Inositol Polyphosphate-5-Phosphatase K (Inpp5k) Enhances Sprouting of Corticospinal Tract Axons after CNS Trauma

Abstract

Failure of CNS neurons to mount a significant growth response after trauma contributes to chronic functional deficits after spinal cord injury. Activator and repressor screening of embryonic cortical neurons and retinal ganglion cells in vitro and transcriptional profiling of developing CNS neurons harvested in vivo have identified several candidates that stimulate robust axon growth in vitro and in vivo Building on these studies, we sought to identify novel axon growth activators induced in the complex adult CNS environment in vivo We transcriptionally profiled intact sprouting adult corticospinal neurons (CSNs) after contralateral pyramidotomy (PyX) in nogo receptor-1 knock-out mice and found that intact CSNs were enriched in genes in the 3-phosphoinositide degradation pathway, including six 5-phosphatases. We explored whether inositol polyphosphate-5-phosphatase K (Inpp5k) could enhance corticospinal tract (CST) axon growth in preclinical models of acute and chronic CNS trauma. Overexpression of Inpp5k in intact adult CSNs in male and female mice enhanced the sprouting of intact CST terminals after PyX and cortical stroke and sprouting of CST axons after acute and chronic severe thoracic spinal contusion. We show that Inpp5k stimulates axon growth in part by elevating the density of active cofilin in labile growth cones, thus stimulating actin polymerization and enhancing microtubule protrusion into distal filopodia. We identify Inpp5k as a novel CST growth activator capable of driving compensatory axon growth in multiple complex CNS injury environments and underscores the veracity of using in vivo transcriptional screening to identify the next generation of cell-autonomous factors capable of repairing the damaged CNS.SIGNIFICANCE STATEMENT Neurologic recovery is limited after spinal cord injury as CNS neurons are incapable of self-repair post-trauma. In vitro screening strategies exploit the intrinsically high growth capacity of embryonic CNS neurons to identify novel axon growth activators. While promising candidates have been shown to stimulate axon growth in vivo, concomitant functional recovery remains incomplete. We identified Inpp5k as a novel axon growth activator using transcriptional profiling of intact adult corticospinal tract (CST) neurons that had initiated a growth response after pyramidotomy in plasticity sensitized nogo receptor-1-null mice. Here, we show that Inpp5k overexpression can stimulate CST axon growth after pyramidotomy, stroke, and acute and chronic contusion injuries. These data support in vivo screening approaches to identify novel axon growth activators.

Keywords: CNS trauma; plasticity; regeneration; spinal cord injury; stroke; transcriptional screening.

Figure 1.

