Experimental design and statistical analysis
The experiments were designed as pre-clinical trials to test the effect of iNSC transplantation during chronic CPZ demyelination. Experiments were carried out according to the Animal Research Reporting: In Vivo Experiments (ARRIVE) guidelines. Technique validation was performed using pilot cohorts. Two parallel trial experiments each used a non-invasive assessment, MRI (structural) or behavior (functional), as the primary outcome measure. A pre-determined study design was used that stated inclusion/exclusion criteria, randomization procedures, and statistical analyses. All experiments were performed with blinding during implementation and analysis. Sample size estimates were based on prior experiments using each primary outcome measure (MRI or behavior). To reduce variability due to the demyelination produced by CPZ, mice were split into groups balanced by weights and assigned to treatment condition using the Excel RAND command for randomization. After completion of the primary outcome assessment, mice in each experiment were perfused for neuropathology and cell type analyses.
Timelines for the mouse cohorts are shown in the corresponding figures and mouse sample numbers are provided in each legend. A total of 100 mice were used across the experiments in the study. Only male mice were used since CPZ toxicity effects the estrus cycle so that sex cannot be appropriately analyzed as a biological variable [36]. Statistical analysis and graphing was performed using GraphPad Prism software version 8.0 (RRID: SCR_002798). The values are shown as the mean ± standard error of the mean (sem). Statistical significance was determined as p < 0.05. The experimental design and statistical analysis for each section is provided below with the technical methods for that study.
Mice and cuprizone model of demyelination
Mice were housed and cared for in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The study protocol was approved by the Institutional Animal Care and Use Committee of the Uniformed Services University of the Health Sciences. Mice were socially housed in 27 cm × 16.5 cm × 12.5 cm cages (2–5 mice per cage) with enrichment objects and maintained on a standard 12 h cycle of daytime light (6:00–18:00). All procedures took place during the daytime light cycle.
C57BL/6 J mice (RRID:IMSR_JAX:000664, Jackson Laboratory, Bar Harbor, MA) at 8 weeks of age were fed pellets containing 0.3% CPZ for 6 weeks or 12 weeks to produce acute or chronic demyelination, respectively. Milled CPZ powder (cat #14690; Sigma-Aldrich; St. Louis, MO) was mixed as 0.3% CPZ in normal chow (diet TD.01453; Harlan Teklad, Frederick, MD) to form pellets that were refreshed every 2–3 days. A dose of 0.3% CPZ in pellets produces CC demyelination approximately equivalent to 0.2% CPZ powder in ground chow [37,38,39]. Cohorts to be evaluated for remyelination received normal chow pellets for an additional 2 weeks after CPZ removal. Naïve mice continuously fed normal chow pellets served as a control non-demyelinated condition performed simultaneously along with CPZ treated mice. Baseline weights at the start of CPZ or normal chow feeding were similar within each cohort: MRI technical validation cohort (mean ± sem = naïve 25.25 ± 0.47 g; CPZ 25.25 ± 0.43 g), MRI iNSC transplant cohort (CPZ veh 26.24 ± 0.42 g; CPZ iNSC 26.33 ± 0.37 g), wheels iNSC transplant cohort (naïve veh 22.87 ± 0.40 g; CPZ veh 23.76 ± 0.56 g; CPZ iNSC 23.76 ± 0.61 g). Mice remained in good health and regained baseline weights by the end of chronic CPZ administration.
Magnetic resonance imaging (MRI)
In vivo MRI was performed on a 7 T small animal (20 cm bore) Bruker BioSpec scanner equipped with 12 cm diameter 650 mT/m gradient coils (Bruker BioSpin GmbH, Reinstetten, Germany). MRI with T2 weighting, magnetization transfer ratio (MTR), and diffusion tensor imaging (DTI) was used for longitudinal in vivo analysis of the effects of chronic CPZ ingestion and subsequent iNSC transplantation in the CC. Prior to image acquisition, mice were separated into yoked pairs of similar weight. During each image acquisition, mice were anaesthetized with 1% isoflurane. MRI slices were established using a sagittal localizer so that seven coronal slices were orientated perpendicular to the length of the CC and each scan was aligned with the midline crossing of the anterior commissure positioned in the same coronal slice [40,41,42].
