Neural Stem Cell Grafts Form Extensive Synaptic Networks that Integrate with Host Circuits after Spinal Cord Injury.

RESOURCE AVAILABILITY Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Steven Ceto ( steven.ceto@epfl.ch ).

Materials Availability

This study did not generate unique reagents.

Data and Code Availability

The datasets supporting this study are available from the corresponding author upon request. The code (CalFDR) for performing the first derivative rasterization of calcium fluorescence traces within the framework of the Romano et al. toolbox is available at https://github.com/slceto/CalFDR .

EXPERIMENTAL MODEL AND SUBJECT DETAILS Mice

Six- to eight-week-old (age at time of spinal cord injury) wild-type C57BL/6J female mice (15-25 g, n = 56) were subjects of this study. Subjects were not involved in previous procedures, and they were drug and test naïve. To generate E12 embryos for neural progenitor cell harvesting, Syn1-Cre ( Zhu et al., 2001 ) males (8 weeks to one year old) were bred to either wild-type C57BL/6J or Ai95D (floxed-STOP GCaMP6f) ( Madisen et al., 2015 ) females (8 weeks to 6 months old). Syn1-Cre x Ai95D embryos were examined for green fluorescence, and only embryos with central nervous system-specific fluorescence were used. All procedures were carried out in accordance with National Institutes of Health (NIH) guidelines for laboratory animal care and safety and were approved by the Veterans Administration San Diego Healthcare System Institutional Animal Care and Use Committee (IACUC). Animals had free access to food and water throughout the study, and they were housed with littermates except after in vivo imaging plate mounting, after which they were singly housed to avoid damage to the preparation and prevent infections. Surgeries were performed under deep anesthesia using a combination (12.5 ml/kg) of ketamine (2.5 mg/ml), xylazine (0.13 g/ml), and acepromazine (0.025 mg/ml); and 1% isoflurane. After surgery, animals received subcutaneous injections of analgesics (banamine) and antibiotics (ampicillin) in 0.3 mL of lactated ringer’s solution. The health of animals was monitored daily for at least ten days after surgery, after which weekly health checks were performed.

METHOD DETAILS Chrimson Virus Cortical Injection

Neonatal pups (P0-P1) were anesthetized via hypothermia on a wet towel on ice and placed on an ice-cold stereotax. The scalp was rubbed with 70% ethanol before and after the injection procedure. Pups were injected with 0.5 μL of AAV1-Syn-ChrimsonR-tdTomato (1 x 10 12 vg/mL) in each hemisphere of the motor cortex using a 34 gauge, 0.375” needle with a 12° tip in a Hamilton syringe attached to syringe pump (NanoJet, Chemyx Inc). The narrow gauge and sharp tip of the needle allowed it to penetrate directly into the cortex through the scalp and skull. An injection rate of 0.75 μL/min was used, and 30 seconds were allowed before removing the needle after the injection completed. Pups recovered on a heat pad until pink and wriggling before being rubbed with bedding and feces from their home cage and returned to the mother.

