Transneuronal delivery of designer-cytokine enables functional recovery after complete spinal cord injury

Spinal cord injury (SCI) often causes severe and permanent disabilities. The current study uses a transneuronal approach to stimulate spinal cord regeneration by AAV-hyper-IL-6 (hIL-6) application after injury. While preinjury PTEN knockout in cortical motoneurons fails to improve functional recovery after complete spinal cord crush, a single, postinjury injection of hIL-6 into the sensorimotor cortex markedly promotes axon regeneration in the corticospinal and, remarkably, raphespinal tracts enabling significant locomotion recovery of both hindlimbs. Moreover, transduced cortical motoneurons directly innervate serotonergic neurons in both sides of the raphe nuclei equally, enabling the synaptic release of hIL-6 and the transneuronal stimulation of raphe neurons in the brain stem. Functional recovery depends on the regeneration of serotonergic neurons as their degeneration induced by a toxin abolishes the hIL-6-mediated recovery. Thus, the transneuronal application of highly potent cytokines enables functional regeneration by stimulating neurons in the deep brain stem that are otherwise challenging to access, yet highly relevant for functional recovery after SCI.


Introduction
Neurons of the adult mammalian central nervous system (CNS) do not normally regenerate injured axons. This regenerative failure often causes severe and permanent disabilities, such as para-or tetraplegia after spinal cord injury. To date, no cures are available in the clinic, underscoring the need for novel therapeutic strategies enabling functional recovery in respective patients.
Besides an inhibitory environment for axonal growth cones caused by myelin or the forming glial scar, the lack of CNS regeneration is mainly attributed to a developmental decline in the neuronintrinsic growth capacity of axons per se [1][2][3] . Among all descending pathways, the corticospinal tract (CST), which controls voluntary fine movements, is the most resistant to regeneration.
Despite numerous efforts aiming to facilitate axon regrowth of the CST over the last decades, such as the delivery of neurotrophic factors [4][5][6] or neutralizing inhibitory cues [7][8][9] , success has remained very limited. However, the conditional genetic knockout of the phosphatase and tensin homolog (PTEN -/-) in cortical motor neurons, which leads to an activation of the phosphatidylinositol-3-kinase/protein kinase B (PI3K/AKT)/mTOR signaling pathway, enables some regeneration of CST axons beyond the site of injury 10 . Although this approach facilitated the most robust anatomical regeneration of the CST after a complete spinal cord crush injury so far 10 , it fails to improve functional motor recovery 11 .
In the optic nerve, the activation of the Janus kinase/signal transducer and activator of transcription 3 (JAK/STAT3) pathway stimulates the regeneration of CNS axons 12,13 . JAK/STAT3 activation is achieved via the delivery of IL-6-type cytokines such as CNTF, LIF, IL-6, and/or the genetic depletion of the intrinsic STAT3 feedback inhibitor: suppressor of cytokine signaling 3 (SOCS3) 12,[14][15][16][17] . However, the low and restricted expression of the cytokine-specific α -receptor subunits in CNS neurons that are required for signaling induction generally limits these pro-regenerative effects of native cytokines. For this reason, a gene therapeutic approach was recently developed using the designer cytokine hyper-interleukine-6 (hIL-6), which consists of the bioactive part of the IL-6 protein covalently linked to the soluble IL-6 receptor α subunit 18 .
In contrast to native cytokines, hIL-6 can directly bind to the signal-transducing receptor subunit glycoprotein 130 (GP130) abundantly expressed by almost all neurons and thereby circumvent the limitation of natural cytokines 19,20 . Hyper-IL-6 is reportedly as potent as CNTF but activates cytokine-dependent signaling pathways significantly stronger in different types of neurons because of its higher efficacy 18 . In the visual system, virus-assisted gene therapy with hIL-6, even when applied only once post-injury, induces stronger optic nerve regeneration than a preinjury induced PTEN knockout (PTEN -/-) 18 . Hence, this treatment is the most effective approach to stimulate optic nerve regeneration when applied after injury.
The current study analyzed the effect of cortically applied AAV-hIL-6 alone or in combination with PTEN -/on functional recovery after complete spinal cord crush (SCC). A single unilateral injection of AAV-hIL-6 applied after complete SCC into the sensorimotor cortex promoted regeneration of CST-axons stronger than PTEN -/-, and, remarkably, also of serotonergic fibers of the raphespinal tract, which enabled locomotion recovery of both hindlimbs. Moreover, this study shows that cortical motoneurons form synapses with raphe neurons deep in the brain stem allowing the axonal transport and synaptic release of hIL-6 to stimulate regeneration of serotonergic axons. Thus, transneuronal stimulation of neurons located deep in the brain stem using highly potent molecules might be a promising strategy to achieve functional repair in the injured or diseased human CNS.
Neither PTEN -/-, hIL-6 expression nor their combination affected the total number of BDApositive CST axons in the medullary pyramid ( Fig. S2 A, B, J) or CST-axonal sprouting in the thoracic spinal cord rostral to the lesion site compared to controls (Fig. S2 C-J). However, axonal dieback of CST-fibers above the injury site, typically seen in control mice, was significantly reduced by PTEN -/and slightly stronger by the AAV2-hIL-6 treatment ( Fig. 2 B, C). The combination of PTEN -/-+ AAV2-hIL-6 showed no additional effect (Fig. 2 B, C).
We then analyzed fiber regeneration beyond the crush site. Contrary to controls (