For DTI [43], a three-dimensional (3D) single-shot echo planar imaging sequence (repetition time/echo time [TR/TE] = 900/36 msec; 1 repetition) was used to acquire 4 unweighted (b = 0 s/mm2) and 2 diffusion-weighted images (b = 600, 1200 s/mm2) in 14 noncollinear diffusion gradient directions using a Stejskal-Tanner diffusion preparation with parameters of Δ = 12 msec and δ = 5 msec, field of view (FOV) = 14 × 11.2 mm2, matrix = 80 × 64 × 24, slice thickness = 750 μm, 24 slices, voxels = 175 × 175 × 750 μm3. A whole brain T2 map was generated using a two-dimensional rapid acquisition with relaxation enhancement (2D RARE, corona [44]) with the following parameters: TR = 4000 msec; TE = 10, 30, 50, 70, 90, 110 msec; RARE factor = 2, number of averages (NA) = 4, FOV = 14 × 12 mm2, matrix = 112 × 96 × 18, slice thickness = 750 μm, 18 slices, voxels = 125 × 125 × 750 μm3. For use in calculating the magnetization transfer ratio (MTR) [45, 46], 2D RARE sequence was used with TR = 6000 msec, TE = 20 msec, RARE factor = 8, NA = 4, FOV = 14 × 12 mm2, matrix = 112 × 96 × 18, voxels = 125 × 125 × 750 μm3, with saturation (Ms) and without (Mo) using an offset saturation pulse (2 kHz offset 66.7° flip angle). Body temperature was maintained at 36 °C by circulating hot water. Respiration and heart rate were monitored throughout each 2-h imaging session.
T2 and MTR maps along with T2 templates were generated from neuroimaging informatics technology initiative (NIFTI) files and analyzed in either VivoQuant (inviCRO, Boston, MA) (chronic CPZ) or MATLAB (MathWorks, Natick, MA)(chronic CPZ + 2 weeks recovery). The CC ROI was defined as extending from the midline bilaterally to the point of ventral curvature in the external capsule, as in our previous studies (Sullivan et al. 2013 [47]; Yu et al. 2017 [42]). For each mouse, an initial ROI was propagated across images using the T2 templates and the voxel placement checked/corrected manually for each slice. The rostrocaudal extent of the CC was contained within 7 slices. Slices 1 and 7 were omitted to avoid regions of varying fiber directions toward the genu and splenium. Slices 2–6 (approximately 1 mm to − 2.5 mm relative to bregma) were combined as the full CC ROI. DTI data was analyzed with TORTOISE software [48, 49]. Diffusion-weighted imaging artifacts, including motion, eddy currents, and concomitant field distortions, were calculated and combined to enable the image correction to be applied in a single step. Fractional anisotropy (FA), Trace, and axial (AD) and radial (RD) diffusivity maps were computed using a nonlinear tensor estimation with RESTORE option [50]. Diffusion direction-encoded color (DEC) maps were generated from the TORTOISE software to demonstrate fiber orientation. The CC ROIs were overlaid onto the DTI maps for further quantitative analysis.
Technical validation for neuroimaging approach
An MRI study with post-imaging neuropathology evaluated the MRI techniques for detection of chronic demyelination at the end of the chronic CPZ period, when iNSCs will be transplanted in the subsequent experiments. This longitudinal design compared naïve mice (n = 6) that received normal chow pellets continuously with CPZ treated mice (n = 6) at baseline (8 weeks of age), 6 weeks ± CPZ (14 weeks of age), and 12 weeks ± CPZ (20 weeks of age). Two-way ANOVA with repeated measures was used to compare naïve vs. CPZ at each time point, with Bonferroni multiple comparisons test.
