A sensitized IGF1 treatment restores corticospinal axon-dependent functions

SUMMARY

A major hurdle for functional recovery after both spinal cord injury and cortical stroke is the limited regrowth of the axons in the corticospinal tract (CST) that originate in the motor cortex and innervate the spinal cord. Despite recent advances in engaging the intrinsic mechanisms that control CST regrowth, it remains to be tested whether such methods can promote functional recovery in translatable settings. Here we show that post-lesional AAV-assisted co-expression of two soluble proteins, namely insulin-like growth factor 1 (IGF1) and osteopontin (OPN), in cortical neurons leads to robust CST regrowth, and the recovery of CST-dependent behavioral performance after both T10 lateral spinal hemisection and a unilateral cortical stroke. In these mice, a compound able to increase axon conduction, 4-aminopyridine-3-methanol, promotes further improvement in CST-dependent behavioral tasks. Thus, our results demonstrate a potentially translatable strategy for restoring cortical dependent function after injury in the adult.

eTOC Blurb

Liu et al. showed that post-lesional AAV-OPN/IGF1 treatment leads to robust regrowth of corticospinal axons and relevant behavioral recovery in both spinal cord injury and cortical stroke models, demonstrating a potentially translatable strategy for restoring cortical function in the adult.

INTRODUCTION

The axons of the corticospinal tract (CST) originate from corticospinal neurons (CSNs) in layer 5 of the motor and somatosensory cortex and innervate all segments of the spinal cord. The CST transmits cortical commands to the spinal cord, allowing willful intention to be translated into observable action. Disruption of CSNs and/or CST axons results in motor functional deficits after traumatic injuries like spinal cord injury and stroke. Therefore, a logical therapeutic approach is to promote CST regrowth in a hope to rebuild functional connections (Maier and Schwab, 2006; Ratan and Noble, 2009; Bradke et al., 2012, Tuszynski and Steward, 2012; Chen and Zheng, 2014; Jin and He, 2016; Carmichael et al., 2017). In general, recovery could be achieved either by regenerative growth of injured CST axons across the lesion site, or by compensatory sprouting of spared axons that innervate the denervated areas. For both types of regrowth, the limited growth ability of adult CSNs is a formidable impediment (Maier and Schwab, 2006; Tuszynski and Steward, 2012; Chen and Zheng, 2014; Jin and He, 2016; Carmichael et al., 2017).

In exploring the molecular mechanisms that control the growth ability of CSNs, several important regulators have been identified, such as mTOR/PTEN (Liu et al., 2010; Zukor et al., 2013; Du et al., 2015), STAT3/SOCS3 (Lang et al., 2013; Jin et al., 2015), KLFs (Blackmore et al., 2012) and Sox11 (Wang et al., 2015). We and others have shown that CSNs undergo a development-dependent and injury-triggered decline of mTOR activity and that activating this pathway by inhibiting the expression of its negative regulator PTEN elicits the regrowth of the adult CST after injury (Liu et al., 2010; Zukor et al., 2013; Lewandowski and Steward, 2014; Jin et al., 2015; Danilov and Steward, 2015; Geoffroy et al., 2015). However, because PTEN is a tumor suppressor, clinical application may require other alternative methods to elevate the growth ability of CSNs.

In seeking such alternatives, it is relevant that the PI3K/mTOR pathway plays several roles, one of which is to mediate the activity of neurotrophins and other growth factors. In cultured neonatal CSNs, insulin-like growth factor 1 (IGF1) and brain derived neurotrophic factor (BDNF) are able to promote the growth and branching of CST axons, respectively (Ozdinler and Macklis, 2006). However, direct administration of these factors has limited effects on promoting CST regrowth in adults (Giehl & Tetzlaff, 1996; Lu et al., 2001; Hollins et al., 2009; Li et al., 2010), suggesting that in comparison to immature neurons, mature CSNs have reduced responsiveness to growth factors. Hence, it would be desirable to develop a sensitizing strategy that enhances the response of mature CSNs to growth factors. A possible means to this end is suggested by our recent studies of optic nerve injury (Duan et al., 2015; Bei et al., 2016). We showed that although IGF1 or BDNF alone failed to promote regeneration, combining either trophic factor with osteopontin (OPN) allowed injured retinal ganglion cells to respond to these growth factors, exhibiting robust axon regeneration in an mTOR-dependent manner (Duan et al., 2015; Bei et al., 2016). However, it remains to be tested whether OPN can sensitize the responses of other types of neurons to these growth factors.

Because of the unique importance of CST axons in controlling spinal cord function, we asked here whether OPN can sensitize CSNs’ responses to growth factors and promote CST regrowth and relevant functional recovery in clinically relevant injury models. Because multiple pathways are interrupted by such injuries, we first defined the behavioral defects that result from damage to the CST. Then, we tested the combinatorial treatment of OPN and IGF1 in two different CST-related injury models, spinal cord T10 lateral hemisection and unilateral cortical stroke. We discovered that this treatment indeed promoted robust CST regrowth with significant yet partial restoration of CST-dependent tasks. We also show that an additional treatment of 4-aiminopyridine- 3-methanol (4-AP-MeOH) in these mice further increases functional recovery. Our results reveal a possibly translatable strategy of promoting CST-dependent functional restoration in adults.

RESULTS

Characterization of CST-dependent Hindlimb Behavioral Deficits after T10 Lateral Hemisection

In the majority of patients of spinal cord injury or stroke, some CST axons are spared, leading to an “incomplete” injury (Raineteau & Schwab, 2001). Accordingly, an ideal model should involve incomplete injury with characterized tract-specific behavioral defects. We used thoracic lateral hemisection, because the resultant lesions and behavioral deficits are reproducible and quantifiable (Figure 1A, Ballermann and Fouad, 2006; Courtine et al., 2008; Takeoka et al., 2014). Furthermore, it is known that different from other types of descending axons (Tetzlaff et al., 1994; Ballermann and Fouad, 2006; Courtine et al. 2008; Alilain et al., 2011; Ruschel et al., 2015), CST axons have limited ability of exhibiting midline crossing after injury in the adult mice (Fouad et al., 2001; Weidner et al., 2001; Bareyre et al., 2004; Liu et al., 2010). Thus we characterized CST-related anatomical and behavioral features of adult mice following T10 spinal cord lateral hemisection. Within weeks after injury, several descending axon tracts (Ballermann and Fouad, 2006; Courtine et al., 2008; Takeoka et al., 2014), including reticulospinal, propriospinal and serotonergic (Figures 1B and 1C), on the intact side sprouted across the midline and innervated the denervated side of the spinal cord. In contrast, CST axons had limited sprouting in the same mice (Figures 1B and 1C).

Figure 1. Functional Deficits after Spinal Cord T10 Lateral Hemisection and Hindlimb CSN Ablation.

(A) Upper: A Cartoon of the T10 lateral hemisection. Bottom: A representative transverse section at epicenter of the lesion stained with anti-GFAP. Scale bar: 500 μm.

(B) Left: Cartoon of a spinal cord at 1 wk (upper) and 12 wk (bottom) post T10 lateral hemisection. Blue and purple lines represent descending serotonergic and corticospinal tracts, respectively. Right: Representative images of transverse sections of the lumbar spinal cord (L3) at 1 wk (upper lane) and 12 wk (bottom lane) post T10 lateral hemisection stained with anti-5-HT (serotonergic axons, left column) or anti-RFP (corticospinal axons, right column), respectively. To label corticospinal axons, animals received bilateral cortical AAV-ChR2-mCherry injection at 2 wk pre-injury. Scale bar: 500 μm.