Hyper-IL-6 promotes functional recovery
Before tissue isolation and analysis, hindlimb movement had been analyzed in all four experimental groups using open-field locomotion tests according to the Basso Mouse Scale (BMS) over the postinjury period of 8 weeks 21 . Consistent with previous reports [21][22][23] , the BMS score dropped down to 0 in all animals 1 day after SCC, also indicating the completeness of the SCC (Fig. 3 A, B; Fig. S4 A-E). Over time, control animals developed only active ankle movements, including spasms (supplementary video 1) as described previously 21,22 , resulting in an average final score of 2 ( Fig. 3 A, B; Fig. S4 A; supplementary video 1). Despite the effect on CST-regeneration ( Fig. 2 D, F, I), PTEN -/did not significantly improve the BMS score compared to AAV2-GFP treated controls (Fig. 3 A, B; Fig. S4 C; supplementary video 2). Strikingly, AAV2-hIL-6 treatment increased the score to 4-5 by restoring plantar stepping with full weight support through the hindlimb followed by lift-off, forward limb advancement, and reestablishment of weight support at initial contact in most of the animals (Fig. 3 A

Cortical AAV2-hIL-6 delivery promotes RpST regeneration
The lack of any functional recovery in PTEN -/mice suggested that improved CST regeneration was not the leading cause for the AAV2-hIL-6 mediated functional recovery. As descending

AAV2-hIL-6 mediated recovery depends on the regeneration of serotonergic neurons
To investigate the relevance of RpST regeneration for functional recovery, adult mice of the same genetic background were subjected to SCC and received bilateral intracortical injections of either AAV2-hIL-6 or AAV2-GFP directly afterward (Fig. 5 A). At week 6, when functional recovery in AAV2-hIL-6 treated mice had reached maximal levels, the neurotoxin 5, 7-Dihydroxytryptamine (DHT) was intracerebroventricularly injected to selectively kill serotonergic neurons ( Fig. 5 A-D) 24,27,28 . One day after DHT application, the BMS in AAV2-hIL-6 treated mice dropped down to similar levels as in control animals, while DHT did not affect the BMS score of controls ( Fig. 5 E, F; Supplementary video 5), suggesting a dependence of the AAV2-hIL-6 mediated locomotory recovery on RpST regeneration although AAV2-hIL-6 had not been applied into the brain stem where the serotonergic neurons originated.
Since genetic backgrounds of mice could potentially affect regeneration, we also tested the postinjury applied AAV2-hIL-6 treatment in non-transgenic BL6 mice, whose low potential for functional recovery has been previously documented 21 . Eight weeks after complete SCC and unilateral intracortical AAV2-hIL-6 injection, BL-6 mice also showed reduced axonal dieback, but less CST-regeneration caudal to the lesion site than accordingly treated PTEN-floxed OLA mice ( Fig. S5 A, B, Fig. S3 B). Nevertheless, the improvement in RpST axon regeneration and functional recovery were very similar in both mouse strains (

