Transforming growth factor α transforms astrocytes to a growth-supportive phenotype after spinal cord injury

Abstract

Astrocytes are both detrimental and beneficial for repair and recovery after spinal cord injury (SCI). These dynamic cells are primary contributors to the growth-inhibitory glial scar, yet they are also neuroprotective and can form growth-supportive bridges on which axons traverse. We have shown that intrathecal administration of transforming growth factor α (TGFα) to the contused mouse spinal cord can enhance astrocyte infiltration and axonal growth within the injury site, but the mechanisms of these effects are not well understood. The present studies demonstrate that the epidermal growth factor receptor (EGFR) is upregulated primarily by astrocytes and glial progenitors early after SCI. TGFα directly activates the EGFR on these cells in vitro, inducing their proliferation, migration, and transformation to a phenotype that supports robust neurite outgrowth. Overexpression of TGFα in vivo by intraparenchymal adeno-associated virus injection adjacent to the injury site enhances cell proliferation, alters astrocyte distribution, and facilitates increased axonal penetration at the rostral lesion border. To determine whether endogenous EGFR activation is required after injury, SCI was also performed on Velvet (C57BL/6J-Egfr(Vel)/J) mice, a mutant strain with defective EGFR activity. The affected mice exhibited malformed glial borders, larger lesions, and impaired recovery of function, indicating that intrinsic EGFR activation is necessary for neuroprotection and normal glial scar formation after SCI. By further stimulating precursor proliferation and modifying glial activation to promote a growth-permissive environment, controlled stimulation of EGFR at the lesion border may be considered in the context of future strategies to enhance endogenous cellular repair after injury.

Figure 1.

EGFR expression is increased and colocalized on astrocytes and progenitors at the lesion site after SCI. A, Western blots of naive (Ctl) and injured spinal cord tissues showing a single band of ∼175 kDa for EGFR, with β-tubulin (∼50 kDa) as a loading control. B, EGFR expression is increased and remains high after spinal cord injury. ANOVA, p < 0.001; *p < 0.05, **p < 0.01 versus Ctl (post hoc tests). C, Wide-field fluorescence image of uninjured spinal cord white matter showing EGFR expression predominately in gray matter, including the neuropil throughout the ventral horn (VH; C′) and punctate staining in lateral white matter (LWM; C″). D–D″, Confocal microscopy shows EGFR expression in profiles throughout naive white matter (red; D, arrows) is colocalized with NF+ axons (D′, blue; D″, magenta profiles) but colocalization with astrocytes (GFAP, green) is rare. E, During the first week after injury, EGFR expression is upregulated primarily in the spared white matter, with minimal expression in the lesion core (*). E–E″, High-power confocal image enlargement of white box in E shows profiles in register reflecting EGFR+ (E′) and GFAP+ (E″) astrocyte processes. F–F″, Confocal micrograph at the lesion border at 3 DPI depicting EGFR+ profiles (F) colocalized with BLBP+ cells of astrocyte lineage (F′, F″, white arrowheads). G, Confocal projection through a 10-μm-thick slice showing BLBP+/EGFR+ astrocytes at the lesion border; a colabeled cell (white arrow) projected along the right and bottom borders of the image (yellow arrows). H, Examples of BLBP+/EGFR+ cells and z-stack from spared white matter at 3 DPI. I, J, Confocal micrographs of the central canal rostral to the injury epicenter with BLBP+/EGFR+ cells at 3 and 7 DPI. Scale bars: C, E, 50 μm; C′, D, F, H′, I′, 10 μm; C″, G, H′, I′, 20 μm.

Figure 2.

TGFα is a potent mitogen for ASCNPCs and astrocytes. A, Dose–response experiment shows the number of BrdU+ profiles 3 d after treatment with TGFα alone, or with the addition of 10 μm AG1478 (+AG). ***p < 0.001, ANOVA and post hoc differences compared with 0 ng/ml TGFα alone. B, Representative images of BrdU+ profiles in ASCNPC cultures after 3 d of treatment with 0 or 25 ng/ml TGFα with or without 10 μm AG1478. C, C′, Representative confocal images showing BrdU+ astrocytes (GFAP+) in the presence of 10% FBS (C) or 25 ng/ml TGFα (C′), stained with GFAP (red) and BrdU (green). Scale bars: B, 50 μm; C, 10 μm.

Figure 3.