Cell Preparation

Neural progenitor cells were prepared for grafting from embryonic day 12.5 (E12.5) embryos following a procedure based on previously described methods ( Adler et al., 2017 ). Embryonic spinal cords were dissected in ice-cold Hank’s Balanced Saline Solution (HBSS), dissociated with 0.05% Trypsin, strained through a 40 μm cell strainer and re-suspended in ice-cold Dulbecco’s Phosphate Buffered Saline (DPBS). Cell viability was assessed by trypan blue exclusion, and aliquots for grafting were prepared immediately before injection at a density of 250,000 cells/μL. In cases where AAV vectors were mixed with graft cells just prior to grafting, virus was added to a final titer of 1.17 x 10 12 vg/mL + 1.65 x 10 12 vg/mL (AAV1- + AAV9-CAG-FLEX-GCaMP6f), 8.3 x 10 11 vg/mL (AAV9-Syn-GCaMP6f), or 4.2 x 10 12 vg/mL (AAV1-Syn-FLEX-ChrimsonR-tdTomato) in the cell suspension. Spinal Cord Lesion and Grafting Dorsal column lesions and acute grafting were performed based on methods previously described ( Adler et al., 2017 ). Briefly, the T12 lamina was removed and a wire knife (McHugh Milieux) was inserted stereotactically to a depth at which the lowest point of the 1 mm-wide extended knife was 1.1 mm below the dorsal surface of the dura mater. Once at this depth, the knife was extended, raised 1.1 mm, and held at this position while the remaining white matter on top of the knife was crushed with a bent insulin syringe needle. The knife was then lowered to its original depth, retracted, and removed from the spinal cord. Neural progenitor cells were immediately grafted using pulled glass micropipettes and a Picospritzer III (Parker Hannifin). Cells were injected directly into the lesion center in a volume of 2 μL DPBS. Because subjects served as their own controls in this study, experimental groups were not randomized, nor experimenters blinded. GCaMP6f Virus Injection AAV9-Syn-GCaMP6f (0.5 μL/site, 1 x 10 12 vg/mL) was injected either directly into the graft center or into the spinal cord gray matter 0.3 mm from midline (bilaterally) over a depth of 1.2 to 0.6 mm with a pulled glass micropipette and Picospritzer.

Live Spinal Cord Slice Preparation

Acute spinal cord slices were prepared following a protocol modified from a previously described procedure ( Husch et al., 2011 ). Animals were placed under deep anesthesia with (28 mL/kg) ketamine (2.5 mg/ml), xylazine (0.13 g/ml), and acepromazine (0.025 mg/ml) and moved to a bed of ice. A block of the spinal column containing the thoracic and lumbar segments of the spinal cord was quickly removed and placed into an icy slurry of oxygenated (95% O 2 –5% CO 2 ) artificial cerebrospinal fluid formulated specifically for dissecting spinal cord (DaCSF) containing (in mM): 2 kynurenic acid, 191 sucrose, 0.75 K-gluconate, 1.25 KH 2 P0 4 , 26 choline bicarbonate 80% solution, 20 dextrose, 1 (+)-sodium L-ascorbate, 5 ethyl pyruvate, 3 myo-inositol, 4 MgSO 4 , and 1 CaCl 2 (~310 mosmol/kgH 2 O, pH 7.35). The animal was then decapitated with scissors. After transferring the spinal block to a dissecting dish with an icy slurry of DaCSF with constant oxygenation, the spinal cord was dissected from the column, with special care taken to cut any scar tissue stuck between the cord and the laminae and attached tissue without pulling on the graft. Due to the presence of scar tissue over the graft site, we did not remove the dura mater prior to slicing. The spinal cord was then transferred to a custom-built chamber filled with 34°C low melting point agarose (0.03 g/ml in DaCSF; A0701; Sigma), and the agarose was allowed to gel on ice with the cord positioned with the lateral aspect facing up. Once gelling completed, the chamber was transferred to an icy DaCSF slurry and the agarose block was trimmed and transferred to the vibratome buffer tray containing continuously oxygenated ice-cold DaCSF. Slices (330 μm thick) were then prepared in the sagittal plane using a Leica VT1000S vibratome and placed with a fine paint brush into a recovery chamber where they were submerged in 34°C oxygenated DaCSF for 30 min. Slices were then transferred to 34°C oxygenated recording buffer (RaCSF) containing (in mM): 121 NaCl, 3 KCl, 1.25 NaH 2 PO 4 , 24 NaHCO 3 , 1.1 MgCl 2 , 2.2 CaCl 2 , 15 dextrose, 1 (+)-sodium L-ascorbate, 5 ethyl pyruvate, and 3 myo-inositol (~310 mosmol/kgH 2 O, pH 7.35) for 30 min. The RaCSF recording chamber was then transferred to room temperature, where the slices remained submerged under oxygenation until transfer to the recording chamber.