AAV2-hIL-6 confers transneuronal stimulation of serotonergic fiber regeneration
To explain the mechanism underlying the bilateral, RpST-dependent functional recovery by a single unilateral AAV2-hIL-6 application into the sensorimotor cortex, we tested the hypothesis that transduced motoneurons project axons to the raphe nuclei in the brain stem and that synaptically released hIL-6-protein stimulates regeneration of serotonergic neurons transsynaptically. To this end, we initially used axon isolation devices to separate somata and axons of cultured sensory DRG neurons 29  Thus, hIL-6 is transported in axons, released at the terminals, and remains active to stimulate axon regeneration.
To investigate whether transduced hIL-6 can also stimulate neurons trans-synaptically in vivo, we injected either AAV2-hIL-6 or AAV2-GFP into the left eye to transduce RGCs. Three weeks later, we tested for STAT3 phosphorylation in their brain targets, the lateral geniculate nucleus The hIL-6 expression in RGCs did not affect the integrity of axons as their number determined after neurofilament staining in the optic nerve and optic tract cross-sections was not reduced compared to untreated controls, excluding any uncontrolled or widespread release of the protein.
We then investigated whether raphe nuclei in the brain stem receive synaptic input from transduced cortical motoneurons and analyzed 5-HT stained brain stem tissue from mice 8 weeks after SCC and AAV2-hIL-6 into the sensorimotor cortex as described above (Fig. 2 A).
To test whether cortical motoneurons were synaptically connected with raphe neurons, we used AAV1, which can transsynaptically transduce supraspinal target neurons in rats 32 . We injected AAV1-Cre into the left sensorimotor cortex of Rosa-tdTomato (RFP) reporter mice and isolated 1 the brain stem tissue after 2 weeks. RFP-expressing (transduced) serotonergic neurons were detected in both sides of the nucleus raphe pallidus (NRPa) and the nucleus raphe magnus (NRM) (Fig. 6 E-L), indicating a synaptic connection to cortical neurons. Hence, unilateral cortical AAV2-hIL-6 application transneuronally activates regenerative signaling pathways in ipsiand contralateral raphe neurons by the release of the protein.

