the skull and continuously perfuse it throughout steps 13–22.
13 Using a 0.7-mm-diameter drill bit (Fine Science Tools; #19007-07), drill a shallow groove in the skull matching the trapezoidal shape of the Crystal Skull
14 Deepen the groove using a 0.5-mm-diameter drill bit (Fine Science Tools; #19007-05) until the trapezoidal bone piece is nearly detached from the skull.
15 Carefully free the trapezoidal piece from the surrounding skull, proceeding from the anterior edge, to the posterior edge and then to the lateral edges.
16 Clean the surgical area with Ringer’s solution, making sure to remove all debris.
17 While continuously perfusing Ringer’s solution, press down on the front edge of the detached skull piece. This elevates the piece’s posterior end, creating a gap between the detached piece and the dura.
18 Pick up the Crystal Skull cover glass using vacuum suction through a blunt needle (20 G).
19 Slide the cover glass into the gap created at the posterior end of the skull. Lubricate the interface between the cover glass and the brain with Ringer’s solution.
20 When 2/3 of the cover glass has been inserted, gently lift and remove the bone piece with forceps.
21 Move the Crystal Skull cover glass into its final position by aligning it to the bregma coordinates. 22 Clean the surgical area again with Ringer’s solution.
23 Dry the edges of the cover glass and skull with surgical eye spears (Butler Schein Animal Health; #1556455).
24 Affix the cover glass to the skull with UV-light-curable glue (Loctite; #4035). Cure the glue with a UV light source (Lightening Enterprises; LED-100) for ~10 s while keeping the brain’s UV light exposure to a minimum and protecting the mouse’s eyes.
25 Position the steel head-plate on the mouse’s head so that the opening is centered on the glass implant.
26 Fill the gaps between the skull and head-plate with UV glue. Cure the glue.
27 Mount the protective cap on the head-plate, to protect the Crystal Skull implant.
28 Transfer the mouse to a recovery cage. When it awakes fully, transfer the mouse to its home cage.
29 Provide water and food on the floor of the home cage, without using a food hopper.
30 For three days after the surgery, provide subcutaneous carprofen (5–10 mg/kg) and dexamethasone (2 mg/kg) once per day. Monitor the mouse’s health at least once daily.
A B C
D E
Craniotomy is complete andthe trapezoidal-shaped bone
is detached from the skull
Figure S1
Access to a million mouse neurons in vivo Kim et al., 2016
p. S6
Figure S1 | Maintenance of normal pressure levels on the brain is a key aspect of
installing the Crystal Skull, Related to Figure 1.
(A) We created a single craniotomy on the mouse skull along a trapezoidal outline. The central
piece (trapezoidal area shown in yellow) is disconnected from the rest of the cranium.
(B) By pressing down gently on the anterior edge of the loose skull piece, we lift up its posterior
end, thereby creating a gap between the skull and the surface of the brain.
(C) We slide the Crystal Skull implant (cyan) underneath the loose bone from its posterior end,
with a slight upward tilt of the glass to avoid scraping the surface of the brain. The interface
between the glass and the brain is lubricated with mammalian Ringer’s solution throughout.
(D) Once about two-thirds of the implant is inserted, we detach the loose skull with fine forceps.
We continue to slide the Crystal Skull implant forward, moving it into its final position.
(E) Once the implant is in position, we dry the edges of the glass and the nearby skull with
surgical eye spears. The implant is affixed to the skull using ultraviolet-light-cured optical glue.
12 mmM1.6 threaded hole
(2 places)
3D printed protectivecap for Crystal Skull
A B
Figure S2
Access to a million mouse neurons in vivo Kim et al., 2016
p. S7
Figure S2 | A custom stainless steel head-plate surrounds the Crystal Skull and protects
the glass implant when it is not in use for imaging, Related to Figure 1.
(A) In addition to the Crystal Skull optical implant, we mount a custom, 1.5-mm-thick stainless
steel head-plate that allows head-fixation of the mouse. The head-plate has two threaded screw
holes for mounting auxiliary apparatus to the head.
(B) We use a custom, 3D-printed plastic cap to protect the large window when optical access is
not needed (e.g. when the animal is resting in its home cage).