(C) The fluorescence intensity of 5-HT or mCherry immunostaining at 1 wk and 12 wk post the spinal cord T10 lateral hemisection. All images were acquired using identical optical parameters and scan setting. In each case, the intensities were normalized to 1 wk post injury. ** and n.s., p < 0.01 and no statistical significance respectively. Student’s t-test. n=3 mice per group. Five sections at L3 were quantified per mouse.

(D–E) Performance on irregular ladder walking post T10 lateral hemisection of the hindlimb from the intact (D) and denervated sides (E). ** and n.s.: p < 0.01 and no statistical significance, n = 7, repeated measures ANOVA followed by post hoc Bonferroni correction.

(F) Schematic diagram of the experimental timeline. Emx1-Cre mice were intraspinally (at the spinal cord T12-L4 segments) injected with HiRet-FLEX-DTR/-GFP (control) at P12–P14, measured for behavioral baseline at P56, injected with diphtheria toxin (DT, i.p.) at P60, re-measured for behavioral performance at P70, P80, respectively before terminal histological analysis.

(G) Representative images of transverse sections of the dorsal spinal cord at cervical (C6) and lumbar (L3 and L5) levels stained with PKCγ in HiRet-GFP (Left column) or HiRet-FLEX-DTR (right column) injected Emx1-Cre mice after DT treatment. Arrowheads indicate the location of the main CST at the dorsal funiculus. Open arrowheads indicate ablation of the CST. Scale bar: 500 μm.

(H) Performance on irregular ladder walking of the forelimb (upper) and hindlimb (bottom) at −1, 10, and 20 days post DT injection in HiRet-GFP and HiRet-FLEX-DTR injected Emx1-Cre mice. For both forelimb and hindlimb, error rates were averaged from both sides. ** and n.s.: p < 0.01 and no statistical significance, Student’s t test, n=6, 5 for HiRet-GFP or HiRet-FLEX-DTR injected mice, respectively.

Next, we performed a variety of behavioral tests to define the functional deficits due to T10 lateral hemisection (Figures S1B–S1J, 1D and 1E). Consistent with previous reports (Ballermann and Fouad, 2006; Courtine et al., 2008; Takeoka et al., 2014), the denervated hindlimbs showed severe locomotor deficits during the first week after injury (Figures S1B–1H). Over the following weeks, recovery varied dramatically among tasks. For example, inter-limb coordination recovered almost completely (Figures S1E–1H); weight support, protraction during walking, and speed tolerance on treadmill recovered partially (Figures S1B–S1D and S1I), and paw dragging of the denervated hindlimb on a treadmill showed no significant recovery (Figure S1J).

In addition, we tested the ability of mice to walk on a horizontal ladder with irregularly spaced rungs (Metz and Whishaw, 2002; Carmel et al., 2010, 2014; Jin et al., 2015). In this task, mice need to continuously adjust their stepping movements by aiming their limbs towards a new rung and then perform an accurate placement. The hindlimbs on the intact side showed a transient defect but then achieved almost full recovery (Figures 1D and S1K). In contrast, the injured hindlimb often missed the rung (miss error) or contacted the rung with a few digits followed by a slip (slip error), thereby resulting in a significantly higher error rate even at 12 weeks post injury (Figures 1E and S1L). Thus, despite spontaneous axonal reorganization and functional recovery, the mice with T10 lateral hemisection failed to show significant CST sprouting and still exhibited significant defects in several locomotor parameters, including paw dragging, speed tolerance on treadmill and precision placement on irregular ladder.

To investigate whether any of these persistent defects following T10 lateral hemisection resulted from loss of the CST, we selectively ablated the CSNs, which give rise to the CST axons that innervate the hindlimb, without damaging other descending tracts. To do this, we took advantage of the efficient retrograde properties of a pseudotyped lentiviral vector (HiRet) (Kinoshita et al., 2012). In a pilot experiment, we injected GFP-expressing vector (HiRet-GFP) into the spinal cord segments (T12-L4) of mice at the age of postnatal day 12–14 (P12–14) when most CST axons have reached their spinal targets (Bareyre et al., 2005) and the pruning of cortical projections is almost complete (O’Leary, 1992; O’Leary and Koester, 1993). As shown in Figures S2A, CSNs in the hindlimb area of the primary sensorimotor cortex were efficiently targeted. To examine the retrogradely labeling efficiency, we co-injected HiRet-mCherry with another commonly used retrograde tracer: fluorescent microspheres (Kamiyama et al., 2015) and found that about 98% of green retrobeads labeled CSNs were co-labeled by mCherry, indicating a high efficiency of HiRet vectors in labeling CSNs (Figure S2B). Next, we injected HiRet vectors carrying flip-excision (FLEX) human diphtheria toxin receptor (DTR; HiRet-FLEX-DTR) into the T12-L4 spinal cord of cortex-specific Emx1-Cre transgenic mice (Bareyre et al., 2005) (Figure 1F). Upon diphtheria toxin (DT) administration, CST axons were efficiently ablated in the lumbar, but not cervical, spinal cord as indicated by immunostaining for PKCγ a CST marker (Liu et al., 2010) (Figures 1G, and S2G). Consistent with this, the retrograde tracing at the lumbar spinal cord revealed less than 10% remaining CST axons in HiRet-FLEX-DTR injected mice (Figures S2C–S2E). The DT-mediated CSNs ablation did not induce significant inflammatory response (Figures S2D, S2F). In these mice, basic locomotor functions, like weight support, excursion length, and limb coordination during ground walking were unperturbed (Figure S3A–S3D), consistent with the previous findings that basic locomotor function achieves almost complete recovery after CST lesions (Metz et al., 1998; Muir & Whishaw, 1999). In addition, these mice are capable of adjusting gait patterns without overt paw dragging towards changing velocities when stepping on the treadmill (Figures S3E–S3G).

On the other hand, similar to the mice with lateral hemisection (Figures 1D and 1E), the performance of the hindlimbs, but not the forelimbs, of these mice is defective in the irregular walking assay permanently (Figures 1H and S3H). Based on these results, we propose that the defects in the irregular walking task, but not the paw dragging and speed tolerance on the treadmill, after T10 hemisection are likely due to the lack of regrowth of spared CST axons and thus persistent CST denervation on the injured side of the lumbar spinal cord.

OPN Sensitizes Adult CSNs to IGF1

Previous studies showed that IGF1 promotes axon growth by activating both PI3K and Erk/MAPK pathways in cultured CSNs isolated from neonatal mice (Ozdinler and Macklis, 2006). However, it is unknown whether IGF1 triggers similar signaling activation in adult CSNs and, if so, whether it could be altered by OPN treatment. Thus we labeled CSNs by spinal injection of HiRet-GFP, performed T10 lateral hemisection, and then bilaterally injected recombinant IGF, OPN or OPN/IGF1 proteins into the cortex (Figure 2A). In 3 days, the cortical sections from both injured and intact sides of adult mice were immunostained with anti-IGFR, anti-phospho-IGFRβ, an indicator for the activation of the IGF-1 mediated signaling pathways (Hernández-Sánchez et al., 1995; Siddle, 2012), and anti-phospho-S6 (p-S6), an established marker of mTOR activation, one of the downstream target of IGF1 mediated signaling pathways (Liu et al., 2010; Pollak, 2008).

Figure 2. OPN Sensitizes the Responsiveness to IGF1 in Adult CSNs.