Slice Imaging

Spinal cord slices were transferred to a submersion type chamber and perfused with oxygenated RaCSF (with or without 100 μM 4-AP) at 34°C. Ten minutes of recovery time was allowed in the recording chamber prior to imaging. Grafts were identified by fluorescence and/or morphology under infrared differential interference contrast (IR-DIC). Corticospinal axon regeneration or graft axon extension were briefly assessed by ChrimsonR-tdTomato fluorescence under a mercury lamp (U-RFL-T, Olympus), after which only low power LED excitation of GCaMP6f was used for imaging to avoid unintentional activation of Chrimson molecules. The slice imaging rig consisted of (in brief) a widefield fluorescence microscope (BX51WI, Olympus), a 10X NA 0.3 water immersion objective (UMPLFLN10X/W, Olympus), blue (470 nm, 750 mW) and orange (617 nm, 650 mW) mounted LEDs (Thorlabs), two 1200 mA LED drivers (Thorlabs), two NA 0.6 aspheric condenser lenses (ACL2520U-A, Thorlabs), 470 nm and 620 nm excitation filters (ET470/40X and AT620/50X, Chroma), a longpass dichroic mirror with 490 nm cutoff (DMLP490R, Thorlabs), a custom filter cube containing a 530 nm dichroic bandpass and 525 nm emission filter (zt530/55dcbp and ET525/50m, Chroma), and a Rolera XR Fast 1394 or Retiga Electro CCD camera (QImaging). Resolution was 348 x 260 pixels with the Rolera XR and 688 x 512 pixels with the Retiga Electro, both with 2 x 2 binning. Blue LED excitation and orange LED stimulation were controlled with a MultiClamp 700b patch-clamp amplifier via a Digidata 1440A digitizer using Clampex software (all from Molecular Devices). Still and video images (33 frames per second) were acquired with Micro-Manager software ( Edelstein et al., 2014 ). Blue excitation irradiance at the slice was kept low at 0.093 mW/mm 2 to avoid off-target Chrimson stimulation, and orange stimulation irradiance was 4.49 mW/mm 2 at maximum current. In Vivo Imaging Hardware Implantation The week prior to imaging, spinal stabilization hardware was implanted as previously described ( Farrar et al., 2012 ). Two magnetic stainless steel bars were placed against the lateral aspect of the spine, centered around the graft site at the T12 spinal level. A stainless steel plate with an opening over the spinal cord was screwed onto the bars and cyanoacrylate glue and dental cement were used to secure the hardware and attach the skin to it to seal the tissue from the environment. On the day of imaging, glue was removed from the surface of the spinal cord. Typically, the dura mater was resected due to scar tissue from the injury preventing clear imaging. A No. 0 glass coverslip was placed over the exposed cord with the graft in the center. Light pressure was applied to the coverslip while it was glued in place to maintain a flush interface with the spinal cord, and the glue was allowed to dry for one hour prior to imaging. In Vivo Imaging All in vivo imaging experiments were acute, terminal procedures. Mice were placed under gas anesthesia with 1.75% isoflurane and restrained at the spine with two rods attached to the spinal plate. They were then placed under an upright two-photon microscope (Sutter Instrument Company) equipped with an 8-kHz-resonant scanner (Cambridge Technology, Inc.), a pulsed femtosecond Ti:Sapphire laser (Chameleon Vision II, Coherent), a T565LPXR beam splitter (Chroma), ET525/70M and ET605/70M emission filters (Chroma), two GaAsP photomultiplier tubes (H10770PA-40 MOD; Hamamatsu) and a 16X 0.8 NA water-immersion objective (CF175; Nikon).

Typical power used for imaging

GCaMP6f in graft cells was 20-30 mW. Image resolution was 512 x 512 pixels at 30.95 frames per second. Pinch stimuli were delivered with a rodent pincher system (2450, IITC Life Science, Inc.) and pressure sensor output was recorded using MCS software (Sutter Instrument Company; sampling rate, 1 kHz). Light touch and hindlimb movement through range of motion were applied manually by the experimenter over predetermined imaging frame acquisition numbers.