Inpp5k enhances neurite outgrowth and regeneration in vitro. A, Schematic shows the approach to identify-intact CSNs undergoing axon growth after PyX. Adult ngr1^+/+ crym-GFP and ngr1^−/− crym-GFP transgenic mice received a PyX, 14-d-postlesion mice received a contralateral infusion of the retrograde tracer Fast Blue, and 28-d-postlesion mice were prepared for laser capture microdissection of intact quiescent CSNs (GFP⁺/FB^–) and intact sprouting CSNs (GFP⁺/FB⁺; Fink et al., 2017). Differential gene expression analysis, calculated using a Wilcoxon rank-sum test, showed that 2738 genes were significantly upregulated and 506 genes were downregulated in sprouting versus quiescent CSNs. B, Volcano plot shows every gene profiled as log2-fold change versus –log10 of the false discovery rate (FDR)-corrected p value. Non-SDE genes (gray dots) are separated from SDE genes (black dots) by p < 0.05 cutoff (stippled) line. Inositol phosphate genes (dark blue) and genes associated with the cytoskeleton (magenta dots) are primarily enriched in sprouting CSNs. Six 5-phosphatases (Inpp5k, Inpp5e, Inpp5j, Ocrl, Sac3, Synj1) and cofilin (CFL) are highlighted. IPA was conducted on genes enriched in sprouting CSNs. C, Table shows the IPA output of significantly upregulated “Canonical Pathways” and “Diseases and Biofunctions” across the whole dataset of differentially expressed genes (with Benjamini–Hochberg corrected p values for multiple comparisons), and canonical pathways associated with inositol phosphate signaling are highlighted in blue, and cytoskeletal dynamics in magenta. D–F, Dissociated E17.5 cortical neurons overexpressing Inpp5k-V5 (E), showed significant longer neurites compared with GFP controls (D, F; average total axon length of GFP (n = 45, gray dots) and Inpp5k-V5 (n = 45, light blue dots) neurons from n = 3 independent experiments (t₍₄₎ = 4.066, *p = 0.015, unpaired two-tailed t test) after DIV4. Data shown are the mean length of axon (in micrometers; biological n, darker dots) ± SEM. G–J, To assess whether Inpp5k-mediated enhanced neurite growth was mTOR dependent, we transduced E17.5 cortical neurons with AAV-mCherry or AAV-Inpp5k+/−300 nm rapamycin in DMSO. K, Neurons expressing mCherry showed minimal regeneration into the scrapped zone, while Inpp5k-expressing neurons showed a significant increase in regeneration compared with control. There was no significant difference in the neurite regeneration on addition of rapamycin to mCherry or Inpp5k-treated neurons (one-way ANOVA, using post hoc Tukey's HSD test: ****p < 0.0001, F_(5,44) = 18.24). Scale bars: D, 100 µm; H, 200 µm.

Figure 2.

Inpp5k increases active cofilin and microtubule advancement in growth cones in vitro. A–D, Assessment of the morphology of GCs of E17.5 cortical neurons transduced with GFP (A, B) and Inpp5k (C, D). E, At DIV4, β-tubulin (green) and phalloidin (blue) labeling showed that significantly more GCs with extending versus looped morphology in Inpp5k-treated cultures compared with controls [GFP (n = 60) and Inpp5k-V5 (n = 60) neurons from n = 3 independent experiments; unpaired two-tailed t test: t₍₄₎ = 5.75, **p = 0.005, data are shown as the mean percentage of extending growth cones ± SEM]. F–N, Assessment of the relative filopodial location and density of EB3⁺ comets in GCs of E17.5 cortical neurons transduced with GFP (F, H, J, L; EB3, red; β-tubulin, green; phalloidin, blue; pink arrows show colocalized EB3⁺ comets and microtubules; dotted line indicates the border of filopodia) and Inpp5k (G, I, K, M) shows that Inpp5k significantly elevated the density of EB3⁺ comets along filopodia (N; F_(1,4) = 10.21, p = 0.033, two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons), and the distance to which EB⁺ comets were found in distal regions of filopodia (F_{(1.339,5.356)} = 6.722, p = 0408, two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons). Data shown are EB3/β-tubulin intensity across the percentage of filopodia length ± SEM. O–W, Assessment of the relative density of (active) cofilin (red) and (inactive) phospho-cofilin (green) in GCs of E17.5 cortical neurons after DIV4 transduced with GFP (P–S) and Inpp5k (T–W) shows that Inpp5k-treated cultures had significantly more active cofilin compared with controls (O; unpaired two-tailed t test, t₍₄₎ = 5.531, **p = 0.005). Data shown are the mean density of active cofilin (biological n, darker dots; lighter dots, GCs; n = 15/condition) ± SEM. Scale bars: C, 20 µm; D, 10 µm; F, 4 µm; W, 4 µm.

Figure 3.

Plasmid maps for in vivo transduction of CSNs. A, B, Plasmid map shows AAV1-CAG-mCherry (A) and AAV1-CAG-Inpp5k-V5 (B) and the restriction sites used for subcloning for adeno-associated viral vector constructs for transduction of corticospinal neurons in vivo (for comprehensive cloning strategy, see Materials and Methods).

Figure 4.

