The mTOR Substrate S6 Kinase 1 (S6K1) Is a Negative Regulator of Axon Regeneration and a Potential Drug Target for Central Nervous System Injury

Abstract

The mammalian target of rapamycin (mTOR) positively regulates axon growth in the mammalian central nervous system (CNS). Although axon regeneration and functional recovery from CNS injuries are typically limited, knockdown or deletion of PTEN, a negative regulator of mTOR, increases mTOR activity and induces robust axon growth and regeneration. It has been suggested that inhibition of S6 kinase 1 (S6K1, gene symbol: RPS6KB1), a prominent mTOR target, would blunt mTOR's positive effect on axon growth. In contrast to this expectation, we demonstrate that inhibition of S6K1 in CNS neurons promotes neurite outgrowth in vitro by twofold to threefold. Biochemical analysis revealed that an mTOR-dependent induction of PI3K signaling is involved in mediating this effect of S6K1 inhibition. Importantly, treating female mice in vivo with PF-4708671, a selective S6K1 inhibitor, stimulated corticospinal tract regeneration across a dorsal spinal hemisection between the cervical 5 and 6 cord segments (C5/C6), increasing axon counts for at least 3 mm beyond the injury site at 8 weeks after injury. Concomitantly, treatment with PF-4708671 produced significant locomotor recovery. Pharmacological targeting of S6K1 may therefore constitute an attractive strategy for promoting axon regeneration following CNS injury, especially given that S6K1 inhibitors are being assessed in clinical trials for nononcological indications.SIGNIFICANCE STATEMENT Despite mTOR's well-established function in promoting axon regeneration, the role of its downstream target, S6 kinase 1 (S6K1), has been unclear. We used cellular assays with primary neurons to demonstrate that S6K1 is a negative regulator of neurite outgrowth, and a spinal cord injury model to show that it is a viable pharmacological target for inducing axon regeneration. We provide mechanistic evidence that S6K1's negative feedback to PI3K signaling is involved in axon growth inhibition, and show that phosphorylation of S6K1 is a more appropriate regeneration indicator than is S6 phosphorylation.

Keywords: S6K; axon regeneration; drug discovery; drug target; kinase; spinal cord injury.

Figure 1.

Ribosomal S6 protein phosphorylation is reduced by most kinase inhibitors that promote neurite outgrowth in primary hippocampal neurons. A, Fourteen small-molecule kinase inhibitors were selected, from a set of neurite outgrowth-promoting compounds, to represent distinct chemical scaffolds and diverse kinase activity profiles (Al-Ali et al., 2015). Neurons were treated with the compounds at 1 μm (or DMSO vehicle as a control) for 10 h before lysis and analysis by Western blotting. Most compounds (9 of 14) resulted in significantly reduced S6 phosphorylation (relative to DMSO control) at Serine^240/244, the substrate site for S6K1, and four others exhibited a trend toward reduced S6 phosphorylation. Data are mean ± SEM. n = 4 experimental replicates from two separate preparations of primary neurons. *p < 0.05 (one-way ANOVA with Dunnett's multiple comparisons test). **p < 0.01 (one-way ANOVA with Dunnett's multiple comparisons test). ***p < 0.001 (one-way ANOVA with Dunnett's multiple comparisons test). B, In vitro kinase activity assays for RSK1 (RPS6KA1), RSK3 (RPS6KA2), RSK2 (RPS6KA3), MSK1 (RPS6KA5), MSK2 (RPS6KA4), RSK4 (RPS6KA6), and S6K1 (RPS6KB1) with the 14 representative compounds. Increasing inhibition is indicated by increased intensity of yellow. Only four of the compounds directly inhibited catalytic activity of S6 kinases in vitro (red arrowheads).

Figure 2.

A selective S6K1 inhibitor promotes neurite outgrowth in primary hippocampal neurons. A, in vitro kinase inhibition profiling of PF-4708671 with a panel of 200 kinases (Al-Ali et al., 2015) shows activity against S6K1 and its isoform S6K2, and little to no activity against other kinases, including other members of the RPS6K family of kinases. B, Hippocampal neurons treated with 10 μm PF-4708671 for 48 h (right) show longer neurites than those treated with vehicle (left). Green represents βIII-tubulin immunostaining. Blue represents nuclear stain. Scale bar, 50 μm. C, Western blot of cell lysates from neurons treated for 1 h with PF-4708671 at the indicated concentrations. D, Red circles represent quantification of NTL of neurons treated with PF-4708671 expressed as percentage of DMSO controls. Neurons were treated at the indicated concentrations for 48 h. Data are mean ± SEM. n = 3 separate preparations of hippocampal neurons with three experimental replicates for each. Blue squares represent quantification of phospho S6 to pan S6 levels in C expressed as percentage of DMSO control. Data are mean ± SEM. E, Western blot of cell lysates from neurons treated for the indicated times with DMSO or 10 μm PF-4708671. F, Blue squares represent quantification of phospho- to pan-S6 band intensities in E and expressed as percentage of DMSO control. Data are mean ± SEM. n = 3 experimental replicates from a single preparation of primary neurons.