Tissue Processing and Immunofluorescence

After imaging, slices were transferred to cold DaCSF and stored at 4°C until the end of the day and then transferred to 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS). Slices were fixed for 20 minutes to overnight at 4°C and transferred to 30% sucrose in PBS for long-term storage. Following in vivo imaging experiments, animals were killed in their home cage using CO 2 asphyxiation at a 20% fill rate, in accordance with IACUC guidelines. Animals were then transcardially perfused with PBS followed by PFA. The spinal column was dissected and placed in PFA at 4°C overnight, then transferred to 30% sucrose in PBS for long-term storage. Slices from ex vivo experiments were either sectioned at 15 μm on a cryostat (CM1950, Leica) and direct-mounted or stained unsectioned, free-floating. Spinal cords from in vivo experiments as well as whole brains were sectioned at 30 μm on a cryostat and stained free-floating. Standard staining techniques were used with primary antibodies against NeuN (guinea pig from Millipore, abn90, at 1:1000 to label neurons); green fluorescent protein (GFP, chicken from Aves, GFP-1020, or rabbit from Life Technologies, A6455, at 1:1000 to label GCaMP6f-expressing cells); and mCherry (goat from Sicgen, AB0040, at 1:1000 to label ChrimsonR-tdTomato). Secondary antibodies were all Alexa Fluor conjugates from donkey and included guinea pig 488 (Jackson Immunoresearch (JI), 706-545-148); goat 555 (Life Technologies (LT), A21432); chicken 488 (JI, 703-545-155); guinea pig 647 (JI, 706-605-148); and rabbit 488 (Invitrogen, A21206). For direct mount staining, primary and secondary antibody concentrations were quadrupled. Stained sections were imaged as 1-2 μm increment z-stacks with a 10X or 20X objective on an automated all-in-one widefield microscope (BZ-X710, Keyence). Composite images were automatically stitched in the X-Y plane using the Full Focus setting with the Advanced Observation – Image Stitching Module of the Keyence software.

QUANTIFICATION AND STATISTICAL ANALYSIS Image Data Processing and Analysis

Slice imaging data were acquired as TIFF stacks and did not require motion correction because slices were held in place by a nylon harp. As imaging data acquisition began before Clampex protocols for excitation and stimulation were initiated, the frame at which the protocol began could be easily distinguished as the first frame with detectable fluorescence signal. A rapid period of photobleaching occurred as soon as the blue LED was turned on, so the first ten seconds after protocol initiation in each image stack were discarded to establish a smoother baseline fluorescence. When quantifying response amplitude or latency, image stacks were unfiltered. When quantifying activity dynamics, stacks were Kalman filtered using ImageJ software ( Schneider et al., 2012 ). Regions of interest (ROIs) were manually drawn on the standard deviation projection of the stack around areas of fluorescence with neuronal morphology and dynamic fluorescence using Fiji ( Schindelin et al., 2012 ). These ROIs were converted to a mask and saved as a text image before being converted to a binary matrix in Matlab (Mathworks) with ones designating pixels belonging to ROIs.

