Transplanted astrocytes derived from BMP- or CNTF-treated glial-restricted precursors have opposite effects on recovery and allodynia after spinal cord injury

Abstract

Background: Two critical challenges in developing cell-transplantation therapies for injured or diseased tissues are to identify optimal cells and harmful side effects. This is of particular concern in the case of spinal cord injury, where recent studies have shown that transplanted neuroepithelial stem cells can generate pain syndromes.

Results: We have previously shown that astrocytes derived from glial-restricted precursor cells (GRPs) treated with bone morphogenetic protein-4 (BMP-4) can promote robust axon regeneration and functional recovery when transplanted into rat spinal cord injuries. In contrast, we now show that transplantation of GRP-derived astrocytes (GDAs) generated by exposure to the gp130 agonist ciliary neurotrophic factor (GDAs(CNTF)), the other major signaling pathway involved in astrogenesis, results in failure of axon regeneration and functional recovery. Moreover, transplantation of GDA(CNTF) cells promoted the onset of mechanical allodynia and thermal hyperalgesia at 2 weeks after injury, an effect that persisted through 5 weeks post-injury. Delayed onset of similar neuropathic pain was also caused by transplantation of undifferentiated GRPs. In contrast, rats transplanted with GDAs(BMP) did not exhibit pain syndromes.

Conclusion: Our results show that not all astrocytes derived from embryonic precursors are equally beneficial for spinal cord repair and they provide the first identification of a differentiated neural cell type that can cause pain syndromes on transplantation into the damaged spinal cord, emphasizing the importance of evaluating the capacity of candidate cells to cause allodynia before initiating clinical trials. They also confirm the particular promise of GDAs treated with bone morphogenetic protein for spinal cord injury repair.

Figure 1

Schematic illustration of the adult rat models of spinal cord injury used in this study. (a) Dorsal view of rat brain and spinal cord. Dorsal column white matter on the right side was transected (shaded area) at the C1/C2 spinal level, and the ability of either BDA-labeled endogenous axons or axons from microtransplanted GFP-expressing adult sensory neurons (DRGs) to cross injuries bridged with GDAs or GRPs was assayed. (b) Horizontal and (c) sagittal views of the dorsal column white-matter pathways at the C1/C2 cervical vertebrae of the spinal cord. (b) Injections of GDAs or GRPs (black diamonds) suspended in medium were made directly into the centers of the injury sites as well as their rostral and caudal margins in the cervical spinal cord. (c) A discrete population of endogenous ascending axons within the cuneate and gracile white-matter pathways of dorsal columns was labeled by BDA injection at the C4/C5 spinal level (5 mm caudal to the injury site, shaded). Alternatively, microtransplants of GFP⁺ DRGs were injected 500 μm caudal to the injury site. (d) The right-side dorsolateral funiculus white matter containing descending axons of the rubrospinal tract was transected at the C3/C4 spinal level and GDAs or GRPs were transplanted as described for dorsal column injuries. CC, central canal; Cf, cuneate fasciculus; CST, corticospinal tract; DF, dorsolateral funiculus; Gf, gracile fasciculus; GM, gray matter; RN, red nucleus; RST, rubrospinal tract; T1, level of the first thoracic vertebra.

Figure 2

GDAs^BMP, GDAs^CNTF and GRPs express different levels of NG2 and phosphacan in vitro. GRPs were induced to differentiate into astrocytes by exposure to BMP or CNTF. Relative levels of expression of NG2 and phosphacan proteins were determined by quantitative Western blot and immunocytochemical analysis. (a,c) Western blot analysis of whole-cell lysates demonstrates that GDAs^CNTF express higher levels of (a) NG2 and (c) phosphacan. The graph shows fold change in protein levels for GDAs compared to GRPs. Error bars represent 1 standard deviation (SD). *p < 0.05. (b,d) Immunofluorescent labeling of cells using (b) anti-NG2 antibodies and (d) anti-phosphacan. Scale bars 50 μm.