Inpp5k increases CST sprouting following PyX. A, Schematics show relative locations of PyX and cortical injection of AAV1-mCherry and AAV1-Inpp5k-V5. Bottom right, Schematic of transverse cervical sections with mCherry (red) labeling in four quadrants that were analyzed; right, experimental timeline. B–E, Photomicrographs show V5 (B, D) and mCherry (C, E) staining in M1 cortex (B, C) and C4 spinal cord (D, E) 42 d after AAV infusion, confirming the expression of Inpp5k and the reporter in CSNs and CST axons. F–I, Photomicrographs show mCherry⁺ CST axon staining in transverse sections of C6 spinal cord from mice that received AAV1-mCherry⁺AAV1-FLEX-GFP 28 d after sham lesion (F) and PyX (G) and Inpp5k treatment 28 d after sham lesion (H) and PyX (I). F′–I′, Insets, PKC^γ-IR in both dorsal column projections in sham-lesioned mice (F′, H′) and intact contralateral dorsal columns after PyX (G′, I′). J, There was no significant difference in the number of CST axons traced between groups (two-way ANOVA with Bonferroni's post hoc comparisons, p > 0.05; data are shown as the average number of labeled CST axons ± SEM). K, There was no significant difference in the number of mCherry⁺ CST axons that crossed the spinal midline between groups (two-way ANOVA with Bonferroni's post hoc comparisons, p > 0.05; data are shown as the average number of labeled CST axons ± SEM). L, Densitometric analysis of mCherry⁺ CST axons in the DI quadrant showed that there was a main effect of Inpp5k treatment (F_(1,25) = 71.78, p < 0.0001, two-way ANOVA with Bonferroni's post hoc multiple comparisons test, ****p < 0.0001; control sham, n = 5; control PyX, n = 11; Inpp5k sham, n = 6; Inpp5k PyX, n = 7), but no effect of lesion. M, There was no significant impact of lesion or treatment on CST density in the VI quadrant. N, In the DC quadrant, there was a main effect of Inpp5k treatment (F_(1,25) = 62.8, p < 0.0001; two-way ANOVA with Bonferroni's post hoc multiple-comparisons test, ****p < 0.0001), and no effect of lesion. O, In the VC quadrant, there was a main effect of Inpp5k treatment (F_(1,25) = 67.45, p < 0.0001, two-way ANOVA with Bonferroni's post hoc multiple-comparisons test: ****p < 0.0001, ***p = 0.004). All densitometric analysis data are shown as average mCherry⁺ signal density ± SEM. CST function was assessed using the grid-walking apparatus. P, Q, No significant differences were found between Inpp5k PyX-treated and control PyX-treated subjects for the forelimbs (P; two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons, p > 0.05; data shown are the average percentage of missteps ± SEM) or the hindlimbs (Q; two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons test, p > 0.05, data shown are the average percentage of missteps ± SEM. Scale bars: B, 1 mm; D, 100 µm; I, 500 µm; I′, 100 µm.

Figure 5.