Neuroimaging analysis of iNSC transplantation effect on CC microstructure
An MRI longitudinal study with post-imaging neuropathology evaluated the effect of iNSC transplantation on CC microstructure during recovery after chronic CPZ. This longitudinal design compared CPZ treated mice injected with vehicle (n = 5, one mouse died) or iNSCs (n = 6). Scans were performed at baseline (8 weeks of age before starting CPZ), 6 weeks CPZ, 12 weeks CPZ, and then mice were returned to normal chow pellets until the final scan at 14 weeks. Mice received iNSC or vehicle intracerebral injections 1 day after the return to normal chow. Mice were scanned in sets of four per day, which were balanced by weights prior to the start of CPZ. After the week scan at 6 weeks of CPZ, mice within each set of four were randomly assigned, using Excel RAND, to receive iNSC or vehicle injection. The T2 values of the assigned groups were checked to ensure similar CPZ effect within the cohorts at 6 weeks. Two-way ANOVA with repeated measures was used for within subjects comparisons to baseline to evaluate CPZ pathology, and for comparison of each time point for iNSC vs. vehicle, with Bonferroni correction for multiple comparisons.
Generation and characterization of mouse iNSCs
Mouse C57BL/6 iNSCs that ubiquitously express farnesylated green fluorescent protein (GFP) were generated as previously detailed [29, 51]. Briefly, iNSCs were produced by direct conversion of mouse embryonic fibroblasts through constitutive expression of Sox2, Klf4, c-Myc, and transient expression of Oct4. iNSCs were then transduced with lentivirus containing a farnesylated GFP construct to target GFP to the inner plasma membrane. Prior to characterization and transplantation experiments, iNSCs were cultured as an adherent monolayer, as detailed previously [29, 51].
For in vitro characterization, iNSCs were cultured as either free floating neurospheres in proliferation medium or seeded onto glass coverslips coated with Matrigel™ (Cat# 356234 Corning, Corning, NY) for growth as adherent monolayers in differentiation medium, as previously detailed [51]. Differentiation medium contained NeuroCults basal medium (Cat# 05700, STEMCELL Technologies, Cambridge, MA) with 10% differentiation supplement (Cat# 05703, STEMCELL Technologies) and 100 units penicillin/0.1 mg streptomycin/mL (Cat# 15140122 Invitrogen, Waltham, MA). For immunocytochemistry, neurosphere cultures were fixed on glass coverslips in 4% paraformaldehyde (PFA, Sigma-Aldrich) 2% sucrose in 1X PBS. Neurospheres were incubated with primary antibodies for neural stem/progenitor cell markers Sox2 (rabbit polyclonal, 1:200; Abcam, Cambridge, MA, ab97959, RRID:AB_2341193) and Nestin (chicken polyclonal, 1:200; Abcam, ab134017, RRID:AB_2753197). Differentiation was examined using immunolabeling to detect markers of either the astroglial lineage (glial fibrillary acidic protein, GFAP; rabbit polyclonal, 1:100; DAKO; Carpinteria, CA, Z0334, RRID:AB_10013382), the neuronal lineage (microtubule associated protein-2, MAP-2, rabbit polyclonal, 1:200; Abcam; ab32454, RRID:AB_776174), or the oligodendroglial lineage (O4 mouse monoclonal, 1:10 [52]). PBS with 0.1% Triton-X 100 and 10% normal goat serum for 1 h was used prior to all primary antibodies, but was not used with O4 to prevent Triton-X 100 degradation of the sulfatide epitope [7, 53]. Secondary antibodies used were goat anti-rabbit IgG conjugated with AF647 (1:200; Thermo Fisher, Waltham, MA; A-21245, RRID:AB_2535813) to detect Sox2, and goat anti-chicken conjugated with AF555 (1:200; Thermo Fisher; A-21437, RRID:AB_2535858) to detect Nestin, donkey anti-rabbit IgG F(ab’)2 conjugated with Cy3 (1:400; Jackson ImmunoResearch, West Grove, PA, 711–166-152, RRID:AB_2313568) to detect GFAP, and goat anti-mouse IgM conjugated with Cy3 (1:50; Jackson ImmunoResearch; 115–166-075, RRID:AB_2338707).
Technical validation for iNSCs preparations
Frozen iNSCs were thawed and cultured for amplification and characterization for each transplantation experiment cohort. An aliquot of each iNSC preparation used for transplantation was analyzed to confirm neural stem cell characteristics. For differentiation analysis, the number of immunolabeled cells was counted in n ≥ 8 non-overlapping fields per sample up to a total of n > 300 cells per aliquot. For growth curve analysis, the total number of cells was quantified after each passage of each subculture.