Discussion
The current study shows significant locomotor recovery in an adult mammal by a single, unilateral application of AAV2-hIL-6 into the sensorimotor cortex after a complete SCC. While pre-injury-induced PTEN -/in cortical neurons failed to facilitate functional recovery, postinjury applied AAV2-hIL-6 promoted longer CST axon growth than PTEN -/and, additionally, regeneration of serotonergic axons in the RpST, enabling significant locomotor recovery of both hindlimbs. Moreover, we provide first direct evidence that cortical motoneurons innervate raphe neurons, thus allowing the axonal transport and synaptic release of highly potent hIL-6 to stimulate the regeneration of serotonergic fibers in the brain stem. Thus, a transneuronal application of highly active molecules is a new and powerful approach to activate regenerative signaling, particularly in neurons located in brain regions that are challenging to access but relevant for functional recovery after spinal cord injury.
In contrast to incomplete, less severe hemisection-or contusion-injury models, permitting spontaneous functional recovery with BMS scores Hence, most previous treatment strategies in this injury model did not reach significant effects beyond a reduction of dieback of CST axons [34][35][36] . Only PTEN -/in cortical motoneurons enabled regeneration of CST-axons and synapse formation with interneurons but still failed to restore locomotion 10,11,33,37,38 . We confirmed these findings and used the PTEN -/model as a reference to validate the effects of AAV2-hIL-6. Although cortical PTEN -/showed a robust effect at shorter distances suggesting a stronger initiation of axon regeneration in the CST, hIL-6 promoted axons to regenerate much further. Moreover, only AAV2-hIL-6 treatment significantly improved serotonergic axon growth in the RpST and, remarkably, enabled locomotion recovery of both hindlimbs. These effects were achieved in mice with different genetic backgrounds by a single, unilateral application of AAV2-hIL-6 after the SCC, making this gene therapeutic approach also a potential strategy to facilitate spinal cord repair in the clinic.
PTEN -/and hIL-6 differently activated regenerative pathways. While PTEN -/expectedly activated PI3K/AKT/mTOR, AAV2-hIL-6 only induced JAK/STAT3-signaling in transduced and adjacent cortical neurons ( Fig. 1) and via the transneuronal route also in raphe neurons. It is therefore conceivable that the combinatorial activation of both signaling pathways by PTEN -/and hIL-6 in cortical motoneurons would also synergistically result in stronger CST-regeneration than each treatment by itself as previously shown in the optic nerve [16][17][18] . These synergistic effects were, however, limited to the regeneration of the CST due to the effect of PTEN -/being restricted to the virally transduced motoneurons in the cortex. Moreover, the finding that hIL-6 alone did not measurably affect PI3K/AKT/mTOR activity but showed even stronger effects than PTEN -/suggests that extensive activation of AKT/mTOR, which is associated with a cancerogenic risk, is not essential to achieve a reduction of axonal dieback or improvement of CST regeneration. AAV2-hIL-6 did not activate the MAPK/ERK pathway either leading to the conclusion that its beneficial effect was mediated by STAT3 phosphorylation as previously shown for RGC-regeneration 12,16,17,39 . Future experiments need to address the question whether a specific knockout or inhibition of STAT3 in cortical neurons reduces the hIL-6 effects on CST-regeneration in the SCC model or whether the effects of hIL-6 on JAK/STAT3 activation and axon regeneration can be further increased in combination with a specific knockout of SOCS3, which normally limits the activity of JAK/STAT3-signaling 16,17 .
Although AAV2-hIL-6 enabled longer CST-regeneration than PTEN -/-, our data suggest that the beneficial effect on functional recovery mostly depended on the improved regeneration of serotonergic fibers of the raphe nuclei. This is because i) the regeneration of CST axons by PTEN -/alone did not enable hindlimb recovery, ii) BL6 mice showed similar effects on RpST-regeneration and recovery as that seen in Ola-PTEN-floxed animals but with less CST regeneration after AAV2-hIL-6 application, and iii) selective elimination of serotonergic fibers by DHT 24,27,28 abolished most of the recovered locomotion after AAV2-hIL-6 treatment without affecting AAV2-GFP treated control animals, thereby also verifying the specificity of the neurotoxin. Consistent with this, the critical role of RpST regeneration for locomotor recovery has been reported after less