Secretion of a mammalian chondroitinase ABC aids glial integration at PNS/CNS boundaries

Schwann cell grafts support axonal growth following spinal cord injury, but a boundary forms between the implanted cells and host astrocytes. Axons are reluctant to exit the graft tissue in large part due to the surrounding inhibitory environment containing chondroitin sulphate proteoglycans (CSPGs). We use a lentiviral chondroitinase ABC, capable of being secreted from mammalian cells (mChABC), to examine the repercussions of CSPG digestion upon Schwann cell behaviour in vitro. We show that mChABC transduced Schwann cells robustly secrete substantial quantities of the enzyme causing large-scale CSPG digestion, facilitating the migration and adhesion of Schwann cells on inhibitory aggrecan and astrocytic substrates. Importantly, we show that secretion of the engineered enzyme can aid the intermingling of cells at the Schwann cell-astrocyte boundary, enabling growth of neurites over the putative graft/host interface. These data were echoed in vivo. This study demonstrates the profound effect of the enzyme on cellular motility, growth and migration. This provides a cellular mechanism for mChABC induced functional and behavioural recovery shown in in vivo studies. Importantly, we provide in vitro evidence that mChABC gene therapy is equally or more effective at producing these effects as a one-time application of commercially available ChABC.

mChABC modulates Schwann cell migration, adhesion and integrin signalling. Using the inverted coverslip migration assay, we confirmed previous findings, showing that Schwann cell migration and adhesion is dependent upon the substrate ( Supplementary Fig. 1a-e) [39][40][41][42] . The number and distance of migration was reduced (~ 20%) when grown on an inhibitory CSPG containing substrate as compared to laminin (Supplementary Fig. 1d). The aggrecan substrate was selected as an inhibitor of Schwann cell migration on astrocytes which is mediated through glial-cell produced aggrecan 41 . Interestingly, combining the laminin and aggrecan substrates led to a ~ 50% reduction in cell number and distance of migration compared to laminin alone. This suggests that the inhibitory effect induced by aggrecan on cellular migration can be partially overcome through the presence of positive migratory cues ( Supplementary Fig. 1a-e). These data were similar to the effect of substrate specific binding upon Schwann cell adhesion ( Supplementary Fig. 1h-j). Adhesion was inhibited on the aggrecan substrate, but significantly increased when conducted in the presence of laminin. This substrate specific effect was independent of cell division ( Supplementary Fig. 1f-g). The number and distance of Schwann cell migration was not different when cell division was inhibited with aphidicolin. These data demonstrate that experimental differences are substrate dependent.
To assess the effect mChABC has upon Schwann cell migration over inhibitory substrates, we performed the inverted coverslip migration assay using the transduced cell populations. When placed upon an inhibitory aggrecan substrate, naïve and fGFP transduced cells typically showed very little migration (~ 25%) and adhesion (~ 15%) of cells. However, mChABC secretion from cells enabled greater numbers of cells to migrate (~ 200%) a longer distance (~ 400%) than the control populations ( Fig. 2a-d). Indeed, these cells migrated the same distance as those on the aggrecan substrate pre-treated with bChABC (Fig. 2f). Interestingly, a greater number of Scientific RepoRtS | (2020) 10:11262 | https://doi.org/10.1038/s41598-020-67526-0 www.nature.com/scientificreports/ cells were found to migrate larger distances with mChABC than with bChABC (Fig. 2e). The increased number and distance of mChABC-expressing cellular migration compared to those on a bChABC-treated substrate was likely due to continuing modification of the ECM throughout the time of the assay. Similar effects were seen with cellular adhesion (Fig. 2g-k). The secretion of mChABC from Schwann cells enabled greater numbers of cells to adhere to the aggrecan substrate. These data were non-divergent from substrate which was pre-treated with bChABC (p > 0.05). These effects were likely caused by the removal of CS-GAG chains from the aggrecan substrate ( Fig. 2l-p). Both mChABC and bChABC were shown to expose the 1B5 stub, indicative of complete CS-GAG removal from the aggrecan substrate, with equal efficiency (Fig. 2n-p; p > 0.05).
To assess the mechanism through which mChABC removal of CS-GAGs may facilitate Schwann cell migration, we assessed integrin signalling. CSPGs have been shown to interfere with aggrecan and laminin binding to  www.nature.com/scientificreports/ cells 31,[43][44][45][46][47] . When active, integrins bind with high-affinity to their specific ligands, which can facilitate neurite outgrowth 48,49 . Further, Schwann cell migration has been shown to be integrin dependent 41 . To assess the effect of mChABC treatment upon this signalling pathway, we determined the level of naïve and transduced Schwann cell total FAK and phospho-FAK (pFAK) of the tyrosine 397 residue when grown on different substrates ( Fig. 2q-u). The tyrosine 397 residue on FAK is the first phosphorylation site after integrin activation. Its state of phosphorylation alters depending upon the substrate upon which cells are seeded 41,44,50 . Regardless of the cell type used or substrate upon which cells were grown, levels of total FAK (phosphorylated and non-phosphorylated) in all cells remained constant (Fig. 2q). Further, all cells showed increases in pFAK when grown on laminin (Fig. 2r) and naïve Schwann cells showed decreases in pFAK staining when grown on aggrecan and laminin/aggrecan substrates. However, cells secreting active mChABC showed significantly increased pFAK staining when grown on aggrecan or laminin/aggrecan compared to untransduced Schwann cells (Fig. 2r-u). The effect of mChABC was not significantly different from that caused by treatment of the cells with bChABC (p = 0.05). These data suggest that ChABC treatment mediates its effect upon Schwann cell migration partly through an integrin dependent pathway due to CS-GAG removal.