(A) Schematic diagram of the experimental timeline. Wild type mice received intraspinal injection (at the spinal cord T12-L4 segments) with HiRet-GFP at P12–P14, T10 spinal cord lateral hemisection at P60, bilateral cortical injection of saline, soluble human recombinant IGF1, OPN or OPN/IGF1 at P74, and were sacrificed at P77 for histological analysis.

(B) Representative images showing intact and axotomized CSNs (green: retrogradely labeled by GFP) co-stained with IGFR, pIGFR, and pS6 in saline (top lanes), IGF1 (middle lanes), and OPN & IGF1 protein (bottom lanes) injected animals. Arrowheads indicate the co-localization of GFP with IGFR, pIGFR, and pS6 immunofluorescence. Scale bar: 20 μm.

(C,D,E) Quantification of immunofluorescence intensities of IGFR, pIGFRβ and pS6 in various conditions. All images were taken using identical optical parameters and scan settings. In each case, the intensities were normalized to that in intact CSNs with saline injection. ** and n.s., p < 0.01 and no statistical significance. One way-ANOVA, followed by post hoc Bonferroni correction. For IGFR quantification, n = 122, 107, 123, 126, 94, 93, 121, 113 for the intact and axotomized CSNs in saline (n = 3), IGF1 (n = 3), OPN (n=3) and OPN/IGF1 (n = 3) injected animals, respectively. For pIGFRβ quantification, n = 104, 119, 107, 122, 104, 96, 126, 118 for the intact and axotomized CSNs in saline (n = 3), IGF1 (n = 3), OPN (n=3) and OPN/IGF1 (n = 3) injected animals, respectively. For pS6 quantification, n = 130, 131, 130, 134, 119, 108, 133, 139 for the intact and axotomized CSNs in saline (n = 3), IGF1 (n = 3), OPN (n=3) and OPN/IGF1 (n = 3) injected animals, respectively. Note that the images of OPN protein injected group were present in Figure S4A.

Levels of IGFR were comparable in intact or axotomized CSNs (Figures 2B, S4A, and 2C). The combination of IGF1 and OPN, but neither IGF1 or OPN alone, slightly increased (around 1.6 fold) the IGFR levels in both intact and axotomized CSNs (Figures 2B, S4A, and 2C). In contrast, levels of p-IGFRβ were barely detectable in intact or axotomized CSNs (Figures 2B, S4A, 2D). While treatment with IGF1 or OPN alone had no overt effect on the p-IGFRβ in either intact or axotomized, mature CSNs, the combination of IGF1 and OPN significantly increased (around 6 fold) the phosphorylation level of IGFRβ (Figures 2B, S4A, and 2D), indicating an activation of the IGFR mediated signaling pathways. Furthermore, treatment with OPN and IGF1, but not IGF1 or OPN alone, significantly increased p-S6 levels in both intact and axotomized CSNs (Figures 2B, S4A, and 2E). Together, these results suggest that in both intact and injured CSNs, OPN is able to enhance the response to IGF1.

OPN/IGF1 Improve the Precision Position Performance and CST Regrowth After T10 Lateral Hemisection

In contrast to unstable proteins, injected AAVs could express OPN and IGF1 for a few weeks. Thus we examined the effects of AAV-assisted OPN/IGF1 treatment on CST regrowth and functional recovery in mice with T10 lateral hemisection. As a first step, we verified that AAV-mediated OPN/IGF1 treatment achieved similar level of IGFR and mTOR activation (Figures S4B–S4D). To mimic clinical conditions, AAVs expressing placental alkaline phosphatase (PLAP, control) or OPN and/or IGF1 were stereotaxically injected to the sensorimotor cortex bilaterally 1 day after injury (Figure 3A). These mice were subjected to behavioral tests biweekly in a double-blind manner. On the irregular walking task, the hindlimb performance of the intact side was similar in both control and treated groups, recovering spontaneously over 4–8 weeks (Figures 3B and 3C). However, there was a significant difference in the hindlimb performance of the denervated side in the double treated group from 8 weeks after injury (Figures 3D and 3E). In the group with treatment of both OPN and IGF1, the error rates dropped from about 86% at 1 week after injury to 46% at 12 weeks after injury, in contrast to a drop only ~70% in controls and groups treated with either OPN or IGF1 treatment (Figures 3B–3E). The improvements appeared to be specific to CST-mediated behaviors, in that mice in the OPN/IGF1 treated group failed to show significant improvements in weight support, hindlimb protraction of the denervated side during ground walking or speed tolerance and hindlimb retraction (denervated side) during treadmill walking when compared to other groups (Figures S5A, S5B).

Figure 3. OPN/IGF1 Treatment Improves Precision Performance after Spinal Cord T10 Lateral Hemisection.

(A) Schematic diagram of the experimental timeline. Wild type mice received baseline behavioral measurement at P56, T10 spinal cord lateral hemisection at P60, bilateral cortical injection of AAV-PLAP, AAV-OPN, AAV-IGF1 or AAV-OPN/IGF1 at P61, and biweekly behavioral measurement from P62–P145 before terminal histological analysis.

(B–E) Performance on irregular ladder walking of the hindlimb from the intact (B, C) and denervated (D, E) sides in AAV-PLAP, AAV-OPN, AAV-IGF1, and AAV-OPN/IGF1 treated animals. C and E: Hindpaw placement categories (breakdown of miss, slip, and hit) on irregular ladder walking. **: p < 0.01 and p < 0.05, respectively. Repeated measures ANOVA followed by post hoc Bonferroni correction. n = 13, 9, 8, and 12 for AAV-PLAP, AAV-OPN, AAV-IGF1, and AAV-OPN/IGF1 treated groups respectively. Hit, miss, and slip in (C) and (E) represent three categories of hindpaw placement on the rungs during walking across the ladder.

Since the mice treated with AAV-OPN/IGF1 were co-injected with AAV-channelrhodopsin (ChR2)-mCherry (AAV-ChR2-mCherry) or AAV-ChR2-YFP to the ipsilateral or contra-lateral cortex, respectively, at the termination of behavioral assessment, we first performed a post hoc examination of the T10 lateral hemisection histology. Mice with incomplete- or over-hemisection could be readily identified by analyzing labeled CST axons at the lumbar spinal cord and excluded from further analysis (Figure S1A). We then examined CST regrowth by preparing horizontal spinal cord sections covering the lesion sites to assess axon regeneration across the lesion site, and transverse sections of lower spinal cord segments to assess midline-crossing of CST axons from intact side. In controls and treatment with either IGF1 or OPN groups, intact axons showed little sprouting across midline (Figures 4A and 4C) and injured axons showed significant die-back from the lesion site (Figures 4B and 4D). In contrast, significant numbers of injured axons regrew across the lesion in the OPN-IGF1 treated group (Figures 4B and 4D). However, while many of these axons projected for over 1 millimeter beyond the lesion (Figures 4B and 4D), they failed to reach the lumbar spinal cord (Figure 4A). Importantly, in these mice with AAV-OPN and IGF1 treatment, CST axons from intact side exhibited significant sprouting into the denervated side, which was observed across all spinal levels below the injury site (Figures 4A and 4C). These sprouted axons terminated broadly in different laminae at the lumbar spinal cord and formed bouton-like structures (Figure S6B), as indicated by co-localization with vGlut1, a presynaptic marker for excitatory synapses (Maier et al., 2008; Liu et al., 2010) (Figures S6A and S6B). Thus, it is likely that these sprouting axons play an important role in the observed functional recovery. Therefore, combined treatment of AAV-OPN and AAV-IGF1 not only promotes CST regrowth but also CST-dependent behavioral performance.