Cell transplantation into the corpus callosum
For all microinjections into the CC, mice were anaesthetized with isoflurane (induction 3%, maintenance 2%) and a small < 1.0 mm diameter burr hole was drilled into the skull. A Hamilton gas tight syringe (Cat# 7653–01; Hamilton Company, Reno, NV) was used with adapters (Cat# 55750–01; Hamilton Company) and a pulled glass micropipette (outer diameter 50 μm) [54]. Each microinjection targeted the left CC using coordinates (− 1.0 AP, 0.5 ML, − 1.3DV) relative to bregma. An iNSC single cell suspension of 1 × 104 cells in 1 μl of PBS, or PBS as a vehicle control, was injected over a 5 min period.
Neurologic analysis with the miss-step running wheel assay
Starting the day after iNSC transplantation, and continuing for 2 weeks, mice were singly housed in home cages with a Miss-step running wheel and an optical sensor to detect wheel revolutions (Mouse Miss-step Activity Wheel system #80821, Lafayette Instruments, Lafayette, IN). The Miss-step running wheels have 16 rungs missing from a standard wheel so that the remaining 22 rungs are distributed in an irregular interval pattern [33]. Whiskers were clipped so that the mice learn to locate each rung by bringing the hind paw forward to grasp the same rung as a forepaw [35]. Activity Wheel Monitor software (Lafayette Instruments) counted wheel revolutions at 6 min intervals during the light phase and 1 min intervals during the dark phase. Results were exported to a Microsoft Excel file every 24 h.
Technical validation for wheel running effects on oligodendrocyte populations in CC
A study using acute CPZ, which undergoes extensive spontaneous remyelination, tested the effect of the Miss-step wheel exposure on the oligodendrocyte lineage response in the CC. Mice were fed normal chow (naïve, n = 6) or CPZ for 6 weeks to produce acute demyelination (acute CPZ, n = 12). After 6 weeks, all mice were fed normal chow. The mice were then singly housed in home cages with (naïve, n = 6, acute CPZ, n = 6), or without (acute CPZ, n = 6), Miss-step wheels for 2 weeks. Mice were perfused for in situ hybridization. Across the three conditions, cell densities were quantified from at least 3 sections per mouse and analyzed using one-way ANOVA with Tukey’s multiple comparisons test.
Miss-step wheel analysis of iNSC transplantation on CC function
Running on Miss-step wheels, with irregularly spaced rungs, presented a novel motor skill task to evaluate effects of iNSC transplantation on sensorimotor function during recovery after chronic CPZ. Cohorts were maintained under the same conditions for simultaneous automated data collection. The study included four cohorts of 12 mice each, which included naïve with vehicle injection (n = 4), CPZ with vehicle injection (n = 4), and CPZ with iNSC transplant (n = 4). Of the four cohorts run, one cohort was excluded due to the CPZ vehicle condition failing to demonstrate a demyelination associated deficit, compared to naïve. Data from 3 independent cohorts of 12 mice each was combined for analysis of naïve (n = 11), CPZ vehicle (n = 12), and CPZ iNSC (n = 11). Death prior to the experimental endpoint required exclusion of one mouse (CPZ iNSC, n = 1). A mouse (naïve, n = 1) that did not achieve criterion (average velocity of 10 m/min by day 7) during the learning phase was excluded from analysis of the plateau phase. A value missing due to technical errors of the wheel system was imputed based on the mean (n = 1 mouse on 1 d). Velocity measures were compared using repeated measures two-way ANOVA with Sidak’s multiple comparison test. For in situ hybridization analysis in tissues from mice with and without wheel exposure, additional mice (naive veh, n = 5; CPZ veh, n = 5) were housed singly without wheels in parallel with the wheel mice from two different cohorts.
Tissue analysis of neuropathology and cellular responses
After completion of imaging or Miss-step wheel assessments, mice were perfused for tissue analysis. Imaging and wheel running cohorts were separately analyzed. Quantification included at least 3 tissue sections per mouse, unless otherwise noted. The number of mice for each condition is provided with each experiment above and is stated in each figure legend. Images for quantification were acquired using an Olympus IX70 fluorescence microscope with a SPOT RT3 camera (Diagnostic Instruments, Sterling Heights, MI). The CC ROI was defined as extending from the midline bilaterally to the point of ventral curvature in the external capsule [42, 47]. Coronal sections within − 0.5 and − 2.0 mm relative to bregma were used for quantification, to focus in the most demyelinated region and minimize variation [40]. Unpaired t-tests were used to compare two conditions, i.e. naïve vs. CPZ, or veh vs. iNSCs. After the Miss-step wheel running, further analysis of oligodendrocyte populations for CPZ veh vs. CPZ iNSCs used two-way ANOVA with Sidak’s multiple comparisons test.