Inpp5k overexpression enhances the sprouting of intact CSNs following stroke. A, Schematic shows unilateral cortical stroke (green) and delivery of AAV1 mCherry and Inpp5k-V5 into contralateral cortex. Inset schematic shows quadrants of cord analyzed as before. B, Experimental timeline. C, D, Photomicrographs show the accumulation of DAPI⁺ nuclei delineating the stroke location (C, stippled oval) in the motor cortex and mCherry⁺ CSNs in contralateral cortex 4 weeks after lesion (D). I, The number of mCherry⁺ CST axons counted in the C4 spinal segment was invariant between groups. E–H, Photomicrographs show transverse sections through C6 showing mCherry⁺ CST axon labeling in control sham (E; n = 6), control stroke (F; n = 11), Inpp5k sham (G; n = 7), and Inpp5k stroke (H; n = 9) 4 weeks postlesion. E′, F′, G′, H′, Insets, PKC^γ-IR revealing complete unilateral denervation of CST in stroke subjects (F′, H′) and intact dorsal columns in sham subjects (E′, G′). K, Densitometric analysis in the DI quadrant showed that there was a main effect of Inpp5k treatment (F_(1,29) = 36.51, p < 0.0001; two-way ANOVA with Bonferroni's post hoc multiple-comparisons test, **p < 0.001, ****p < 0.0001). M, There was no significant impact of lesion or treatment on CST density in the VI quadrant. L, In the DC quadrant, there was a main effect of Inpp5k treatment (F_(1,29) = 27.96, p < 0.0001; two-way ANOVA with Bonferroni's post hoc multiple-comparisons test, ****p < 0.0001). N, In the VC quadrant, there was a main effect of Inpp5k treatment (F_(1,29) = 32.13, p < 0.0001; two-way ANOVA with Bonferroni's post hoc multiple-comparisons test, ****p < 0.0001). All densitometric analysis data are shown as the average mCherry⁺ signal density ± SEM. J, The number of midline crossing CST axons was also significantly elevated in mice treated with Inpp5k after stroke (F_(1,29) = 9.396, ##p = 0.005; two-way ANOVA with Bonferroni's post hoc comparisons test). O, P, Assessment of CST function using grid-walking analysis revealed that there were no significant differences between Inpp5k stroke-treated and control stroke-treated subjects in the percentage of missed steps for forelimbs (O; two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons test, p > 0.05) and hindlimbs (P; two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons test, p > 0.05, data shown are the average percentage of missteps ± SEM). Scale bars: C, 2 mm; I, 500 µm; I′, 100 µm.

Figure 6.

Inpp5k overexpression increases sprouting of CST axons after acute contusion injury. A, Schematic shows bilateral T11 contusion injury (green) and delivery of AAV1 mCherry and Inpp5k-V5 into the right side of motor cortex. Inset, Schematic shows the quadrants of cord analyzed, with laterality determined via the injection side as the lesion is bilateral, so the left side of the cord for contusion experiments is labeled contralateral. Inset shows the experimental timeline. B, C, Photomicrographs show horizontal sections through the lesion site 4 weeks after severe contusion injury in control mice (B; mCherry⁺ CST axons, red; GFAP, green; asterisks indicate lesion site) and Inpp5k-treated mice (C). No mCherry⁺ CST axons grew into or past the lesion site. H, The average number of mCherry⁺ CST axons counted in the dorsal columns was significantly different between groups (F_(1,21) = 5.215, p = 0.033, two-way ANOVA with Bonferroni's post hoc comparisons test; data shown are the average number of mCherry⁺ axons ± SEM); therefore, subsequent densitometric data were normalized to the number of CST axons per animal. D, E, F, G, Photomicrographs show transverse sections through C6 showing mCherry⁺ CST axon labeling in control sham (D; n = 5), control SCI (E; n = 9), Inpp5k sham (F; n = 4), and Inpp5k SCI (G; n = 7). D′, E′, F′, G′, Insets, PKC^γ-IR revealing intact lumbar dorsal columns for sham-lesioned mice (D′, F′), and complete absence after SCI subjects (E′, G′). J, Densitometric analysis in the DC quadrant showed that there was a main effect of Inpp5k treatment (F_(1,21) = 6.608, p = 0.0178, two-way ANOVA with Bonferroni post hoc multiple-comparisons test, *p = 0.0387). L, In the VC quadrant, there was a main effect of Inpp5k treatment (F_(1,21) = 6.776, p = 0.0166, two-way ANOVA with Bonferroni's post hoc multiple-comparisons test). K, M, There was no significant impact of lesion or treatment on CST density in the DI (K) or VI (M) quadrants. All densitometric data are shown as the average number of mCherry axons normalized compared with the number of dorsal column axons per animal ± SEM. I, There was no significant difference in the number of midline crossing CST axons between lesion and treatment groups (data shown are the average number of mCherry axons normalized to the number of dorsal column axons ± SEM). N, Behavioral assessment using the BMS revealed that there was no significance between Inpp5k-treated and control-treated SCI mice (two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons, p > 0.05; data shown are the average BMS score ± SEM). Scale bars: C, 2 mm; H, 500 µm; H′, 100 µm.