Figure 4. OPN/IGF1 Treatment Promotes CST Regrowth After Spinal Cord T10 Lateral Hemisection.

(A) AAV-ChR2-mCherry (red) or AAV-ChR2-YFP (green) was injected to the intact (assessing collaterally sprouted axons) or axotomized (assessing regenerating axons) cortex. Representative images in (A) of transverse sections of the lumbar spinal cord (L3) stained with anti-RFP and anti-GFP at 13–14 wk post spinal cord T10 lateral hemisection in AAV-PLAP, AAV-OPN, AAV-IGF1, and AAV-OPN/IGF1 treated groups Notice that no GFP⁺ axons were detected at the denervated lumbar spinal cord. Scale bar: 500 μm.

(B) Representative images of horizontal sections of the spinal cord around the lesion site (marked with star) stained with anti-GFP to mark regenerating axons at 13–14 week post spinal cord T10 lateral hemisection in each group. Scale bar: 1 mm.

(C) Quantification of midline crossing axons counted in different regions of the lumbar spinal cord (L3 and L5) for each treatment group. The number was normalized against that of labeled CST axons counted at the pyramidal level (see the method section) in each condition. The schematic drawing on the upper-right corner illustrates the division of different regions of the spinal cord. Mid, midline; Z1 and Z2, different lateral positions. ** and *, p< 0.01 and p < 0.05, One-way ANOVA followed by Bonferroni post hoc test. n = 3 per group. Five sections at L3 and L5 were quantified per mouse.

(D) Quantification of labeled axons in the spinal cord caudal to the lesion site in each treatment group. *, p < 0.05, repeated measures ANOVA followed by post hoc Bonferroni correction, n = 3 mice per group. Five sections were quantified per mouse.

Further Improvements in Precision Performance in the OPN/IGF-treated Mice by 4- aminopyridine-3-methanol (4-AP-MeOH) Treatment

Because mice treated with OPN and IGF1 showed partial improvement on the irregular ladder walking task (Figure 3), we next tested whether additional treatments could further improve their performance. As different neuro-modulatory small molecule compounds have been shown to impact locomotor function in SCI models (Fong et al., 2005; Courtine et al., 2009; Murray et al., 2010), we tested several commonly used modulators in the injured mice with AAV-OPN/IGF1 or AAV-PLAP treatments at 14 weeks after injury. Previous studies showed that systemic administration of a cocktail of serotonin receptor agonists and dopamine receptor agonists facilitates the transformation of lumbosacral circuits from dormant to highly functional states after injury (Musienko et al., 2011, van den Brand et al., 2012). However, quipazine, (a 5-HT_2A agonist), 8-OH-DPAT (a 5-HT_1A agonist), and SKF (a D1/D5 agonist) failed to improve their performance in irregular ladder, overground, or treadmill walking (Figures 5A, 5B, S5C and S5D), with one (SKF) even deteriorated the hindlimb performance in irregular ladder walking on the denervated side (Figure 5B).

Figure 5. 4-AP-MeOH Treatment Further Enhances Precision Performance in OPN/IGF1 Treated Animals.

(A) Schematic diagram of the experimental timeline. Wild type mice received baseline behavioral measurement at P50–P56, T10 spinal cord lateral hemisection at P60, bilateral cortical injection of AAV-PLAP or AAV-OPN/IGF1 at P61, pharmacological treatment at P160 and re-measurement of behavioral tests immediately after the treatment.

(B–C) Performance on irregular walking of hindlimbs from both intact and denervated sides in AAV-PLAP and AAV-OPN/IGF1 treated groups with systematic administration of saline, 5-HT receptor agonists (B, quipazine and 8-OH-DPAT), dopamine receptor agonist (B, SKF), 4-AP (C) and 4-AP-MeOH (C). n.s. and *, no statistical differences, and p < 0.05, respectively. One-way ANOVA followed by post hoc Bonferroni correction (to the saline treated group), n= 7, and 8 for AAV-PLAP and AAV-OPN/IGF1 treated groups.

Our previous studies in an optic tract injury model showed that 4-aminopyridine (4-AP) or its derivative 4-aminopyridine-3-methanol (4-AP-MeOH), both voltage-gated potassium channel blockers (Bostock et al., 1981; Sun et al., 2010), were able to improve nerve conduction of regenerated retinal axons and led to behavioral improvements in a visual task (Bei et al., 2016). Thus, we tested whether these compounds could improve performance in locomotor tasks. As shown in Figures S6E–S6F, neither 4-AP nor 4-AP-MeOH had significant effects on the recovery of treadmill walking performance when compared to the control. However, on the irregular ladder walking task, 4-AP at the dose of 1 mg/kg showed a trend of reducing the error rates in mice with OPN/IGF1 treatment, but failed to reach statistical significance (Figure 5C). We did not obtain reliable results with higher doses of 4-AP treatments because of seizures as a result of such treatments in injured mice. However, in the OPN/IGF1 treated group, at the dose of 1 mg/kg, 4-AP-MeOH treatment significantly reduced the error rate from about 48% to 30% (Figure 5C), approaching to the intact hindlimb performance (error rate about 20%). These results suggest that improving the conduction of regrowing CST axons can facilitate the recovery of CST-dependent skilled locomotor function.

AAV-OPN/IGF1 Stimulates the Sprouting of Cortical Axons and Improves Functional Recovery in a Cortical Stroke Model

Results from T10 hemisection prompted us to test whether OPN/IGF1 treatment might be beneficial in cortical stroke models in which CSNs and their axons are disrupted. With a photothrombosis-based protocol, previous studies have established a reproducible cortical infarction that destroys the sensorimotor cortex unilaterally leading to deficits in skilled locomotor function (Watson et al., 1985; Li et al., 2015; Wahl et al., 2014). We optimized this procedure in adult mice (Figure S7A) and showed that it resulted in consistent lesion of the sensorimotor cortex, as evident by both TTC staining (3 days post lesion) and Nissl staining (12 weeks post lesion) (Figure S7B). Behaviorally, these lesioned mice exhibited significant unilateral defects on the irregular ladder walk task for both forelimbs and hindlimbs (Figure S7C), and the food pellet retrieval task for the forelimbs (Figure S7D), which has been shown to be highly relevant to CST function (Farr and Whishaw, 2002, Wahl et al., 2014). In contrast, mice gained almost full functional recovery over ground locomotion (Figure S7E).

To test whether OPN/IGF1 treatment could promote recovery following stroke, we injected AAVs expressing OPN/IGF1 (treatment group) or PLAP (control) to the intact sensorimotor cortex at 3 days after photothrombotic cortical lesion (Figure 6A). Treatment with OPN/IGF1 resulted in significant recovery in both behavioral assays, starting from 8 weeks after injury (Figures 6B and 6C). Furthermore, similar to treated mice with T10 lateral hemisection (Figure 5), the addition of 4-AP-MeOH resulted in further improvements on both assays (Figure 6D) in these ischemic mice treated by OPN/IGF1.

Figure 6. OPN/IGF1 Treatment Improves Skilled Locomotion after Unilateral Cortical Stroke.

(A) Schematic diagram of the experimental timeline. Wild type mice received baseline behavioral training at P56, unilateral cortical photothrombotic stroke at P60, unilateral cortical injection (the intact side) of AAV-PLAP or AAV-OPN/IGF1 at P63, biweekly behavioral tests from P67–P145 before terminal histological analysis.