Figure 3

Differential expression of Olig2 protein by different astrocyte populations. (a) GDAs^BMP do not express Olig2. (b) In sharp contrast, GDAs^CNTF are uniformly immunopositive for Olig2 in vitro. (c) A subset of endogenous GFAP⁺ astrocytes in the margins of untreated dorsal column spinal cord injuries is also Olig2-immunoreactive. Survival, 8 days post-injury. Note the nuclear localization of Olig2 in GDAs^CNTF in vitro and in reactive, endogenous GFAP⁺ astrocytes in vivo. Scale bars: (a,b) 50 μm; (c) 25 μm.

Figure 4

GDAs^CNTF express GFAP and neurocan after transplantation into spinal cord injuries. (a) Intra-injury GDAs^CNTF are uniformly GFAP⁺ within acute dorsal column injuries. Note the co-localization (yellow) of human placental alkaline phosphatase (hPAP, red) with GFAP (green). GDAs^CNTF have also failed to align host astrocytic processes within injury margins. Survival, 8 days post-injury/transplantation. (b) High-magnification confocal image of neurocan immunoreactivity at the injury margin and within a GDA^CNTF-transplanted injury site at 8 days after injury/transplantation. Note that some GDAs^CNTF are immunoreactive for neurocan (green). In contrast, intra-injury transplanted GDAs^BMP (not shown) do not express GFAP or neurocan, and can align host astrocytic processes within injury margins [14]. Scale bars 100 μm.

Figure 5

NG2 immunoreactivity in GDA^CNTF-transplanted dorsal column injuries. (a) Transplanted hPAP⁺ GDAs^CNTF (arrowheads) at 8 days post injury/transplantation. (b) The same slide stained for NG2 (green) showing that the transplanted cells (arrowheads) show immunoreactivity for NG2. (c) Co-localization (yellow) of NG2 and hPAP immunoreactivity in regions containing higher densities of GDAs^CNTF (arrowheads). In general, regions of the injury site that contained higher densities of hPAP⁺ GDAs^CNTF had a higher density of NG2 immunoreactivity. Scale bars 50 μm.

Figure 6

Transplanted GDAs^CNTF express neurocan and NG2, but suppress host expression of these two CSPGs at 4 days post-injury/transplantation. (a) At 4 days after injury, control dorsal column injury margins express dense neurocan immunoreactivity (green) mainly associated with GFAP^- processes. Note the absence of neurocan immunoreactivity in the injury center (to the left). (b,c) While neurocan immunoreactivity in host white matter was markedly lower and mainly associated with astrocyte cell bodies, many intra-injury GDAs^CNTF within injury centers displayed neurocan immunoreativity. (d) NG2 immunoreactivity in control injuries is high in both injury centers and margins. (e,f) Although overall levels of NG2 immunoreactivity were reduced within injury centers and margins of GDA^CNTF-transplanted injury sites compared to untreated control injuries (compare (d) and (f)), levels of NG2 immunoreactivity were still higher than that previously observed for identical dorsal column injuries transplanted with GDAs^BMP [14]. Scale bars 200 μm.

Figure 7

Failure of axons to regenerate across GDA^CNTF or GRP transplanted dorsal column injuries. (a) Biotinylated dextran amine (BDA)-labeled endogenous, ascending dorsal column axons (green) fail to cross GDA^CNTF-transplanted injury sites and instead form dystrophic endings within caudal injury margins. While a few axons sprout towards the injury center, BDA⁺ axons are rarely detected beyond the injury/transplantation site at 8 days post-injury/transplantation. Scale bar 200 μm. (b) In contrast, transplanted GDAs^BMP support extensive axon growth across dorsal column injuries at 8 days after injury/transplantation. Scale bar 200 μm. (c) Quantification of numbers of regenerating BDA⁺ axons in GDA- or GRP-transplanted dorsal column white matter at 8 days after injury and transplantation. BDA-labeled axons were counted in every third sagittally oriented section within the injury center and at points 0.5 mm, 1.5 mm and 5 mm rostral to the injury site and within the dorsal column nuclei (DCN). Note that 55% of BDA⁺ axons reached the centers of GDA^BMP-transplanted injuries, and 36% to 0.5 mm beyond the injury site. After GDA^CNTF or GRP transplantation, however, only 7% and 5.3% of BDA⁺ axons, respectively, were observed within injury centers, with only 4.6% and 4.2% of the axons observed at 0.5 mm beyond the injury site. No BDA⁺ axons were detected beyond 1.5 mm rostral to the injury site in GDA^CNTF- or GRP-transplanted spinal cords. Error bars represent 1 SD.