severe spinal cord injuries, which allow some endogenous sprouting of serotonergic axons 24,[40][41][42][43] . Although these data demonstrate that AAV2-hIL-6mediated functional recovery depends on the regeneration of serotonergic fibers, we cannot exclude the possibility that hIL-6 might have also stimulated other nuclei in the brain, such as the red nuclei, which also receive collateral input from cortical motoneurons 31 . Consistently, intracortical AAV2-hIL-6 application also induced STAT3 activation in the red nucleus, verifying the transneuronal hIL-6 delivery. Whether the regeneration of rubrospinal tracts was also affected or contributed to the beneficial effects of AAV-hIL-6 is currently unknown.
Moreover, improved regeneration of the CST on top of the RpST could have contributed to the beneficial effect of hIL-6, despite CST regeneration induced by PTEN -/alone being insufficient.
This would explain the finding that the stronger CST regeneration in the AAV2-hIL-6/PTEN -/group compared to hIL-6 animals correlated with an improvement seen in the BMS subscore, although the RpST regeneration remained similar. So improved CST-regeneration might only affect functional recovery on top of RpST regeneration, which first enables the basic walking behavior. Accordingly, CST axon regeneration reportedly improves voluntary movements and skilled locomotion in less severe pyramidotomy-and contusion-injury models, which leave other spinal tracts still intact 26,[44][45][46] . Future experiments need to investigate to what extent regeneration of serotonergic fibers alone can mimic the full AAV2-hIL-6 effect on functional recovery and whether it also reveals beneficial effects in less severe spinal cord injury models (e.g., pyramidotomy, hemisection, and contusion).
Although the collateral projection of cortical motor neurons into various brain areas, e.g., in the striatum and the thalamus [47][48][49] , as well as the red nucleus and reticular formation, are well documented [50][51][52] , the innervation of raphe nuclei in the brain stem has not yet been directly shown. Only data from a recent study constructing a connectome map of the whole brain point to a cortical projection of layer V motoneurons into the nucleus raphe magnus without clearly addressing this possibility 31 . The current study used AAV1, which can trans-synaptically transduce neurons 32 and demonstrates in transgenic mice that cortical neurons project almost equally into the ipsi-and contralateral sides of the raphe nuclei. Finally, the activation of STAT3 (pSTAT3) in serotonergic raphe neurons by unilateral AAV2-hIL-6 application in the motor cortex in our study provides further evidence for this innervation target.
The current study also demonstrates that virally expressed hIL-6 is not only released from the soma to stimulate adjacent motoneurons in a paracrine fashion but that it is also transported over long distances in axons of either RGCs in the visual system or cortical neurons. The release of active hIL-6 at axonal terminals/synapses was verified in cell culture experiments and by pSTAT3 staining in visual target areas as well as directly in the raphe nuclei themselves. In this context, it is worth mentioning that hIL-6 is highly potent, so that even the smallest quantities can activate all types of neurons expressing gp130, the receptor to which it directly binds 18 .
The high potency of hIL-6 and the almost equal activation of neurons in the ipsi-and contralateral sides of the raphe nuclei also explains why the unilateral application improved the recovery of both hindlimbs to a similar degree, and why the bilateral application had no additional beneficial effect on RpST regeneration or functional recovery.
In conclusion, the finding that transneuronal application of hIL-6 enables functional recovery opens many new possibilities to further improve the functional outcome by combining it with other strategies, such as neutralizing extracellular inhibitors at the lesion site 8,42,53 or bridging the lesion site with permissive grafts 54,55 . These combinatorial strategies, also in less severe
C57BL/6J mice were obtained from Janvier Labs. Rosa26 loxP-stop-loxP-tdTomato (Rosa-tdTomato) mice were obtained from Jackson Laboratories (Stock No: 007914). All animals were housed under the same conditions for at least ten days before the start of experiments and generally maintained on a 12 h light/dark cycle with ad libitum access to food and water. All experimental procedures were approved by the local animal care committee and conducted in compliance with federal and state guidelines for animal experiments.