Image stacks with accompanying

ROIs were then processed and analyzed with Matlab routines modified from a previously published calcium imaging toolbox ( Romano et al., 2017 ). Briefly, background fluorescence from neuropil surrounding each ROI but not including neighboring ROIs was scaled by a factor of 0.9 and subtracted from the ROI’s fluorescence before the change in fluorescence over a slowly varying baseline (ΔF/F) was calculated for each ROI over the course of the imaging session. To quantify response strength, we took the mean ΔF/F over a 1 s period following stimulus onset. Response latency was defined as the temporal midpoint to peak fluorescence following stimulus onset, which corresponds to a time midway through the calcium spike. Significant fluorescence transients were identified by their exceeding a threshold that was dynamically calculated based on the scale of the noise of the ROI’s fluorescence as well as the calcium reporter’s decay time constant. We then deviated from the Romano toolbox and took the first derivative of the ΔF/F trace, d(ΔF/F)/t, during these periods of significant fluorescence and subtracted 2.5 x (the standard deviation of d(ΔF/F)/t during non-significant periods). Those imaging frames with positive d(ΔF/F)/t values after noise subtraction were counted as periods during which true spiking activity took place within the cell, since the fluorescence decay period is not generally associated with activity ( Carrillo-Reid et al., 2008 ). This data was used to generate raster plots of ROI activity. Clusters of cells with similar activity dynamics were identified with scripts from the Romano toolbox based on the first derivative rasterization described above. Briefly, principal components analysis (PCA) was run on the z-scored activity of each ROI, and non-orthogonal factor rotation, or promax ( Hendrickson and White, 1964 ) was applied to allow for non-exclusive cell assemblies, or clusters. A cell was designated as belonging to a cluster based on its z-scored loading on the principal component defining that cluster. A matching index was then used to quantify significant activations of the clusters over the imaging session, with the threshold p-value set at < 0.05 ( Romano et al., 2017 ). In vivo imaging data was converted from MDF format to TIFF stacks and corrected for lateral motion artifacts using TurboReg ( Thévenaz et al., 1998 ). Uncut image stacks were Kalman filtered and ROIs were drawn around areas with dynamic fluorescence. For in vivo imaging only, ROIs included cell processes as well as soma, potentially allowing different regions of the same cell to be quantified separately. ΔF/F traces were quantified as for slice image stacks, but with 0.4-factor or no background neuropil subtraction.

Statistics

Statistical details of individual experiments may be found in the figure legends and Results section. Two-group comparisons were tested by Welch’s t-test using Matlab software. Significance values were set at p = 0.05 (*) and p = 0.01 (**). Data are presented as mean ±SEM. Animals were included in analyses only if the majority of the host-graft interface was uninterrupted by cavitation or scarring and consistent spontaneous calcium activity was observed (in slice experiments) or if any consistent calcium activity (spontaneous or evoked) could be observed (in vivo experiments).

Materials Availability

This study did not generate unique reagents.

EXPERIMENTAL MODEL AND SUBJECT DETAILS Mice

Six- to eight-week-old (age at time of spinal cord injury) wild-type C57BL/6J female mice (15-25 g, n = 56) were subjects of this study. Subjects were not involved in previous procedures, and they were drug and test naïve. To generate E12 embryos for neural progenitor cell harvesting, Syn1-Cre ( Zhu et al., 2001 ) males (8 weeks to one year old) were bred to either wild-type C57BL/6J or Ai95D (floxed-STOP GCaMP6f) ( Madisen et al., 2015 ) females (8 weeks to 6 months old). Syn1-Cre x Ai95D embryos were examined for green fluorescence, and only embryos with central nervous system-specific fluorescence were used. All procedures were carried out in accordance with National Institutes of Health (NIH) guidelines for laboratory animal care and safety and were approved by the Veterans Administration San Diego Healthcare System Institutional Animal Care and Use Committee (IACUC). Animals had free access to food and water throughout the study, and they were housed with littermates except after in vivo imaging plate mounting, after which they were singly housed to avoid damage to the preparation and prevent infections. Surgeries were performed under deep anesthesia using a combination (12.5 ml/kg) of ketamine (2.5 mg/ml), xylazine (0.13 g/ml), and acepromazine (0.025 mg/ml); and 1% isoflurane. After surgery, animals received subcutaneous injections of analgesics (banamine) and antibiotics (ampicillin) in 0.3 mL of lactated ringer’s solution. The health of animals was monitored daily for at least ten days after surgery, after which weekly health checks were performed.