(B–C) Performance on single pellet retrieval task (B) and irregular ladder walking (C) respectively. * and **, p < 0.05 and p< 0.01, Repeated measures ANOVA with Bonferroni post hoc correction, n = 8, 9 for AAV-PLAP, AAV-OPN/IGF1 injected animals.

(D) Performance on single pellet retrieval task and irregular walking in AAV-PLAP and AAV-OPN/IGF1 treated groups with systematic administration of 4-AP-MeOH. *, p < 0.05, Student’s t -test. n = 8, 9 for AAV-PLAP, AAV-OPN/IGF1 injected animals.

At 12 weeks after the lesion, these mice were subjected to anatomical analysis. As AAV-ChR2-mCherry vectors were co-injected with AAV-OPN/IGF1 or AAV-PLAP to the intact side of the cortex (Figure 7A), we could monitor the projection patterns of the labeled cortical axons. As shown in Figures 7 and S8, AAV-ChR2-mCherry-labeled axons from the intact, treated sensorimotor cortex showed increased sprouting not only in the cervical and lumbar spinal cord (Figures 7B and 7E), but also in other subcortical regions such as the medullary reticular formation (both ventral and dorsal parts; MdV & MdD, bilaterally), the spinal trigeminal nucleus (Sp5O, bilaterally) and the gigantocellular reticular nucleus (GiV, bilaterally) in the brainstem, the ipsilateral (relative to the AAV-mCherry injected side) pontine nucleus (PnO), the contralateral red nucleus (R) and ipsilateral superior colliculus (SC) in the midbrain, and the striatum (St) beneath the lesioned cortex (Figures 7C, 7D and S8).

Figure 7. OPN/IGF1 Treatment Promotes CST axon Sprouting in the Spinal Cord and Subcortical Areas after Unilateral Cortical Stroke.

(A) Schematic drawing of the experimental procedure. The cortical injection (intact side) of AAV-PLAP (Left) or AAV-OPN/IGF1 (right), along with AAV-ChR2-mCherry was performed 3 days after the unilateral photothrombotic stroke. Notice the collateral sprouting occurred at the striatum, the midbrain, the brain stem, and the spinal cord from the corticospinal/corticofugal axons after the injury in control (AAV-PLAP) and experimental (AAV-OPN/IGF1) conditions.

(B–C) Representative images of serial transverse sections at the spinal cord (C6 and L3, B), the midbrain, and the brain stem (C) stained with anti-RFP in AAV-PLAP and AAV-OPN/IGF1 treated animals. Scale bar in B: 500 μm, in C: 1 mm. Aq: cerebral aqueduct; mRT: mesencephalic reticular formation; GiV: gigantocellular reticular nucleus; Sp5O: spinal trigeminal nucleus; and Py: pyramidal tract.

(D) Quantification of fluorescence intensity of corticofugal projections from the intact side at multiple subcortical areas in AAV-PLAP and AAV-OPN/IGF1 treated animals. For each subcortical position, fluorescence intensity was normalized to that of the cortical area injected with AAV-ChR2-mCherry. St: striatum; R: red nucleus, PnO: pontine reticular nucleus; GiV: gigantocellular reticular nucleus; Sp5O: spinal trigeminal nucleus; MdD and MdV: medullary reticular formation, dorsal and ventral parts. Note: images of St, PnO, MdD and MdV from AAV-PLAP or AAV-OPN/IGF1 injected animals are in Supplementary Figure 8. **, p< 0.01, Student’s t-test. n = 3 mice per group. Five sections at C7 and L3 were quantified per mouse.

(E) Quantification of midline crossing axons counted in different regions of the cervical and lumbar spinal cord (C6 & L3) in AAV-PLAP and AAV-OPN/IGF1 injected groups. ** and *, p< 0.01 and p < 0.05, Student’s t-test. n = 3 mice per group. Five sections at C6 and L3 were quantified per mouse.

Contribution of CST Sprouting in the Cervical Spinal Cord to the Functional Recovery Induced by OPN/IGF1

The observed axon sprouting in subcortical regions such as the red nucleus and brainstem might relay the cortical signal to the denervated spinal cord raised the possibility that these new pathways could mediate functional recovery instead of or along with connections resulting from sprouting in the spinal cord (García-Alías et al., 2015). To assess the contribution of sprouted CST axons in the spinal cord to the observed functional recovery, we analyzed the effects of ablating CSNs that send collaterally sprouted axons to the denervated side of the cervical spinal cord (C5–C7) using a viral vector-assisted intersectional targeting strategy (Kinoshita et al., 2012; Wahl et al., 2014). With an optimized stereotaxic injection protocol (Jin et al., 2015), we first unilaterally injected pseudotyped HiRet-FLEX-DTR into the denervated side of the cervical spinal cord (C5–C7) at 14 weeks post injury (Figure 8A) in a set of adult mice with AAV-OPN/IGF1 treatment as described above. 3 days later, AAV-Cre (Ablation) or AAV-PLAP (Control) was then injected into the unlesioned side cortex (Figure 8A). 2 weeks later, DT was administrated intraperitoneally. By doing this, only Cre⁺ CSNs that sprouted midline-crossing axons into the cervical, but not lumbar, spinal cord would express DTR, which would be ablated by DT injection. We verified that behavioral performance was unaltered by these intraspinal and cortical injections (Figures 8B and 8C).

Figure 8. Ablation of Sprouted CSNs Abolishes Recovered Skilled Locomotor Performance.

(A) Schematic drawing of the experimental procedure. Mice that received AAV-OPN/IGF1 treatment after stroke were subjected to intraspinal lentivirus (HiRet-FLEX-DTR) injection on the denervated side of the cervical spinal cord (C5–C7) at P160 (1), and cortical AAV-Cre (ablation) or AAV-PLAP (control) injection (intact side) at P163 (2). At P175, DT was administrated (i.p.) (3).

(B, C) Performance on the single pellet retrieval task (B) and irregularly horizontal ladder (C), respectively. Notice that for irregular horizontal ladder walking test, the performance of both intact and denervated sides of forelimbs and hindlimbs was analyzed. *, p < 0.01 or 0.05, n.s., not significant, Student’s t-test. n = 6 and 5 for AAV-Cre (ablation) or AAV-PLAP (control) injected group, respectively.

(D) Representative images of transverse spinal cord sections at C6 and L3 immunostained with anti-RFP to label the CST axons originated from the intact side in cortical AAV-PLAP (control) or AAV-Cre (ablation) injected animals. Scale bar: 500 μm.

(E–F) Quantification of midline crossing axons (left) and axon density of the intact side (right) in the cervical (E) and lumbar (F) spinal cord (C6 & L3) in cortical AAV-PLAP (control) or AAV-Cre (ablation) injected groups. **, p=0.06, and n.s.: p< 0.01, borderline p value, and no statistical significance, respectively. Student’s t-test. n = 3 mice per group. Five serial sections at C6 and L3 were quantified per mouse.

However, at two weeks after DT administration, the improved performance by OPN/IGF1 treatment on single pellet retrieval task and irregular ladder walking of the denervated forelimb significantly declined (Figures 8B and 8C). Such ablation-induced behavioral decline was seen only in the forelimbs, but not the hindlimbs (Figure 8C). Consistently, the ablation of CST axons was seen in the denervated side of the cervical, but not lumbar, spinal cord (Figures 8D–8F), likely due to the fact that HiRet-FLEX-DTR was selectively injected to the cervical spinal cord. On the other hand, the performance of the intact forelimb on irregular ladder walking showed a decreasing trend, although without statistical difference (Figure 8C). In this regard, we found that the CST axons on the intact side of the cervical spinal cord were also reduced (Figures 8D–8F), consistent with the notion that the CST axons in the denervated side were primarily sprouted from the intact side of the spinal cord. Thus, although we cannot rule out a contribution of sprouting axons in the subcortical regions for functional improvement, our results suggest that the sprouted axons in the spinal cord is required for the recovery of skilled motor performance after unilateral photothrombotic stroke.