Figure 3.

siRNA-mediated knockdown of S6K1 promotes neurite outgrowth in primary hippocampal neurons. A, Western blot of cell lysates (3 d after transfection) from neurons transfected with S6K1-targeting or scrambled siRNAs. B, C, Quantification of A expressed as percentage of scramble control. Data are mean ± SD. D, Quantification of NTL in neurons transfected for 5 d with α-S6K1 siRNA or scrambled control. Data are mean ± SEM. n = 6 experimental replicates from a single preparation of primary neurons. **p < 0.01 (Student's t test, one-tailed). ***p < 0.001 (Student's t test, one-tailed).

Figure 4.

PF-4708671 induces PI3K signaling and stimulates mTORC1 and mTORC2 activities in primary hippocampal neurons. A, Western blot of cell lysates from neurons treated with PF-4708671, rapamycin, and Torin-2, individually or in combination for the indicated times. B–E, Quantification of band intensities in A expressed as percentage of DMSO controls (dashed lines). Data are mean ± SEM. n = 3 experimental replicates from three separate preparations of primary neurons.

Figure 5.

Treatment with mTOR inhibitors counteracts mTORC1 and mTORC2 induction by PF-4708671. A, Western blot of cell lysates from neurons treated with DMSO or PF-4708671, PF-4708671 plus rapamycin (cotreated), PF-4708671 plus Torin-2 (cotreated), rapamycin then PF-4708671 (pretreated), or Torin-2 then PF-4708671 (pretreated) at the indicated concentrations for 1 h. B, C, Quantification of band intensities in A expressed as percentage of DMSO controls (dashed lines). Data are mean ± SD. n = 2 or 3 experimental replicates from a single preparation of primary neurons.

Figure 6.

Inhibiting both mTORC1 and mTORC2 abolishes neurite outgrowth promotion by PF-4708671. A, Rapamycin treatment; B, Torin-2 treatment. Quantification of NTL from neurons treated with PF-4708671 alone, rapamycin or Torin2 alone, pretreated with rapamycin or Torin-2 before treatment with PF-4708671, or cotreated with rapamycin or Torin-2 in combination with PF-4708671, expressed as percentage of DMSO controls (dashed lines). Data are mean ± SEM. n > 15 technical replicates. *p < 0.05, treatment versus DMSO control (one-way ANOVA with Dunnett's multiple comparisons test). **p < 0.01, treatment versus DMSO control (one-way ANOVA with Dunnett's multiple comparisons test). ***p < 0.001, treatment versus DMSO control (one-way ANOVA with Dunnett's multiple comparisons test). ^##p < 0.01 (one-way ANOVA with Tukey's multiple comparisons test). ^###p < 0.001 (one-way ANOVA with Tukey's multiple comparisons test).

Figure 7.

Ribosomal S6 protein knockdown induced neurite outgrowth. A, qPCR quantification of S6 mRNA in neurons treated (24 h) with four different S6-targeting siRNAs or scrambled control (Accell siRNA oligos). α-S6 siRNAs 2–4 significantly reduced S6 mRNA levels. B, C, Quantification of NTL and length of the longest neurite (LOLN) in neurons transfected for 5 d with α-S6 siRNA, expressed as percentage relative to scrambled control (dashed line). α-S6 siRNAs 2–4 significantly increased LOLN, and α-S6 siRNAs 2–3 significantly increased NTL. Data are mean ± SEM. n = 4 experimental replicates from a single preparation of primary neurons. *p < 0.05 (one-way ANOVA with Dunnett's multiple comparisons test). **p < 0.01 (one-way ANOVA with Dunnett's multiple comparisons test). ***p < 0.001 (one-way ANOVA with Dunnett's multiple comparisons test).

Figure 8.