METHOD DETAILS Chrimson Virus Cortical Injection

Neonatal pups (P0-P1) were anesthetized via hypothermia on a wet towel on ice and placed on an ice-cold stereotax. The scalp was rubbed with 70% ethanol before and after the injection procedure. Pups were injected with 0.5 μL of AAV1-Syn-ChrimsonR-tdTomato (1 x 10 12 vg/mL) in each hemisphere of the motor cortex using a 34 gauge, 0.375” needle with a 12° tip in a Hamilton syringe attached to syringe pump (NanoJet, Chemyx Inc). The narrow gauge and sharp tip of the needle allowed it to penetrate directly into the cortex through the scalp and skull. An injection rate of 0.75 μL/min was used, and 30 seconds were allowed before removing the needle after the injection completed. Pups recovered on a heat pad until pink and wriggling before being rubbed with bedding and feces from their home cage and returned to the mother.

Cell Preparation

Neural progenitor cells were prepared for grafting from embryonic day 12.5 (E12.5) embryos following a procedure based on previously described methods ( Adler et al., 2017 ). Embryonic spinal cords were dissected in ice-cold Hank’s Balanced Saline Solution (HBSS), dissociated with 0.05% Trypsin, strained through a 40 μm cell strainer and re-suspended in ice-cold Dulbecco’s Phosphate Buffered Saline (DPBS). Cell viability was assessed by trypan blue exclusion, and aliquots for grafting were prepared immediately before injection at a density of 250,000 cells/μL. In cases where AAV vectors were mixed with graft cells just prior to grafting, virus was added to a final titer of 1.17 x 10 12 vg/mL + 1.65 x 10 12 vg/mL (AAV1- + AAV9-CAG-FLEX-GCaMP6f), 8.3 x 10 11 vg/mL (AAV9-Syn-GCaMP6f), or 4.2 x 10 12 vg/mL (AAV1-Syn-FLEX-ChrimsonR-tdTomato) in the cell suspension. Spinal Cord Lesion and Grafting Dorsal column lesions and acute grafting were performed based on methods previously described ( Adler et al., 2017 ). Briefly, the T12 lamina was removed and a wire knife (McHugh Milieux) was inserted stereotactically to a depth at which the lowest point of the 1 mm-wide extended knife was 1.1 mm below the dorsal surface of the dura mater. Once at this depth, the knife was extended, raised 1.1 mm, and held at this position while the remaining white matter on top of the knife was crushed with a bent insulin syringe needle. The knife was then lowered to its original depth, retracted, and removed from the spinal cord. Neural progenitor cells were immediately grafted using pulled glass micropipettes and a Picospritzer III (Parker Hannifin). Cells were injected directly into the lesion center in a volume of 2 μL DPBS. Because subjects served as their own controls in this study, experimental groups were not randomized, nor experimenters blinded. GCaMP6f Virus Injection AAV9-Syn-GCaMP6f (0.5 μL/site, 1 x 10 12 vg/mL) was injected either directly into the graft center or into the spinal cord gray matter 0.3 mm from midline (bilaterally) over a depth of 1.2 to 0.6 mm with a pulled glass micropipette and Picospritzer.