Figure 8

Grid-walk analysis of locomotor recovery. Graph showing the average number of missed steps per experimental group from 1 day before injury (baseline pre-injury) to 28 days after injury for all GDA-transplanted/dorsolateral funiculus injured rats versus the control-injured animals. GDA^BMP-transplanted animals (green) performed significantly better than GDA^CNTF-transplanted animals and injured control animals at all post-injury time points (p < 0.05). Note that the performance of GDA^CNTF-transplanted animals was not different from untreated control injured rats at all time points (two-way repeated measures ANOVA, *p < 0.05). N = 9 rats per group.

Figure 9

Neuroprotection of red nucleus neurons. Injured left-side red nuclei contained an average of 52% of the neurons counted in uninjured right-side red nuclei at 5 weeks after transection of the right-side rubrospinal tract. The numbers of neurons in the injured left-side red nuclei of GRP- and GDA^CNTF-transplanted animals were no different from controls, and contained an average of 55% and 51%, respectively, of the neurons counted in the uninjured right-side nuclei. In contrast, the number of neurons in the injured left-side red nuclei of GDA^BMP-transplanted animals was 81% of the total number of neurons in uninjured right-side nuclei. *p < 0.01. Error bars represent 1 SD.

Figure 10

Von Frey filament and hot-plate analysis of mechanical and thermal allodynia. (a) Withdrawal threshold of the right front paw to a mechanical stimulus (force in grams). Measurements were made on GRP- or GDA-transplanted and injured control animals at 2, 3, 4 and 5 weeks after dorsolateral funiculus injury/transplantation. (b) Latency (in seconds) to paw withdrawal from a heat source. Note that injury alone and GDA^BMP transplantation do not induce statistically significant mechanical or thermal allodynia at any time point. However, the mechanical threshold and latency to withdrawal from a heat source are significantly lower in GDA^CNTF- and GRP-transplanted rats beginning at 2 and 3 weeks, respectively, post-injury/transplantation. Asterisks denote a statistical difference from time-matched control animals (two-way repeated measures, ANOVA, p < 0.05). Error bars represent 1 SD.

Figure 11

Aberrant CGRP⁺ c-fiber sprouting into lamina III of GDA^CNTF- or GRP-transplanted spinal cords that have received dorsolateral funiculus transection injuries. The density of pixels within images of lamina III of the right-side dorsal horn caudal to the injury and transplantation site in GDA- or GRP-transplanted, or injury-only control animals is presented as the average percentage of CGRP⁺ pixels per total pixels (area) of lamina III. (a) Averages of 5.7% and 6.2% of the total pixels in lamina III were CGRP⁺ in GDA^CNTF- and GRP-transplanted spinal cords, respectively. In contrast, only 2.2% and 3.4% of lamina III pixels were CGRP⁺ in GDA^BMP-transplanted and injury-only spinal cords. The asterisk indicates significant difference from both control injury-only and GDA^BMP-transplanted groups (two-way repeated measures ANOVA, p < 0.05). Error bars represent 1 SD. (b-e) Sample images of sections labeled with anti-CGRP antibodies from rats transplanted at the spinal C6 level: (b) control, (c) GDA^BMP, (d) GDA^CNTF and (e) GRP. Area enclosed with a dashed line in (b-e) indicates lamina III. Note the increased density of CGRP⁺ immunoreactivity within lamina III of the dorsal horn of (d) GDA^CNTF- and (e) GRP-treated spinal cords compared to (b) control injured and (e) GDA^BMP-treated spinal cords. Scale bar 200 μm.