DISCUSSION

Previous studies have shown that, by neutralizing inhibitory factors in the environment and elevating neuronal activity of affected neurons, several methods are able to promote regrowth of CST axons and resultant functional recovery after spinal cord injury and cortical stroke (García-Alías et al., 2009; Wahl et al., 2014; Carmel et al., 2010, 2014; Li et al., 2015). In this study, we present experimental evidence showing the efficacy of activating the intrinsic growth ability of adult CSNs to achieve functional recovery in both spinal cord injury and stroke models. Because both OPN and IGF1 are soluble proteins, they could serve as the basis for a highly translatable avenue of promoting neural repair.

OPN Sensitizes CSNs’ Responses to IGF1

Despite the fact that IGF1 could promote axon growth from cultured CSNs isolated from neonatal mice (Ozdinler and Macklis, 2006), it failed to promote CST regrowth in adult mice in vivo, consistent with previous findings (Hollins et al., 2009; Li et al., 2010). Thus, our results reveal a potentially important difference between young and adult neurons in the CNS, in terms of their responsiveness to growth factors. In this regard, previous studies indicated that despite well-established roles of neurotrophins and other growth factors in promoting neuronal survival in young and cultured neurons, these factors have limited efficacy in protecting neurons in disease models such as ALS (Thoenen and Sendtner, 2002). Our results suggest an exciting possibility that OPN could at least partially improve neuronal responsiveness to IGF1. Osteopontin is able to sensitize CSNs’ signaling responses to IGF1, indicated by both increased phosphorylation of IGF1 receptor and S6 kinase (Figures 2, S4B–4D), pointing to a likely possibility that it acts on the plasma membrane of CSNs. Previous studies revealed that OPN can interact with different types of integrins and other cell adhesion molecules such as CD44 (Kazanecki et al., 2007; Wang and Denhardt, 2008; Kahles et al., 2014). In non-neuronal cells, IGFR and integrins have been shown to be associated with lipid rafts (Salani et al., 2009). Thus, a working hypothesis is that by interacting integrins or other cell surface proteins, OPN could mobilize or cluster the IGF1 receptors so that their responsiveness to the ligand would be enhanced. Future studies will examine this and other possibilities.

CSN Dependent Behavioral Tasks

Despite ample evidence of CSNs and their CST projections functioning in skilled forelimb locomotion, CST-dependent behavioral tasks of the hindlimbs are not well characterized. In this study, we found that adult mice with ablated CSNs innervating low thoracic and lumbar spinal cord showed selective defects in an irregular walking task, in which these mice have to constantly rely on cortically mediated sensorimotor integration to avoid missteps. These results support a proposed role of corticospinal projections in precision walking tasks (Liddle and Phillips, 1944, Georgopoulos and Grillner, 1989; Drew et al., 1993; Carmel et al., 2010; 2014).

Furthermore, we demonstrate that mice with T10 lateral hemisection have few spontaneous sprouting of CST axons across the midline and exhibit persistent behavioral deficits in this irregular walking task, suggesting a possible causal relationship between such anatomical and behavioral events. This is further supported by our results that sensitized IGF1 treatment is able to promote CST regrowth and specific functional recovery in both T10 lateral hemisection and unilateral cortical stroke models. In the case of T10 lateral hemisection, we observed that the OPN/IGF1 treatment elicited both regenerative growth from injured CST axons and compensatory sprouting from spared axons. However, regenerated axons grew only a few millimeters, far away from the lumbar segments, and are thus unlikely to contribute to the observed functional recovery. On the other hand, in both T10 lateral hemisection and unilateral cortical stroke models, midline-crossing axons sprouted from the intact side robustly innervate the denervated side in different spinal cord levels. Our finding that restored skilled locomotion function was dependent on CSNs that sprouted midline-crossing axons into the cervical spinal cord in the unilateral cortical stroke model (Figure 8) reinforces the potential reparative effects of promoting sprouting responses of spared axons in these disease models.

Combinatorial strategies of maximizing functional recovery

Considering sub-optimal numbers of regrowing axons and their un-refined termination patterns, it is not surprising that only partial functional recovery was observed with OPN/IGF1 treatment. Instead of pharmacological treatments that increase neuronal excitability, those that improve nerve conduction are able to further improve behavioral performance. As sprouted axons are unlikely to make functional connections with their original targets in numbers approximating normal circuitry, improving axon conduction may facilitate the transmission of cortical commands carried by these detour connections in the spinal cord. Importantly, compared to clinically approved 4-AP with more serious side effect (Blight et al. 1991; Donovan et al. 2000), 4-AP-MeOH showed significantly better effects with broader safety doses, and should be considered as a candidate for further clinical investigations. In addition, rehabilitation-based methods have been shown as an additional means to facilitate functional recovery in an activity-dependent manner (Cai et al., 2006; García-Alías et al., 2009; Courtine et al., 2009; Wahl et al., 2012, van den Brand et al., 2012; Rossignol et al., 2015). Thus, further studies should test whether the observed functional recovery could be improved by such accessory manipulations. In summary, our results demonstrate a potentially translatable strategy of achieving functional restoration that is applicable for the treatment of both spinal cord injury and stroke.

EXPERIMENTAL PROCEDURES

STAR*METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken monoclonal anti-GFP	Abcam	Cat#ab13970
Rabbit polyclonal anti-RFP	Abcam	Cat#ab34771
Rabbit anti-GFAP	DAKO	Cat# Z0334
Rabbit polyclonal anti-PKC gamma	Santa Cruz	Cat#sc211
Rabbit anti-5-HT	Immunostar	Cat#20080
Rabbit anti-IGFR	Santa Cruz	Cat#sc-712
Rabbit anti-pIGFRβ	Cell Signaling Technology	Cat#3024
Rabbit anti-pS6	Cell Signaling Technology	Cat#4857
Rat anti-CD68	Bio-Rad	Cat#MCA1957
Guinea pig-anti-Vglut1	Synaptic Systems	Cat#135304
Goat-anti-ChAT	Millipore	Cat#AB144P
Biological Samples
N/A	N/A	N/A
Chemicals, Peptides, and Recombinant Proteins
Quipazine	Sigma	Cat#Q1004
SKF- 82197	Tocris	Cat#1447
8-OH-DPAT	Tocris	Cat#0529
4-AP	Sigma	Cat#275875
4-AP-MeOH	Santa Cruz	Cat# sc-267247
Diptheria toxin	Sigma	Cat# D0564
Human recombinant IGF1	Peprotech	Cat# 100–11
Human recombinant Osteopontin (OPN)	Peprotech	Cat# 120–35
Critical Commercial Assays
N/A	N/A	N/A
Deposited Data
N/A	N/A	N/A
Experimental Models: Cell Lines
N/A	N/A	N/A
Experimental Models: Organisms/Strains
Mouse/C57Bl/6	Charles River	Strain code#027
Mouse/Emx1Cre	The Jackson Laboratory	Jax#5628
Recombinant DNA
AAV-PLAP	Liu et al., 2010	Cat#N/A
AAV-tdTomato	Addgene	Cat#59462
AAV-Cre	This paper	Cat#N/A
AAV-OPN	Duan et al., 2015	Cat#N/A
AAV-IGF1	This paper	Cat#N/A
AAV-CaMKIIa-ChR2-mCherry	Addgene	Cat# 26975
AAV- Syn-ChR2-EYFP	Addgene	Cat#26973
Lenti-HiRet-GFP	This paper	Cat#N/A
Lenti-HiRet-FLEX-DTR	Jin, et al., 2015	Cat#N/A
Sequence-Based Reagents
N/A	N/A	N/A
Software and Algorithms
ImageJ2	NIH	https://imagej.nih.gov/ij/index.html
Prism 7.0	GraphPad Software, Inc.	Cat#http://www.graphpad.com/
STATA 12	College station, TX, USA	www.stata.com
Other
N/A	N/A	N/A