Immunohistochemistry
Mice were perfused transcardially with 4% PFA, post-fixed overnight at 4 °C, and processed for frozen sectioning. Coronal cryosections (15 μm) were immunostained with primary antibodies to detect either myelin based on myelin oligodendrocyte glycoprotein (MOG, mouse monoclonal, 1:100; EMD Millipore, Burlington, MA; mab5680, RRID:AB_1587278), astrocytes with GFAP (rabbit polyclonal, 1:500; DAKO; Z0334, RRID:AB_10013382), microglia/macrophage with IBA1 (rabbit polyclonal, 1:500; Wako, Richmond, VA; 019–19,741, RRID:AB_839504), oligodendrocyte lineage cells with Olig2 (rabbit polyclonal, 1:500; EMD Millipore, ab9610, RRID:AB_570666), OPCs with NG2 (rabbit polyclonal to ectodomain; generous gift from William Stallcup; RRID:AB_2572306) and neural stem/progenitor cells with Sox2 (rabbit polyclonal, 1:200; Abcam, ab97959, RRID:AB_2341193). Cells undergoing active proliferation were identified by Ki67 (rabbit polyclonal, 1:500; Abcam ab 15,580, RRID:AB_443209). Axons were identified by neurofilament heavy chain protein (NF-H) rabbit polyclonal antibody (1:500; RPCA-NF-H Encor Biotechnology, Gainesville, FL, RRID:AB_2572360) with the proportion with axon damage immunolabeled for SMI32 mouse IgG1 monoclonal antibody, which recognizes nonphosphorylated neurofilaments (1:1000; Sternberger monoclonal SMI32; 801,701 Biolegend, San Diego, CA, RRID:AB_2564642). The secondary antibodies used were donkey anti-rabbit IgG F(ab’)2 (Jackson ImmunoResearch, West Grove, PA) conjugated with Cy3 (1:400 or 1:500; 711–166-152, RRID:AB_2313568) or AF647 (1:100; 711–606-152, RRID:AB_2340625), donkey anti-mouse IgG IgG F(ab’)2 conjugated with Cy3 (1:100; Jackson ImmunoResearch; 715–166-150, RRID:AB_2340816), goat anti-rabbit IgG conjugated with AF647 (1:200; Thermo Fisher; A21245, RRID:AB_2535813) and goat anti-mouse IgG conjugated with AF555 (1:500; Thermo Fisher; A21422, RRID:AB_2535844). Nuclei in all sections were counterstained with DAPI (Sigma-Aldrich). The GFP expression of iNSCs was sufficient for direct detection without using additional antibody detection.
In situ hybridization
In situ hybridization using a proteolipid protein (Plp1) riboprobe was performed as previously described [55, 56]. The Enpp6 plasmid vector (pCMV-SPORT6, gift from Dr. William D. Richardson, University College London) was used to generate the Enpp6 riboprobe (Xiao et al. 2016 [57]). In 15 μm coronal cryosections, hybridized Plp1 or Enpp6 riboprobe was detected with alkaline phosphatase-conjugated sheep anti-digoxigenin antibody and incubation in substrate solution (nitroblue tetrazolium chloride/5–bromo-4–chloro-3–indolyl-phosphate [NBT/BCIP]; Dako).
Quantification details for CC area, myelin, astrogliosis and microglia activation
Immunolabeling within the CC ROI was quantified on images acquired with a 10x objective. Metamorph software (RRID:SCR_002368; Molecular Devices, Downington, PA) was used to measure the total CC ROI area in coronal sections immunolabeled for MOG along with DAPI staining of nuclei for cytoarchitecture of CC as distinct from adjacent regions. Myelination of the CC was measured based on pixel intensity values to determine the MOG immunolabeled pixels above background levels using the Metamorph thresholding function [20]. Similar thresholding was used to quantify astrogliosis and microglia activation based on GFAP and IBA1 immunoreactivity, respectively.
