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

Abstract

Neural stem/progenitor cell (NSPC) grafts can integrate into sites of spinal cord injury (SCI) and generate neuronal relays across lesions that can provide functional benefit. To determine if and how grafts become synaptically organized and connect with host systems, we performed calcium imaging of NSPC grafts in SCI sites in vivo and in adult spinal cord slices. NSPC grafts organize into localized and spontaneously active synaptic networks. Optogenetic stimulation of host corticospinal tract axons regenerating into grafts elicited distinct and segregated neuronal network responses throughout the graft. Moreover, optogenetic stimulation of graft-derived axons extending from the graft into the denervated spinal cord also triggered local host neuronal network responses. In vivo imaging revealed that behavioral stimulation likewise elicited focal synaptic responses within grafts. Thus neural progenitor grafts can form functional synaptic subnetworks whose activity patterns resemble intact spinal cord.

Keywords: all-optical; calcium imaging; graft; neural progenitor cell; neural stem cell; neuronal relay; optogenetics; slice preparation; spinal cord injury; transplant.

Figure 1:. Chrimson-expressing corticospinal axons regenerate robustly into GCaMP6f-expressing neural progenitor cell grafts

(A) Experimental timeline. Animals received cortical injections of AAV-ChrimsonR at postnatal day 0 (P0). T12 dorsal column lesions and acute neural progenitor cell grafting were performed at 6 – 8 weeks of age. Calcium imaging was performed 6 – 8 weeks later. (B) Sagittal section of adult whole brain shows abundant expression of ChrimsonR-tdTomato in cortex following neonatal AAV injections. Neurons labeled with NeuN. Inset in (B) shown at higher magnification in (C). (D) Chrimson-labeled corticospinal tract (CST) axons regenerate extensively into neural progenitor cell grafts expressing GCaMP6f in neurons in a graft-specific, Cre-dependent manner. GCaMP6f (E) and Chrimson (F) in separate channels. NeuN labels neurons in (E). In all images, the dorsal aspect of the slice is at the top and the rostral aspect is to the left. All images are composites of multiple adjacent images (see STAR Methods). Scale bar, 2 mm (B); 200 μm (C-F). See also Figure S1.

Figure 2:. Clusters of graft neurons exhibit spontaneous, correlated activity

(A) Fluorescence traces of 40 graft cells exhibiting spontaneous activity. Red portions of traces indicate periods of significant activity as determined by first derivative rasterization. Individual regions of interest exhibit the onset of calcium activity concurrently, and intermittently repeat this concurrent activation pattern. Moreover, many of these cells are located adjacent to one another, as clusters of cells in grafts (see panel B). Additional traces in Figure S3. (B) Standard deviation projection of the calcium activity run shown in (A) with all regions of interest (ROIs) outlined in yellow; clusters of neurons with similar activity dynamics are indicated by fill color. Many cells exhibiting concurrent onset of calcium activity are spatially adjacent to one another; we refer to these clusters as “assemblies.” (C) First derivative raster plot (top) of traces from (A), with cells grouped into the assemblies shown in (B). Cells above red line belonged to assemblies. Significant activations of color-coded assemblies are plotted below. (D) The mean activity of cells belonging to assemblies was not significantly different from that in control, random assemblies or surrogate distance-control assemblies (see text; N = 4 animals per group analysis). (E) Cells in assemblies showed significantly more pairwise correlation than those of surrogate assemblies (N = 4 animals). (F) Series of serial imaging frames (ΔF/F) showing temporal rise of calcium activity in separate assemblies in the same imaging run as in (A-C). Generally, peak calcium activity is reached over ~500-1000 msec. Arrowheads indicate same cells within three separate assemblies over sequential frames. The colors of the arrowheads indicate the assemblies in (B) to which the cells belong. In all images, the dorsal aspect of the slice is at the top and the rostral aspect is at left. Scale bar, 200 μm (B, F). Data presented as mean ±Sem with significance determined by Welch’s t-test (*p<0.05; NS, not significant). See also Figures S2-S4.

Figure 3:. Graft neurons are activated by optogenetic stimulation of regenerating corticospinal axons