Figure 7.

Inpp5k increases CST sprouting after chronic contusion injury. A, Schematic shows bilateral T11 contusion injury (green) and delivery of AAV1 mCherry and Inpp5k-V5 into the right side of motor cortex. Inset schematic shows quadrants of cord analyzed and the experimental timeline: AVVs were delivered 28 d post-SCI. B, C, Photomicrographs show horizontal sections through the lesion site 13 weeks after severe contusion injury in control mice (mCherry⁺ CST axons, red; GFAP, green; asterisks indicate lesion site; B) and Inpp5k-treated mice (C). No mCherry⁺ CST axons grew into or past the lesion site. D, No significant differences were found in the number of mCherry⁺ CST axons counted in the C4 spinal segment between treatment groups (data shown are the average number of puncta ± SEM). B, E, Photomicrographs show transverse sections through C6 showing mCherry⁺ CST axon labeling in control SCI (E; n = 8) and Inpp5k SCI (B; n = 8). E′, F′, Insets show the complete absence of PKC^γ-IR in injured mice. H, J, K, Densitometric analysis of mCherry⁺ CST axons in gray matter showed that Inpp5k treatment was significant in the DC quadrant (H; t_(8.116) = 2.597, *p = 0.0314, unpaired t test with Welch's correction), the VC quadrant (J; t_(8.270) = 3.600, **p = 0.0066, unpaired t test with Welch's correction), and the VI quadrant (K; t_(13.88) = 5.530, ****p < 0.0001, unpaired t test with Welch's correction). I, There was no effect of Inpp5k on CST density in the DI quadrant. G, Data are shown as the average mCherry⁺ CST density ± SEM. Inpp5k increased the number of midline crossing CST axons compared with control (t₍₁₄₎ = 2.348, p = 0.034, unpaired two-tailed t test; data shown are the average number of axons ± SEM). L, Behavioral assessment using the BMS revealed that there was no significance between Inpp5k-treated and control-treated SCI mice (two-way ANOVA with repeated measures with Bonferroni's post hoc comparisons test, p > 0.05; data shown are the average BMS score ± SEM). Scale bars: C, 2 mm; E, 500 µm; E′, 100 µm.

Figure 8.

Model for the mechanism of Inpp5k-induced axon extension. PI3K is activated by growth factors binding to receptor tyrosine kinases phosphorylating PI(4,5)P2 forming PI(3,4,5)P3, initiating mTOR-dependent (rapamycin sensitive) actin polymerization and enhanced neurite outgrowth. A, Dephosphorylation of basal levels of PI(4,5)P₂ to PI(4)P by Inpp5k leads to the unbinding of a pool of active cofilin from the plasma membrane that can engage in cytoskeletal remodeling (Saarikangas et al., 2010). Two predominant hypothetical models by which cofilin regulates cytoskeletal remodeling, leading to protrusion and extension of growth cones. In model 1, cofilin binds to ATP-actin at the pointed end of filaments, depolymerizing them. This increases the availability of monomers converted to ATP-actin and polymerization by profilin. In model 2, capping proteins prevent polymerization. B, Cofilin severs F-actin creating new barbed ends for polymerization. C, LUT (look-up table) confocal images of paused looped growth cone (top left) and extending the growth cone (bottom left). Right, Diagram of morphology of microtubules (blue) and F-actin (red). In the looped paused growth cone (top right), F-actin is disorganized, leading to abnormal microtubule growth. In the extending growth cone (bottom right), F-actin is organized, which directs microtubules in a radial orientation leading to directional growth.