Quantification details for structure tensor analysis of astrocytes and myelin
NIH ImageJ software (ImageJ, RRID:SCR_003070) with the OrientationJ Plug-in (RRID:SCR_014796, http://bigwww.epfl.ch/demo/orientation/) was used for structure tensor analysis [58]. Images were acquired with a 10x objective. Using the polygon tool, the ROI was selected within the CC under the medial extension of the cingulum. This CC region avoids the curvature toward the midline and the crossing fibers that are present more laterally. The program computes the microscopic, or local, orientation and local coherence for each pixel. The local orientation uses a color map to represent the directional distribution. The local coherence is a measure of the alignment of anisotropic domain tensors. Both the anisotropy of a local domain and the coherence of domains within a voxel contribute to fractional anisotropy [59].
Quantification details for oligodendrocyte lineage populations
Oligodendrocyte counts in the CC were based on in situ hybridization and quantified from bright field images with the CC ROI area measured using Spot Advanced Software (RRID: SCR_014313; Spot imaging solutions, Sterling Heights, MI). Plp1low expressing cells had mRNA transcripts localized mainly in the perinuclear cytoplasm; in Plp1high expressing cells, darker substrate reaction was evident in the cell body and extended out into processes [20, 47, 60]. Only cells with strong substrate reaction for Enpp6 transcript levels were counted as specific labeling of newly formed oligodendrocytes [57]. Quantification of proliferating OPCs in the CC and cingulum were identified based on Ki67 immunoreactive nuclei and NG2 immunolabeling of the cell body and processes. Ki67 and NG2 analysis included only one section per mouse due to the limited availability of tissue within the defined coronal levels.
Quantification details for axon damage
Confocal images were acquired at 63x and quantified in maximum intensity projections of the ROI (59.70 μm, y: 59.70 μm, z: 1.60 μm) in the cingulum. The ROI was positioned adjacent to the CC and centered under the peak of the cingulum. Individual axons were manually counted as immunolabeled for NF-H with or without co-labeling for SMI32. Nuclei were counted simultaneously. Ipsilateral and contralateral sides were quantified in at least 3 sections per mouse.
Transplanted iNSC localization and differentiation in vivo
Transplanted iNSCs were quantified by direct visualization of GFP expression using a 40x objective on an Olympus IX-70 microscope. Tissue sections were analyzed from mice in the imaging (n = 6) and Miss-step wheel assessments (n = 11) that were used for quantification of MOG, GFAP, and IBA1 immunoreactivity. The in situ hybridization reaction for Plp1 precluded identification of GFP expression from iNSCs. Additional tissue sections were immunostained for labeling of iNSCs with cell type markers. Overall, this iNSC cell type quantification included at least 6 mice per cell type immunostain with at least 3 sections analyzed per mouse combining to approximately 200 iNSCs each for Sox2 and for Olig2, with approximately 600 iNSCs counted for GFAP which included sections from the neuropathology analysis of astrogliosis.
Transplanted iNSCs were analyzed in vivo only within coronal sections from rostrocaudal levels matching the neuropathology ROI (− 0.5 mm to − 2.0 mm from bregma), which focused on the most demyelinated CC region and minimize variation [40]. Regions of the cerebral cortex and cingulum above the CC, and hippocampus below the CC, were quantified within the same coronal sections as the CC. In total, 1585 iNSC cells were counted from among mice combined from MRI and Miss-step wheel cohorts and graphed to represent the distribution within the tissue as white matter (CC, cingulum) and adjacent gray matter (cortex, hippocampus). Contingency tables were used to analyze in vivo differentiation of iNSCs after transplantation with Fisher’s exact test for statistical significance.
Additional images of GFP cells were acquired on a Zeiss 700 laser scanning confocal microscope (Carl Zeiss, San Diego, California; RRID:SCR_017377) with individual laser lines sequentially collected for each channel using a Plan-Apochromat 63x/1.4 oil objective. An optical image stack was acquired, and maximum intensity projection images were generated from each image stack with Zen software (Carl Zeiss, ZEN Digital Imaging for Light Microscopy, RRID:SCR_013672).