A
B
C
D
E
AP = -1.3 mm
A B C
AP = -2.9 mm
D E
GFAPDAPI
GCaMP6f
Figure S3
Positive control Negative controlG H
F
Access to a million mouse neurons in vivo Kim et al., 2016
p. S8
Figure S3 | Histologic analysis reveals minuscule levels of astrocytic activation in the
neocortex of Crystal Skull mice, Related to Figure 1.
(A–E) Immunohistological analysis revealed comparable levels of neocortical astrocytes in
Crystal Skull mice as in negative control mice (Figure 1C) that had no surgeries. We took brain
tissue samples 2–5 weeks after surgery and stained them with DAPI, which labels the nuclei of
all cells (blue), and anti-glial fibrillary acidic protein (anti-GFAP; red). Images are of five
randomly selected neocortical tissue sites in a GCaMP6f-tTA-dCre mouse with a Crystal Skull.
Images in the middle and right columns are shown at a higher magnification than images in the
left column. Images in the right column also show the green fluorescence from GCaMP6f. Scale
bars are 50 µm in the middle and right columns and 100 µm in the left column.
(F) Coronal section maps, based on those in (Paxinos and Franklin, 2012), show the
approximate locations of the tissue samples in A–E.
(G, H) Brain tissue samples, stained with DAPI and anti-GFAP, from a positive control mouse,
G, in which we deliberately nicked the brain during surgery to cause tissue damage, and a
negative control mouse, H, that had no surgeries and for which the tissue sample underwent the
immunostaining protocol but with the primary antibody deliberately withheld. For both G and H
the scale bars are 200 µm (left panels) and 50 µm (right panels).
Access to a million mouse neurons in vivo Kim et al., 2016
p. S9
Movie Legends
Supplemental Movie 1 | Epi-fluorescence Ca2+ imaging through the Crystal Skull across the
cortex of a GCaMP6f-tTA-dCre transgenic mouse reveals neuropil and single cell Ca2+ activity.
Note that due to the curvature of the brain, it was not possible to focus upon cortical layer 2/3
simultaneously across all of neocortex. Hence, active neurons are apparent in some cortical
regions, but other regions are out-of-focus and appear blurrier. Playback speed is 10× real-time.
The video was taken 8 weeks after implantation of the curved glass window. Scale bar: 1 mm.
Supplemental Movie 2 | Epi-fluorescence Ca2+ imaging through the Crystal Skull over the
retrospenial area of a GCaMP6f-tTA-dCre transgenic mouse reveals both neuropil and single
cell Ca2+ activity. As this video has a higher magnification than Movie 1, a higher density of
neurons is visible than in the lower magnification recording. Due to the curvature of the brain, it
was not possible to focus upon cortical layer 2/3 simultaneously across all of the tissue shown.
Hence, active neurons are apparent in some areas, but other areas are out-of-focus and appear
blurrier. Playback speed is 10× real-time. The video was taken 8 weeks after implantation of the
curved glass window. Scale bar: 0.5 mm.
Supplemental Movie 3 | An epi-fluorescence video of Ca2+ activity in the same mouse as in
Movie 2, but acquired three weeks after Movie 2, at 11 weeks after implantation of the curved
glass window. Left, Raw epi-fluorescence Ca2+ video data across a field of view approximately
4 mm × 3 mm in size. The square box encloses the 1.3 mm × 1.3 mm sub-region shown in the
right panel. Scale bar is 0.5 mm. Right, Video of the relative changes in fluorescence (ΔF/F)
across the boxed region of the left panel, highlighting the somatic Ca2+ activity of single
neurons. Playback speed is 3× real-time.
Access to a million mouse neurons in vivo Kim et al., 2016
p. S10
Supplemental Movie 4 | Two-photon Ca2+ imaging through the Crystal Skull in a tetO-
GCaMP6s/CaMK2a-ttA transgenic mouse, from the red region-of-interest marked in Figure 4C.
From left to right, the individual Ca2+ videos were acquired at 100 µm, 200 µm and 350 µm
beneath the cortical surface. Playback speed is 6× real-time. Scale bar: 100 µm.
Access to a million mouse neurons in vivo Kim et al., 2016
p. S11
Supplemental Reference
Paxinos, G., and Franklin, K. (2012). The Mouse Brain in Stereotaxic Coordinates, 4th edition,
(San Diego: Academic Press).
top related