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Zhigang He (Zhigang.He@childrens.harvard.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice Strains

All experimental procedures were performed in compliance with animal protocols approved by the Institutional Animal Care and Use Committee at Boston Children’s Hospital or National Institution of Health. C57Bl/6 wild type mouse (Charles River, Strain code#027) and Emx1-Cre (Jax#5628), mouse strains were maintained on C57Bl/6 genetic background. For behavioral measurement, experimental animals used were from different littermates. Both male and female mice were used. Animals were fed ad libitum and maintained in the same room under a 12:12-hour light/dark photoperiod at 22°. The surgeries were performed around postnatal day 60 (see figures for details). The body weight and sexes were randomized and assigned to different treatment groups, and no other specific randomization was used for the animal studies. Behavioral tests were videotaped and examined blindly.

METHOD DETAILS

Chemicals and Antibodies

For systematic administration (i.p.), Quipazine [Sigma (Q1004), 0.2 mg/kg), SKF- 82197 [Tocris (1447) 0.1 mg/kg], and 8-OH-DPAT [Tocris (0529), 0.1 mg/kg), 4AP [Sigma (275875), 1mg/kg, 3mg/kg), 4AP-MeOH [Santa Cruz (sc-267247), (1mg/kg)] were dissolved in saline. Tamoxifen (Sigma, 10540-29-1) was dissolved in oil. For diphtheria toxin mediated cell ablation, we purchased the diptheria toxin from Sigma (D0564). For immunostaining, the primary antibody used were chicken anti-GFP [Abcam (Cat: ab13970)], rabbit anti-RFP [Abcam (Cat: ab34771)], rabbit anti-PKCγ[Santa Cruz (sc211)], rabbit anti-GFAP [DAKO (Z0334)]; rabbit anti-5-HT [Immunostar (20080)], rabbit-anti-IGFR [Santa Cruz (sc-712)] rabbit-anti-pIGFRβ [Cell signaling technology (3024)], rabbit-anti-pS6 [Cell signaling technology (4857)], rat anti-CD68 [Bio-Rad (MCA1957)], and Guinea pig-anti-Vglut1 [Synaptic Systems (135304)].

Injury Models

The procedure of T10 lateral hemisection was similar to that described elsewhere (Ballermann and Fouad, 2006; Courtine et al., 2008; Takeoka et al., 2014). Briefly, a midline incision was made over the thoracic vertebrae, followed by a T10 laminectomy. The unilateral hemisection was then performed carefully using both scalpel and micro-scissors, avoiding, to the greatest extent, the damage of the spinal cord dura. The muscle layers were then sutured and the skin was secured with wound clips. All mice received post hoc histological analysis and those with spared CST axons (incomplete lateral hemisection) or with significant less CST axons on the contra-lesional side (over lateral hemisection) at the lumbar spinal cord (L3), exemplified in Figure S1A, were excluded for behavioral analysis.

The procedure of unilateral photothrombotic stroke was similar to that described elsewhere (Watson et al., 1985; Wahl et al., 2014). Briefly, mice were fixed in a stereotactic frame, with the skull exposed. To unilaterally cover the sensoromotor cortex, the cold light source (Zeiss, CL 1500HAL, 3000K) was positioned over an opaque template with an opening (a circle with a diameter of 2.5mm) centered at (−0.5, 2.0 mm, anterior and lateral to the bregma) on the cortex contralateral to the preferred paw in the food pellet retrieval task. Rose Bengal (10 mg/kg body weight, 5 mg/ml Rose Bengal in saline) was injected (i.p.) 10 min before the brain was illuminated through the intact skull for 15 min. Lesion volumes were calculated when mice brains were fixed at the end point of the experiments.

Virus and Protein Injection

AAV2/1-IGF1, AAV2/1-OPN, AAV2/1-PLAP, AAV2/1-ChR2-YFP, AAV2/1-ChR2-mCherry, AAV2/1-Cre, AAV-2/9- GFP and mCherry (all AAV titers were adjusted to 0.5–5×10¹³ copies/ml for injection, produced by Boston Children’s Hospital, viral core) or recombinant human IGF1 (Peprotech, 1μg/1μl) and/or osteopontin (OPN) (Peprotech, 1 μg/μl) were injected to the mouse sensorimotor cortex as described previously (Liu et al., 2010, Zukor et al., 2013). Vectors of HiRet-GFP, HiRet-mCherry, HiRet-FLEX-DTR (all lenti-virus titers were adjusted to 1.6–2×10¹² copies/ml for injection) were constructed based on the HiRet-lenti backbone (Kinoshita et al., 2012).

Immunohistochemistry and Imaging

The paraformaldehyde fixed tissues were cryo-protected with 30% sucrose and processed using cryostat (section thickness 40 μm for spinal cord and 60 μm for brain). Sections were treated with a blocking solution containing 10% normal goat serum with 0.5 % Triton-100 for 2 hours at room temperature before staining. The primary antibodies (4°, overnight) used are rabbit anti-GFAP [DAKO (Z0334), 1:1000]; rabbit anti-5-HT [Immunostar (20080), 1: 5,000]; chicken anti-GFP [Abcam (ab13970), 1:400]; rabbit anti-RFP [Abcam (ab34771), 1:400]; rabbit anti-PKCγ [Santa Cruz (sc211),1:100]; rabbit-anti-IGFR [Santa Cruz (sc-712), 1:200] rabbit-anti-pIGFRβ [Cell signaling technology (3024), 1:100], rabbit-anti-pS6 [Cell signaling technology (4857), 1:200], rat anti-CD68 [Bio-Rad (MCA1957), 1:400], and Guinea pig-anti-Vglut1 [Synaptic Systems (135304), 1:1000]. Secondary antibodies (room temperature, 2h) include Alexa Fluor 488-conjugated goat anti chicken and rabbit, Alexa Fluor 594-conjugated goat anti rabbit (all from Invitrogen). Spinal cord transverse and horizontal sections and brain transverse sections were imaged with a confocal laser-scanning microscope (Zeiss 700 or Zeiss 710). To quantify and compare fluorescence intensity of IGFR, pIGFRβ and pS6 in GFP⁺ CSNs (Figures 2C–2E), CST axons in the dorsal funiculus (Figure S2G) and CD68 staining (Figure S2F) and corticofugal projections at multiple subcortical areas (Figure 7B), and axon density of the intact side in the spinal cord (Figure 8E and 8F), all images used for analysis under multiple conditions were taken using the same optical parameters and avoided for saturation. Densitometry measurement was taken by using FIJI software, after being sub-thresholded to the background and normalized by area.