TGFα encourages wound closure in confluent ASCNPC and mixed glial cell cultures. A, A′, Confocal images of nestin immunoreactivity in ASCNPC cultures without (A) and with (A′) 25 ng/ml TGFα present in the medium for 3 d. White lines depict borders used for analysis. B, TGFα induces a dose-dependent wound closure response that is blocked by AG1478 (ANOVA treatment, dose, interaction; post hoc, ***p < 0.001). C, C′, GFAP immunoreactivity reveals formation of glial bridges in mixed cultures with progenitors and astrocytes incubated with TGFα for 5 d. D, TGFα increases wound closure (ANOVA, p < 0.05), with significant post hoc difference at the 50 ng/ml dose (p < 0.05). Migration is blocked by AG1478. E, E′, Fully differentiated astrocytes switch from polygonal morphology seen in 10% FBS (E) to an elongated morphology resembling radial glia when incubated 5 d in 25 ng/ml TGFα (E′). F, Astrocytes do not migrate into the scratch area. Scale bars, 50 μm.

Figure 4.

TGFα induces elongation of both ASCNPCs and astrocytes while exerting opposite effects on astrocyte marker expression intensity. A, ASCNPCs stained with anti-GFAP develop elongated and aligned processes after incubation in 10 ng/ml TGFα. B, Decreased expression of GFAP staining intensity with TGFα treatment. C, D, ASPNPCs show decreased BLBP expression with serum starvation (C′) and no additional change after incubation in TGFα (C″). E–H, Differentiated astrocytes in FBS (E, G) become elongated with evidence of stress fiber formation when serum starved (E′, G′). In contrast, they develop into radial glia-like elongated profiles after treatment for 5 d with 25 ng/ml TGFα (E″, G″). Both GFAP (F) and BLBP (H) stain intensity is increased after removal of FBS and further enhanced with the addition of 25 ng/ml TGFα. p < 0.05 for all graphs (ANOVAs). *p < 0.05, **p < 0.01, ***p < 0.001 (post hoc comparisons). Scale bars, 50 μm.

Figure 5.

TGFα-transformed astrocytes support axonal growth. A, Camera lucida drawings of representative DRG neurons that were plated on cellular and acellular substrates, including laminin alone and laminin in the presence of 10 ng/ml TGFα, or on astrocytes pretreated with either 10% FBS or 25 ng/ml TGFα before plating. B, C, Confocal images of NF+ axons (red) plated for 24 h on astrocytes (green) after treatment in control medium (10% FBS) or 25 ng/ml TGFα for 5 d. Scale bars, 20 μm. D, The addition of TGFα did not inhibit neurite outgrowth of DRG neurons plated on laminin (Lam/10) compared with laminin with no TGFα. In contrast, differentiated astrocytes prepared in 10% FBS (Astros/FBS) were inhibitory to axon growth. Astrocytes treated with 25 ng/ml TGFα for 5 d (Astros/25) were as permissive as those plated on laminin alone. p < 0.05, ANOVA; ***p < 0.001, post hoc comparison.

Figure 6.

AAV is incorporated into spinal cord neurons and astrocytes and increases TGFα mRNA and peptide expression after SCI. A, Schematic showing location of injection sites at the T9 laminectomy site (top) and the estimated viral spread (bottom, green) and site of contusion injury administered 2 weeks later (gray/red oval). B, Representative size of GFP-AAV injection site in midthoracic spinal cord at 2 weeks after injection. C, RT-PCR of human TGFα mRNA expression after AAV injection and SCI. + Control, TGFα-transfected HEK 293 cells; Naïve, naive tissue. D, Results of ELISA confirm expression of TGFα peptide in spinal cord at 2 weeks after TGFα-AAV injection, compared with naive spinal cord, laminectomy with eAAV, or 10 d after SCI after eAAV injection. Values represent mean ± SEM for two to three samples per group; *p < 0.05. E, F, Confocal images of GFP (E, F), NeuN (E′, F′), and GFAP (E″, F″) showing neurons and astrocytes expressing GFP 2 weeks after GFP-AAV injection. Scale bars: B, 50 μm; E, F, 5 μm.

Figure 7.