(A) One second average of ΔF/F video prior to stimulation onset shows several spontaneously active graft neurons. (B-C) The same field as (A), immediately following stimulation onset. GCaMP6f fluorescence is activated in several graft neurons in response to stimulation of corticospinal axons that have regenerated into the graft (cyan numbered arrows in (C)). (D) Sample traces from cells labeled in (C) during a 50 sec period over which four corticospinal stimuli (vertical cyan bars) are delivered. Stimuli consisted of 500 msec-long 20 Hz trains of 10 msec pulses of 617 nm light. Irradiance at the slice was 4.49 mW/mm² Significant fluorescence transients are highlighted in red. Cells display varying response strengths as well as spontaneous activity. Notably, clusters of neurons within grafts repeatedly responded to corticospinal stimulation (see also Fig. 4). (E) Mean response traces with ±SEM shade of n = 7 cells in one slice to 12 trials in standard recording conditions, + 100 μM 6,7-dinitroquinoxaline-2,3-dione (DNQX), and after DNQX washout. Responses were abolished by the AMPA and kainite receptor antagonist DNQX, indicating the presence of excitatory synaptic connections from corticospinal axons onto graft neurons. (F) Mean response strengths from (E), quantified over one second following stimulus onset. (G) Mean response strength quantified as in (F) in response to increasing 617 nm LED irradiance. Responses in n = 16 cells in one slice increased in strength with increasing irradiance and plateaued at higher irradiance. (H) The percentage of cells activated (activation threshold = 100% ΔF/F) by corticospinal stimulation also increased with increasing 617 nm LED irradiance. In all images, the dorsal aspect of the slice is at the top and the rostral aspect is at left. Scale bar, (A-C), 100 μm. Two-group comparisons were tested with Welch’s t-test (**p<0.01; NS, not significant). Data are represented as mean ±SEM. See also Figure S4.

Figure 4:. Spontaneously active graft neuron assemblies are activated by corticospinal stimulation

A) Same graft as in (Figure 2B) with assemblies extracted from a run with optogenetic corticospinal stimulation. (B) As in (Figure 2C), during a run with four trials of 500 msec, 20 Hz corticospinal stimulation (indicated by arrows and vertical gray bars). (C-F) One second, serial 33-frame average ΔF/F images of graft responses to four consecutive corticospinal stimulation trials show graft regions (ellipsoids) and individual neurons (arrowheads) that respond consistently (cyan) or inconsistently (magenta) to stimulation. Red circle indicates a region that was spontaneously active but did not respond to corticospinal stimulation (region was already activated prior to stimulus in (E)). (G) One second average from the same field of view as (C-F) during a period of low activity in an imaging run with no stimuli given. (H) The same field during a period of spontaneous activation of the regions in the red and magenta ellipsoids. In all images, the dorsal aspect of the slice is at the top and the rostral aspect is at left. Scale bar, 200 μm (A,C-H). See also Figures S5 and S6.

Figure 5:. Host neurons are activated by optogenetic stimulation of axons extending from grafts

(A) Graft that expresses Chrimson extends axons into host spinal cord rostral and caudal to the lesion site. (B) Inset from (A) showing graft axon innervation of regions of GCaMP6f-expressing host spinal cord neurons caudal to the lesion. (C) One second average ΔF/F image of host spinal cord neurons prior to stimulus onset. (D-E) In the same field as (C), one second average following onset of 500 msec optogenetic stimulation of graft axons (cyan numbered arrows in (E) label analyzed cells). (F) Individual cell fluorescence traces of the cells in (E) show abundant spontaneous activity as well as consistent responses in some cells to graft axon stimulation. (G) Mean ±SEM traces of n = 5 responding cells from one slice before and after wash-in of 100 μM DNQX, showing that grafts form excitatory synaptic connections with host neurons below the lesion. In all images, the dorsal aspect of the slice is at the top and the rostral aspect is at left. (A) and (B) are derived from a composite of multiple adjacent images (see STAR Methods). Scale bars, 300 μm (A); 150 μm (B); 100 μm (C-E).

Figure 6:. Graft neurons respond to sensory stimuli in vivo

(A-B) Baseline average ΔF/F image of a graft neuron over a one second period just prior to (A) and one second following onset of a spontaneous calcium transient (B) in a single graft neuron with graft-specific GCaMP6f expression in an anesthetized animal. (C) Fluorescence trace of the calcium transient shown in (A-B). (D) Maximum projection image of in vivo calcium activity during a period of three hindpaw pinches. ROIs (yellow outlines) include both neuronal soma and processes. (E) Fluorescence traces of the activity in the ROIs shown in (D) during hindpaw pinch (purple trace shows pinch strength). (F) Standard deviation projection of viral GCaMP6f expression in vivo during light touch. ROIs are highlighted in yellow. (G) Several ROIs respond consistently to light touch. (H) Two cropped ROIs from the same field of view as (E) respond consistently to movement of the hindlimb through its range of motion by the experimenter. These ROIs, which may be part of a single neuron or neuron cluster, were on the left side of the graft and responded to left hindlimb movement (green bars) but not right hindlimb movement (pink bars). Grafts in (A-E) were derived from Syn1-Cre/Ai95D embryos (fully transgenic GCaMP6f expression), and grafts in (F-H) were derived from Syn1-Cre embryos with AAV-FLEX-GCaMP6f mixed with cells at the time of grafting. Scale bars, (A-B), 25 μm; (D), 30 μm; (F), 100 μm.