Live Spinal Cord Slice Preparation

Acute spinal cord slices were prepared following a protocol modified from a previously described procedure ( Husch et al., 2011 ). Animals were placed under deep anesthesia with (28 mL/kg) ketamine (2.5 mg/ml), xylazine (0.13 g/ml), and acepromazine (0.025 mg/ml) and moved to a bed of ice. A block of the spinal column containing the thoracic and lumbar segments of the spinal cord was quickly removed and placed into an icy slurry of oxygenated (95% O 2 –5% CO 2 ) artificial cerebrospinal fluid formulated specifically for dissecting spinal cord (DaCSF) containing (in mM): 2 kynurenic acid, 191 sucrose, 0.75 K-gluconate, 1.25 KH 2 P0 4 , 26 choline bicarbonate 80% solution, 20 dextrose, 1 (+)-sodium L-ascorbate, 5 ethyl pyruvate, 3 myo-inositol, 4 MgSO 4 , and 1 CaCl 2 (~310 mosmol/kgH 2 O, pH 7.35). The animal was then decapitated with scissors. After transferring the spinal block to a dissecting dish with an icy slurry of DaCSF with constant oxygenation, the spinal cord was dissected from the column, with special care taken to cut any scar tissue stuck between the cord and the laminae and attached tissue without pulling on the graft. Due to the presence of scar tissue over the graft site, we did not remove the dura mater prior to slicing. The spinal cord was then transferred to a custom-built chamber filled with 34°C low melting point agarose (0.03 g/ml in DaCSF; A0701; Sigma), and the agarose was allowed to gel on ice with the cord positioned with the lateral aspect facing up. Once gelling completed, the chamber was transferred to an icy DaCSF slurry and the agarose block was trimmed and transferred to the vibratome buffer tray containing continuously oxygenated ice-cold DaCSF. Slices (330 μm thick) were then prepared in the sagittal plane using a Leica VT1000S vibratome and placed with a fine paint brush into a recovery chamber where they were submerged in 34°C oxygenated DaCSF for 30 min. Slices were then transferred to 34°C oxygenated recording buffer (RaCSF) containing (in mM): 121 NaCl, 3 KCl, 1.25 NaH 2 PO 4 , 24 NaHCO 3 , 1.1 MgCl 2 , 2.2 CaCl 2 , 15 dextrose, 1 (+)-sodium L-ascorbate, 5 ethyl pyruvate, and 3 myo-inositol (~310 mosmol/kgH 2 O, pH 7.35) for 30 min. The RaCSF recording chamber was then transferred to room temperature, where the slices remained submerged under oxygenation until transfer to the recording chamber.

Slice Imaging

Spinal cord slices were transferred to a submersion type chamber and perfused with oxygenated RaCSF (with or without 100 μM 4-AP) at 34°C. Ten minutes of recovery time was allowed in the recording chamber prior to imaging. Grafts were identified by fluorescence and/or morphology under infrared differential interference contrast (IR-DIC). Corticospinal axon regeneration or graft axon extension were briefly assessed by ChrimsonR-tdTomato fluorescence under a mercury lamp (U-RFL-T, Olympus), after which only low power LED excitation of GCaMP6f was used for imaging to avoid unintentional activation of Chrimson molecules. The slice imaging rig consisted of (in brief) a widefield fluorescence microscope (BX51WI, Olympus), a 10X NA 0.3 water immersion objective (UMPLFLN10X/W, Olympus), blue (470 nm, 750 mW) and orange (617 nm, 650 mW) mounted LEDs (Thorlabs), two 1200 mA LED drivers (Thorlabs), two NA 0.6 aspheric condenser lenses (ACL2520U-A, Thorlabs), 470 nm and 620 nm excitation filters (ET470/40X and AT620/50X, Chroma), a longpass dichroic mirror with 490 nm cutoff (DMLP490R, Thorlabs), a custom filter cube containing a 530 nm dichroic bandpass and 525 nm emission filter (zt530/55dcbp and ET525/50m, Chroma), and a Rolera XR Fast 1394 or Retiga Electro CCD camera (QImaging). Resolution was 348 x 260 pixels with the Rolera XR and 688 x 512 pixels with the Retiga Electro, both with 2 x 2 binning. Blue LED excitation and orange LED stimulation were controlled with a MultiClamp 700b patch-clamp amplifier via a Digidata 1440A digitizer using Clampex software (all from Molecular Devices). Still and video images (33 frames per second) were acquired with Micro-Manager software ( Edelstein et al., 2014 ). Blue excitation irradiance at the slice was kept low at 0.093 mW/mm 2 to avoid off-target Chrimson stimulation, and orange stimulation irradiance was 4.49 mW/mm 2 at maximum current. In Vivo Imaging Hardware Implantation The week prior to imaging, spinal stabilization hardware was implanted as previously described ( Farrar et al., 2012 ). Two magnetic stainless steel bars were placed against the lateral aspect of the spine, centered around the graft site at the T12 spinal level. A stainless steel plate with an opening over the spinal cord was screwed onto the bars and cyanoacrylate glue and dental cement were used to secure the hardware and attach the skin to it to seal the tissue from the environment. On the day of imaging, glue was removed from the surface of the spinal cord. Typically, the dura mater was resected due to scar tissue from the injury preventing clear imaging. A No. 0 glass coverslip was placed over the exposed cord with the graft in the center. Light pressure was applied to the coverslip while it was glued in place to maintain a flush interface with the spinal cord, and the glue was allowed to dry for one hour prior to imaging. In Vivo Imaging All in vivo imaging experiments were acute, terminal procedures. Mice were placed under gas anesthesia with 1.75% isoflurane and restrained at the spine with two rods attached to the spinal plate. They were then placed under an upright two-photon microscope (Sutter Instrument Company) equipped with an 8-kHz-resonant scanner (Cambridge Technology, Inc.), a pulsed femtosecond Ti:Sapphire laser (Chameleon Vision II, Coherent), a T565LPXR beam splitter (Chroma), ET525/70M and ET605/70M emission filters (Chroma), two GaAsP photomultiplier tubes (H10770PA-40 MOD; Hamamatsu) and a 16X 0.8 NA water-immersion objective (CF175; Nikon).