TGF-AAV increases astrocyte migration and axonal extension into the site of a contusion injury (*) at 10 DPI. A, B, Fluorescence images of longitudinal sections through the lesion border in GFP-AAV-treated (A) and TGFα-AAV-treated (B) mice, stained with anti-GFAP. A′, B′, High-power images of the lesion edge. A″, B″, Confocal micrographs showing the relationship of astrocytes (GFAP, red), axons (NF, blue), and GFP expression (green) at the lesion border in GFP-AAV-treated (A″) and TGFα-AAV-treated (B″) mice. C, Representative images of GFAP immunoreactivity in cross sections from 0.4 mm rostral to 0.4 mm caudal to the lesion epicenter in PBS-, eAAV-, and TGFα-AAV-treated mice. D, Volume of the GFAP-devoid region (n = 3–9 per group; p < 0.05, ANOVA main effect; corrected post hoc tests nonsignificant). E, Images of NF immunoreactivity at the lesion border (0.4 mm rostral to the epicenter) from representative PBS-, eAAV-, and TGFα-AAV-treated specimens. Blue outlines denote the area used to measure NF immunoreactivity. F, Proportional area (PA) of sample box occupied by NF profiles. p < 0.05, two-way ANOVA main effects of treatment and distance across the lesion; *p < 0.05 for TGFα-AAV vs PBS and eAAV at 0.4 mm rostral to the epicenter (post hoc treatment effect). Scale bars: A, B, 100 μm; A′, A″, B′, B″, 20 μm.

Figure 8.

Axons growing along newborn astrocytes after TGFα-AAV treatment. A–A‴, C, Confocal image of TGFα-AAV-treated mouse section from just rostral to the injury epicenter. New axons (GAP43+, green; A′) grow along GFAP+ profiles (purple; A″) in association with laminin (red; A′). C, Enlargement from boxed area in A‴ shows an axon profile along a GFAP+/laminin+ process (arrows). B, B‴, Confocal images of astrocytes (GFAP, green) and centrally derived axons (5-HT, red) at the lesion border of a PBS-treated (B) and TGFα-AAV-treated (B′) specimen. 5-HT axons (arrowheads) are found both adjacent to and away from the GFAP+ profiles within the lesion. D, Low-power image of BrdU+ nuclei (green) and BLBP+ staining surrounding the central canal just rostral to the site of contusion. Some BrdU+ nuclei are colocalized with BLBP+ cells (yellow arrowheads), whereas some others are not (white arrows). E, High-magnification confocal image of BrdU+ (green) astrocytes (BLBP, red) with axons (NF, blue) growing alongside them (white arrows). F, Confocal z-stack and projection images (right and bottom) of a BrdU+/BLBP+ cell in white matter near the lesion border. G, G″, TGFα-AAV increases proliferation of cells at the lesion border. G, Low-power wide-field image series of BLBP immunostaining at 200 μm intervals spanning the lesion in a TGFα-AAV-treated specimen. Counting frames (white boxes) are shown at the rostral and caudal borders. G′, Enlargement of confocal image showing BrdU+ nuclei and BLBP+ staining with a counting frame from a representative section. G″, More BrdU+ nuclei were found at the lesion borders in TGFα-AAV-treated mice than PBS- or eAAV-treated mice. H, High-power confocal image and projection of BrdU+/BLBP+ cell intertwined with NF+ axons within the caudal lesion border. p < 0.01, two-way ANOVA main-effect treatment; **p < 0.01, ***p < 0.001 vs PBS (post hoc comparisons); ⁺p < 0.05, versus eAAV. Scale bars: A–A‴, C, D, G, 50 μm; F, 20 μm; B, B′, E, H, 10 μm.

Figure 9.

Mice with dominant Velvet gene show impaired locomotor recovery and enlarged GFAP-negative area at the lesion epicenter after SCI. A, Locomotor recovery is impaired in Velvet mice compared with WT littermates. Two-way ANOVA revealed main effects of time (p < 0.001), treatment (p < 0.01), and interaction (p < 0.01); *p < 0.05, **p < 0.01 (post hoc). B, C, Images of GFAP immunoreactivity at the lesion epicenter of a WT (C) and Velvet (D) specimen showing abnormal scar formation in the mutant mouse. D, E, BrdU+ nuclei (green) are found throughout the GFAP+ stained region at the lesion borders. F, Two-way ANOVA revealed a significant main effect of genotype on the number of BrdU+ nuclei per sample region. *p < 0.05, by post hoc corrected Bonferroni's test. Scale bars: B, C, 100 μm; D, E, 40 μm.