Intracortical injection of pups
Postnatal (P1) PTEN floxed (C57BL/6;129/J-TgH(Pten-flox)) pups were fixed under a stereotactic frame and continuously supplied with 2% isoflurane for anesthesia via a mouthpiece. A midline incision into the skin was made to expose the skull using microscissors.
Since the skull of P1 mice is still soft, the cortex could be accessed using a 30-gauge needle to create two small holes in the left skull hemisphere with the following coordinates: -0.2 mm and 0.3 mm anteroposterior, 1.0 mm lateral to bregma. For AAV2-GFP or AAV2-Cre application, 770 nl of the virus suspensions were injected at a depth of 0.5 mm into the two sites using a pulled glass pipette connected to a nanoliter injector (Nanoject II, Drummond). To inject 770 nl, we applied 11 pulses of 70 nl at a rate of 23 nl/s and waited for 10 s after each pulse to allow distribution of the virus solution. After injection, the pipette was left in place for 1 minute before being carefully withdrawn, and the incision carefully closed with a 4-0 black silk suture.

Intracortical injection of adult mice
For intracortical injection, adult PTEN floxed (C57BL/6;129/J-TgH(Pten-flox)), or adult wt (C57BL/6) mice were anesthetized by intraperitoneal injections of ketamine (120 mg/kg) and xylazine (16 mg/kg), and then placed in a stereotaxic frame. A midline incision was made over the skull to open the skin and to reveal the bregma. A microdrill with a 0.5 mm bit was used to open a 2 x 2 mm window on each side of the skull to expose the sensorimotor cortex. The respective AAV2 was injected into the cortex layer V through a glass pipette attached to a nanoliter injector (Nanoject II, Drummond). To this end, four injections of 490 nl each were given either unilaterally (left hemisphere) or in both hemispheres, at the following coordinates: 1.5 mm lateral, 0.6 mm deep, and 0.5 mm anterior; 0.0 mm, 0.5 mm, and 1.0 mm caudal to bregma. For injecting 490 nl into each injection site, we applied 7 pulses of 70 nl at a rate of 23 nl/s and waited for 10 s after each pulse to allow distribution of the virus solution. The needle was left in place for 1 minute before moving to the next site, and the brain was kept moist during the procedure by moving the skin over the exposed area after each injection. After surgery, the skin was closed with sutures. The virus transduced mainly neurons in layer 5 of the primary motor cortex (M1). We observed only very rare transduction of astrocytes or neurons of other M1 layers ( Fig. S1 A-E). All GFP-expressing neurons also expressed hIL-6 ( Fig. S1 F). Therefore, hIL-6 transduction was identified by GFP expression during the whole study.

Injection into the red nucleus
Adult wt mice were anesthetized by intraperitoneal injections of ketamine (120 mg/kg) and xylazine (16 mg/kg) and then placed in a stereotaxic frame. A midline incision was made over the skull to open the skin and to reveal the bregma. A microdrill with a 0.5 mm bit was used to open a 1 x 1 mm window in the skull to expose the cortex. Biotinylated dextran amine (BDA, 10,000 MW, 10% solution in water, Invitrogen, D1956) was injected into the red nucleus through a glass pipette attached to a nanoliter injector (Nanoject II, Drummond). To this end 500 nl was given as previously described 26 at the following coordinates: 0.6 mm lateral, 3.5 mm deep, and 2.5 mm caudal to bregma. For injecting 500 nl, we applied 5 pulses of 100 nl at a rate of 5 nl/s and waited for 10 s after each pulse to allow distribution of the virus solution. The needle was left in place for 1 minute before removing, and the brain was kept moist during the procedure. After surgery, the skin was closed with sutures.
The fat and muscle tissue were cleared from thoracic vertebrae 7 and 8 (T7, T8). While holding onto T7 with forceps, we performed a laminectomy at T8 in order to expose the spinal cord.
Afterward, the complete spinal cord was crushed for 2 s with forceps that had been filed to a width of 0.1 mm for the last 5 mm of the tips to generate a homogeneously thin lesion site. To ensure that the full width of the spinal cord was included we took care to gently scrape the forceps' tips across the bone on the ventral side of the vertebral canal so as not to spare any axons ventrally or laterally. After surgery, the muscle layers were sutured with 6.0 resorbable sutures and the skin secured with wound clips. The completeness of the injury in the SCC model was verified by an astrocyte free gap and the absence of any spared CST or raphe spinal tract (RpST) fibers caudal to the lesion site shortly after injury (Fig. S7 E-J).

CST tracing
To trace axons from cortical motoneurons of PTEN floxed (C57BL/6;129/J-TgH(Pten-flox)), or wt (C57BL/6) mice, we injected the axon tracer biotinylated dextran amine (BDA, 10,000 MW, 10% solution in water, Invitrogen, D1956) into the sensorimotor cortex 2 weeks before the mice were sacrificed. Therefore, the skin was opened, and 490 nl BDA were applied to each injection site using the same procedure and coordinates as described for AAV2 injection. After surgery, the skin was closed with sutures. In groups with the PTEN knockout, we verified that 70-80% of BDA-traced neurons were also Cre-positive (Fig. S7 A-D).