Experimental design of in vivo injury model and compound injection. A, Schematic drawing of the timeline, procedures, and behavior assessments. B, Schematic drawing of the tracing of dorsal (purple) and lateral (green) CST from the motor cortex down to the spinal cord. A dorsal hemisection was made at the C5/C6 level that completely transected both the dorsal and lateral CST. In this lesion model, the dorsolateral portion of the lateral funiculus remained intact to facilitate animal survival after the cervical injury. Regeneration of CST axons (red) caudal to the injury were examined up to 9 mm from the lesion site. C, D, After PF-4708671 and BDA injections at two different time points, a coronal section of the brain shows an injection site with BDA labeling (C, arrow). Neuronal nuclei (NeuN) staining of the layer V motoneurons showed that the NeuN⁺ motoneurons remained intact at the injection site (D). GFAP staining shows that the injections did not trigger strong astrocytic glial responses (arrow) at the site of PF-4708671/BDA injections (E). These observations can be further appreciated in a merged image (F).

Figure 9.

PF-4708671 promoted robust axonal regeneration of the CST through and beyond the dorsal hemisection. A, A cross section of the cervical spinal cord shows bilateral distribution of BDA-labeled main CST tracts (both dorsal and lateral) and their sprouting into the spinal gray matter. Red lines indicate sagittal/parasagittal sections spaced 125 μm apart. B, A cross section of caudal spinal cord segment 9 mm distal to the injury shows a lack of BDA labeling in the dorsal and lateral CSTs, indicating that these tracts were completely transected at C5/C6. Some regenerated CST axons were found in the gray matter of the spinal cord (arrows), indicating that these axons extended for a distance of up to 9 mm beyond the injury site. C–F, Parasagittal sections of the spinal cord, 125 μm apart, from the same animal that received 10 mm PF-4708671 show CST axonal regeneration across and beyond the lesion gap. The regenerated axons elongated within the distal spinal gray matter for considerable distances (blue arrows). G, Neurolucida reconstruction of the single section from F shows the lesion gap far beyond the main dorsal CST (between two arrows). Varicosities of regenerated CST axons extended within the distal spinal gray matter for considerable distances. C1–F1, High magnification of the lesion area, boxed in C–F, clearly shows the lesion gap (dashed line) surrounded by relatively mild astroglial responses (GFAP-IR, red, zigzag arrows). Axonal regeneration through the narrow lesion gap and beyond could be clearly seen (blue arrows). Notably, the Vibraknife cut produced a complete and deep dorsal hemisection, ensuring the transection of all CST axons at this level. Scale bars: A, B, 200 μm; C–F, 400 μm; C1–F1, 100 μm.

Figure 10.

Concentrations of PF-4708671 from 1 to 10 mm increase CST regeneration after a C5/C6 dorsal hemisection. In the DMSO control group, BDA-labeled CST axons stopped at the lesion border (DMSO control). In all three PF-4708671 treatment groups (1, 5, and 10 mm PF-4708671), numerous BDA-labeled CST axons were found to regenerate through and beyond the lesion gap and elongated within the distal spinal cord gray matter for considerable distances. High magnifications of boxed areas in representative sections show CST axonal regeneration across the lesion gap only in the PF-4708671-treated groups. Neurolucida drawings under the same representative images of the four experimental groups show detailed growth patterns of the CST axons across and beyond the lesion gap. Bottom, Quantitative analysis of BDA-labeled CST axons regenerated at different distance zones from the lesion site. In general, PF-4708671 significantly enhanced CST axonal regeneration beyond the lesion gap; 10 mm PF-4708671 promoted higher numbers of regenerative axons at different zones distal to the injury, but a statistically significant difference was only found at the 0–0.5 mm zone compared with the 1 and 5 mm PF-4708671 groups. Data are mean ± SEM. *p < 0.05 (two-way repeated-measures ANOVA with Bonferroni post-test). **p < 0.01 (two-way repeated-measures ANOVA with Bonferroni post-test). ***p < 0.001 (two-way repeated-measures ANOVA with Bonferroni post-test).

Figure 11.

PF-4708671 enhanced functional recoveries of mice after a C5/C6 dorsal hemisection. Rotarod (A) forelimb drops at grid-walking (B) adhesive removal (C, D) and pellet retrieval (E–H) show improved behavioral recoveries in groups treated with 1, 5, and 10 mm PF-4708671, compared with the control groups. At one or more time points for several tests, 10 mm PF-4708671 showed significant differences from control when 1 and 5 mm treatments did not (e.g., Rotarod at 4 and 6 weeks). Data are mean ± SEM. *p < 0.05 (two-way repeated-measures ANOVA with Bonferroni post-test). **p < 0.01 (two-way repeated-measures ANOVA with Bonferroni post-test). ***p < 0.001 (two-way repeated-measures ANOVA with Bonferroni post-test).