Specific Ablation or inhibition of Hindlimb Corticospinal Neurons

To specifically target hindlimb CSNs, 2 μl HiRet viruses (HiRet-GFP/mCherry for labeling HiRet-FLEX-DTR for ablation) were injected to the lower thoracic to lumbar spinal cord (T12-L4) guided by ultrasound (detailed method see Arlotta et al., 2005) and carried out at postnatal day 12–14 (P12–P14) in Emx1-Cre mice. Diphtheria toxin (DT,100 μg/kg) or tamoxifen (75 mg/kg) was administered (i.p.) in adult animals. The high efficient ablation was verified by the absence of PKCγ staining in the dorsal funiculus of the lumbar, but not cervical, spinal cord (Figures 1 and S2G) and also with retrograde labeling at lumbar spinal cord (Figure S2C–2E).

Axon Counting and Quantification

To quantify the number of sprouting axons, a horizontal line was firstly drawn through the central canal and across the lateral rim of the gray matter. Three vertical lines (Mid, Z1, and Z2) were drawn to divide the horizontal line into three equal parts, starting from the central canal to the lateral rim. While Mid denotes midline crossing fibers, Z1 and Z2 are for sprouting fibers at different distance from the midline. Only fibers crossing the three lines were counted on each section. The results were presented after normalization with the number of counted CST fibers at the medulla level.

For quantifying total labeled CST axon, AAV-ChR2-YFP (Figure 4D) or AAV-ChR2-mCherry (Figures 4C, 7E and 8E) labeled CST fibers were counted at the level of medulla oblongata 1 mm proximal to the pyramidal decussation. Axons were estimated by counting 4 rectangular areas (about 10000 μm²/area) per section on two adjacent sections.

To quantify the regenerating axons (Figure 4D), the number of intersections of chR2-YFP-labeled fibers with a dorsal-ventral line positioned at a defined distance caudal from the lesion center was counted under a 25X objective. Fibers were counted on 3 sections with the main dorsal CST and 1–3 lateral sections with collaterals in the gray matter. The number of counted fibers was normalized by the number of labeled CST axons in the medulla and divided by the number of evaluated sections. This resulted in the number of CST fibers per labeled CST axons per section at different distances (fiber number index).

Behavioral experiments

Ground walking, Swimming and Treadmill Walking

For ground walking and swimming, mice were placed in the MotoRater (TSE Systems, Zorner et al., 2010) and all kinematic analysis was performed based on data collected by the MotoRater.

For treadmill walking, mice were placed on the DigiGate at various speed. Speed tolerance was defined as the maximal speed a mouse can walk on the treadmill without falling. All trials were video recorded (Hotshot e64, 100 fps) for the measurement of the paw dragging distance on the treadmill.

Irregular Ladder Walking

In this assay, mice in different groups were tested to walk on a horizontal ladder with irregular spacing between rungs, following the procedure described previously (Metz and Whishaw, 2002; Carmel et al., 2010, 2014; Jin et al., 2015). Briefly, the ladder was elevated 30 cm above the ground. Animals were trained to cross the ladder until their performance achieved the plateau (with an average error rate about 20%). To prevent animals from learning the pattern, the irregular pattern was changed from trial to trial. All trials were video recorded (Hotshot e64, 100 fps) and paw placement was analyzed twice by blinded observers. We define steps with precise placement of the center of the palm on the rung (for both forelimbs and hindlimbs) and digits closed (for forelimbs) as correct steps (hit). All other steps were recorded as errors, which included two types: 1) Miss: when crossing the ladder, the forelimb/hindlimb either completely miss the rung or contact the rung with the wrist/heel instead of the paw; 2) Slip: when crossing the ladder, the mice use a few digits instead of the paw to place on the rungs, causing the subsequent slip on the rungs. The results were expressed as both percentage of total errors and percentage of different placement categories (hit, miss and slip).

Single pellet Retrieval

The single-pellet reaching task was carried out following previously established procedures with slight modification (Farr and Whishaw, 2002). The training chamber was built from clear Plexiglas (1 mm thickness, dimensions 203cm × 153cm × 8.53cm), with a vertical slit (0.53cm wide; 133cm high) located on the front wall of the box. An exterior shelf with 1.5 cm height was affixed to the wall in front of the slits to hold a sugar pellet (dustless precision pellet, 20 mg, bioserv). After one day of habituation to the chamber with sugar pellet inside the chamber, mice were food-restricted for one night before training and were maintained above 90% of free feeding weight throughout the training session. Mice were digitally videotaped at 60 frames/sec while reaching for a maximum of 40 pellets within 20 min. The success rate was calculated as: number of successful retrievals/total attempts per trail *100. The animals without intention to retrieve the sugar pellet or consistently using the tongue instead of the forelimb to retrieve the sugar pellet were excluded from receiving the unilateral photothrombotic stroke.

Pharmacological Treatment

Ten to fifteen minutes (van den Brand et al., 2012) prior to behavioral tests (irregular ladder walking, grounding walking or treadmill walking, all of which were performed individually), mice received systematic administration (i.p.) of neural modulators [quipazine (0.2 mg/kg), SKF- 82197 (0.1 mg/kg), or 8-OH-DPAT (0.1 mg/kg)]. We did pilot experiment and determined that 4AP (1mg/kg, 3mg/kg) and 4AP-MeOH (1mg/kg) achieved their maximal effects within 1–3 hours post systematic administration (i.p.). All behavioral tests were then accomplished between 1–3 hours post administration.

Selective Ablation of CSNs with Sprouted Axons to the Denervated Side of the Spinal Cord

Mice received unilateral photothrombotic stroke at P60, OPN/IGF1 treatment at P63 respectively. Fourteen weeks after injury, a laminectomy was performed at cervical spinal cord. The viruses (1×10¹² copies/ml) generated by a HiRet- carrying the FLEX-DTR were stereotaxically injected into the denervated side of the cervical (C5–C7) spinal cord of the OPN/IGF1 treated mice with procedures established in Jin et al., 2015. AAV2/1-Cre (ablation) or AAV2/1-PLAP (control) (1×10¹³ copies/ml) was then injected into the unlesioned sensoromotor cortex at 3 days post HiRet virus injection. After 2 weeks, animals were tested for the irregularly spaced horizontal ladder walking and/or single pellet food retrieval task to reassess their performance of the skilled limb movement. Diphtheria toxin was then administrated (100 μg/kg, i.p.). Animals were tested for the horizontal ladder walking and/or single pellet food retrieval task again at 2 and 4 weeks after diphtheria toxin administration.

QUANTIFICATION AND STATISTICAL ANALYSIS

The normality and variance similarity were measured by STATA (version 12, College station, TX, USA) before we applied any parametric tests. Two-tailed Student’s t-test was used for the single comparison between two groups. The rest of the data were analyzed using one-way or two-way ANOVA depending on the appropriate design. Post hoc comparisons were carried out only when a main effect showed statistical significance. P-value of multiple comparisons was adjusted by using Bonferroni’s correction. Error bars in all figures represent mean ± S.E.M. The mice with different litters, body weights and sexes were randomized and assigned to different treatment groups, and no other specific randomization was used for the animal studies.

Highlights.

• Osteopontin (OPN) sensitizes the responses of adult corticospinal neurons to IGF1

• OPN/IGF1 promotes CST regrowth and relevant functional recovery after spinal cord injury

• CST sprouting induced by OPN/IGF1 mediates functional recovery after stroke

• 4-AP-MeOH further improves behavioral performance after OPN/IGF1 treatment