DHT injection
Mice received bilateral intracerebroventricular injections of 30 µg of the serotonin neurotoxin 5,7dihydroxytryptamine (DHT, Biomol) dissolved in 0.5 μ l of 0.2% ascorbic acid in saline to deplete serotonergic inputs to the lumbar spinal cord. To this end, mice were anesthetized and fixed under a stereotactic frame as described above for intracortical injections. The tip of a glass micropipette attached to a nanoliter injector (nanoject II, Drummond) was positioned at the following coordinates: 0.6 mm posterior, 1.6 mm lateral to bregma, and 2 mm deep from the cortical surface. DHT was injected at the same rate as described for the AAV2 injections, and the pipette was left in place for 1 minute before the withdrawal. Thirty minutes before DHT injection, the monoamine uptake inhibitor, desipramine (Sigma), was administered at 25 mg/kg intraperitoneally to prevent the uptake of DHT into noradrenergic neurons.

DRG neuron two-compartment cultures
Dorsal root ganglion-(DRG) neurons were harvested from adult mice, as previously described 57 .

Western blot
For cortical lysate preparation, mice were killed, and a cuboid piece of cortical tissue of 1.5 x 2 mm with 1 mm depth from the cortical surface was dissected and isolated from the sensorimotor cortex around the coordinates used for viral injection. For lysates of brain stem raphe nuclei, a 2.5 mm long and 1.5 mm wide piece of tissue was isolated around the midline of the medulla with a depth of 1 mm starting above pyramidal tracts, which were thereby also removed. Tissues Signaling Technologies; RRID: AB_10892860). Antigen-antibody complexes were visualized by using enhanced chemiluminescence substrate (Bio-Rad) on a FluorChem E detection system (ProteinSimple). Western blots were repeated at least three times to verify results. Band intensities were quantified relative to respective loading controls by using ImageJ software.

Tissue clearing
Brain and spinal cord tissue were isolated from mice after perfusion with PBS and 4% PFA as described above. Tissue was postfixed overnight at 4 °C before further processing. Wholemount immunostaining was performed as described previously 58 2)) and imaged with a confocal laser scanning microscope (SP8, Leica).

Quantification of the total number of BDA traced CST axons
To The area of the entire pyramid and the total area inside a counted square were measured and used to extrapolate the total number of BDA-labeled axons per pyramid.

CST sprouting proximal to the lesion site
The sprouting index was determined as described previously 10,45 .

Evaluation of CST-axon retraction
The retraction of axons rostral to the injury site was quantified as an axon density index by measuring the BDA staining intensity as described previously 10,11,33

Basso Mouse Scale (BMS)
The locomotory behavior of mice was tested and scored according to guidelines of the Basso Mouse Scale 21 . Therefore, each mouse was placed separately in a round open field of 1 m in diameter and observed by two testers for 4 minutes. Scoring was based on different parameters such as ankle movements, paw placement, stepping pattern, coordination, trunk instability, and tail position, with a minimum score of 0 (no movement) to a maximum score of 9 (normal locomotion). For animals that have attained frequent plantar stepping (BMS ≥ 5), we additionally determined the BMS subscore which discriminates more precisely the fine details of locomotion such as coordination or paw position that may not be differentiated by the BMS 11,21 . The subscore starts with a minimum score of 0 whereas the maximum value is 11. Before testing, mice were acclimated to being handled and the open field environment. BMS tests were performed before the injury, on days 1, 3, 7, and then weekly (over eight weeks) after spinal cord injury. A BMS score of 0 at 1 d after injury and the absence of BDA staining in distal spinal cord cross-sections, as described above were used as quality criteria for completeness of the lesion.
Mice that did not meet these criteria were excluded from the analysis.