Typical power used for imaging

GCaMP6f in graft cells was 20-30 mW. Image resolution was 512 x 512 pixels at 30.95 frames per second. Pinch stimuli were delivered with a rodent pincher system (2450, IITC Life Science, Inc.) and pressure sensor output was recorded using MCS software (Sutter Instrument Company; sampling rate, 1 kHz). Light touch and hindlimb movement through range of motion were applied manually by the experimenter over predetermined imaging frame acquisition numbers.

Tissue Processing and Immunofluorescence

After imaging, slices were transferred to cold DaCSF and stored at 4°C until the end of the day and then transferred to 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS). Slices were fixed for 20 minutes to overnight at 4°C and transferred to 30% sucrose in PBS for long-term storage. Following in vivo imaging experiments, animals were killed in their home cage using CO 2 asphyxiation at a 20% fill rate, in accordance with IACUC guidelines. Animals were then transcardially perfused with PBS followed by PFA. The spinal column was dissected and placed in PFA at 4°C overnight, then transferred to 30% sucrose in PBS for long-term storage. Slices from ex vivo experiments were either sectioned at 15 μm on a cryostat (CM1950, Leica) and direct-mounted or stained unsectioned, free-floating. Spinal cords from in vivo experiments as well as whole brains were sectioned at 30 μm on a cryostat and stained free-floating. Standard staining techniques were used with primary antibodies against NeuN (guinea pig from Millipore, abn90, at 1:1000 to label neurons); green fluorescent protein (GFP, chicken from Aves, GFP-1020, or rabbit from Life Technologies, A6455, at 1:1000 to label GCaMP6f-expressing cells); and mCherry (goat from Sicgen, AB0040, at 1:1000 to label ChrimsonR-tdTomato). Secondary antibodies were all Alexa Fluor conjugates from donkey and included guinea pig 488 (Jackson Immunoresearch (JI), 706-545-148); goat 555 (Life Technologies (LT), A21432); chicken 488 (JI, 703-545-155); guinea pig 647 (JI, 706-605-148); and rabbit 488 (Invitrogen, A21206). For direct mount staining, primary and secondary antibody concentrations were quadrupled. Stained sections were imaged as 1-2 μm increment z-stacks with a 10X or 20X objective on an automated all-in-one widefield microscope (BZ-X710, Keyence). Composite images were automatically stitched in the X-Y plane using the Full Focus setting with the Advanced Observation – Image Stitching Module of the Keyence software.

Supplementary Material 1 Document S1. Figures S1-S6. 2 Video S1: Graft response to optogenetic corticospinal stimulation, Related to Figure 4 . 3 Video S2: Host response to optogenetic graft stimulation, Related to Figure 5 . 4 Video S3: In vivo graft response to light tough, Related to Figure 6 .