Statistics
Significances of intergroup differences were evaluated using Student's t-test or analysis of variance (ANOVA) followed by Holm-Sidak, or Tukey post hoc test using the Sigma STAT 3.1 software (Systat Software). Statistical significance of intergroup differences was defined as the following: *p < 0.05; **p < 0.01; ***p < 0.001. Statistical details for individual experiments are presented in the corresponding figure legend.

Video S1: Open field locomotion after SCC and AAV2-GFP treatment
The video shows a PTEN-floxed Ola mouse (PTEN +/+ ) that received an injection of AAV2-GFP into the left sensorimotor cortex at postnatal day 1. After 7 weeks, the mouse was subjected to complete spinal cord crush (SCC) (T8) and subsequently received an intracortical (left) injection of AAV2-GFP (see Fig. 1A). Videos were recorded at 1 and 8 weeks post SCC.

Video S2: Open field locomotion after SCC and AAV2-hIL-6 treatment
The video shows a PTEN-floxed Ola mouse (PTEN +/+ ) that received an injection of AAV2-GFP into the left sensorimotor cortex at postnatal day 1. After 7 weeks, the mouse was subjected to complete spinal cord crush (SCC) (T8) and subsequently received an intracortical (left) injection of AAV2-hIL-6 (see Fig. 1 A). Videos were recorded at 1 and 8 weeks post SCC.

Video S3: Open field locomotion after PTEN -/and SCC
The video shows a PTEN-floxed Ola mouse that received an injection of AAV2-Cre (PTEN -/-) into the left sensorimotor cortex at postnatal day 1. After 7 weeks, the mouse was subjected to complete spinal cord crush (SCC) (T8) and subsequently received an intracortical (left) injection of AAV2-GFP (see Fig. 1A). Videos were recorded at 1 and 8 weeks post SCC.

Video S4: Open field locomotion after PTEN -/and SCC with AAV2-hIL-6 treatment
The video shows a PTEN-floxed Ola mouse that received an injection of AAV2-Cre (PTEN -/-) into the left sensorimotor cortex at postnatal day 1. After 7 weeks, the mouse was subjected to complete spinal cord crush (SCC) (T8) and subsequently received an intracortical (left) injection of AAV2-hIL-6 (see Fig. 1A). Videos were recorded at 1 and 8 weeks post SCC.

Video S5: Open field locomotion after DHT-mediated depletion of raphe spinal input
The video shows a PTEN-floxed Ola mouse (PTEN +/+ ) that was subjected to complete spinal cord crush (SCC) (T8) and subsequently received bilateral intracortical injections of AAV2-hIL-6.
To destroy raphe spinal input, the animal was injected intracerebroventricularly with the serotonin neurotoxin 5,7-dihydroxytryptamine (DHT) 6 weeks after SCC. Videos were recorded at 1 and 6 weeks post SCC and at 1d and 1 week after DHT application. Additionally, another mouse of the same background that received a similar treatment but was injected with AAV2-GFP instead of AAV-hIL-6 was recorded 6 weeks after SCC and 1 day after DHT treatment.

Video S6: Open field locomotion determined in a BL6 mouse after SCC and AAV2-hIL-6 treatment
The video shows a C57BL/6J wildtype mouse (BL6) that was subjected to complete spinal cord crush (SCC) (T8) and subsequently received an intracortical injection of AAV2-hIL-6 into the left sensorimotor cortex. Videos were recorded at 1 and 8 weeks post SCC.

Video S7: Intracortical hIL-6 application induces STAT3 phosphorylation in raphe nuclei
Three-dimensional projection of transverse confocal scan through 150 µm of cleared brain stem tissue (presented as projection in Fig. 7D) after unilateral (left) intracortical injection of AAV2-hIL-6 and BDA treatment. Serotonin (5-HT) was stained in blue, phosphorylated STAT3 (pSTAT3) in red, and BDA labeled CST axons in green.