mChABC increases Schwann cell migration and adhesion on astrocytes.
With the success of mChABC to enable Schwann cell migration and adhesion on pure aggrecan, we next determined if similar results could be generated when Schwann cells were seeded onto an astrocytic substrate. We confirmed that naïve and fGFP-expressing Schwann cells show little migration and adhesion on astrocytic monolayers due to inhibitory molecules secreted by these cells (Fig. 3a-k). However, secretion of mChABC caused significant increases in the number of adherent cells and maximum distance of cellular migration (~ 500%, Fig. 3a-f). These data were not divergent from that obtained following pre-treatment of the astrocytic monolayer with bChABC (p = ns). Interestingly, performing a similar experiment on the astrocytic matrix produced a similar trend ( Fig. 3g-h). However, numbers of cells and maximum distance migrated were fewer once normalised to control cell (naïve) populations, although, more cells were able to travel, on average, a greater distance upon the matrix. Schwann cell migration is mediated through a balance of inhibitory and permissive factors. Most CSPGs produced by astrocytes are secreted and not retained on the cell surface 6,51,52 . These data suggest that the matrix produced by astrocytes alone contains components inhibitory to Schwann cell migration and adhesion. However, there are a greater number of positive cues retained on the surface of the glial cells which are exposed following CS-GAG removal which can aid cellular migration. Indeed, immunostaining revealed that while neither bChABC pre-treatment nor mChABC activity diminished laminin staining on the astrocytic monolayer ( Fig. 3l-n), the activity of both forms of the enzyme acted to expose the aggrecan protein core bound to the cell surface shown through anti-aggrecan staining ( Fig. 3o-q). Furthermore, this increase in staining associated with significant removal of CS-GAGs were confirmed by a reduction in CS-56 staining (Fig. 3r-t). The removal of CS-GAGs from the surface of the cellular monolayer, through the action of either mChABC or bChABC, was shown to cause an increase in the adhesion of Schwann cells to astrocytes (~ 340%, Fig. 3i-k). Together, these data show mChABC can digest inhibitory CS-GAGs from the astrocyte surface and ECM with the equivalent efficiency to commercial bChABC. The breakdown of these sugar chains facilitates Schwann cell migration and adhesion on astrocytes, two factors important in cellular boundary integration.

mChABC increases Schwann cell intermingling of cells at astrocyte boundaries.
In order to assess the effect mChABC has upon cellular intermingling, we performed the Schwann cell-astrocyte confrontation assay ( Fig. 4a-d). This closely models the barrier which forms between the two glial cell types following tissue transplantation 29,39,42,53 . We plated dense populations of naïve or transduced Schwann cells together with astrocytes 0.5 mm apart on a PDL and laminin coated slide. Using an established protocol 53 , the cell populations requiring bChABC treatment were exposed to 100 mU of the enzyme for 1 h daily while the boundary formed. Fresh medium was placed on all assays, regardless of treatment group, every 24 h. After 5 days, the cell populations had grown together, forming a distinct straight boundary with only a few Schwann cells penetrating the astrocyte population (Fig. 4a). All treatment groups exhibited more intense GFAP immunoreactivity at points of contact with anti-p75 positive Schwann cells. This suggests that neither mChABC nor bChABC caused a change in the astrocytic reaction to Schwann cell contact 54 . Nonetheless, secretion of the enzyme did affect cell behaviour. mChABC secretion from Schwann cells produced an increase in the number of Schwann cells intermingling with the astrocytes (~ 640%, Fig. 4b) and the maximum distance of migration into the astrocyte population ( Fig. 4e-f). This was most evident up to 400 μm from the edge of the boundary where mChABC was shown to facilitate a greater number of cells penetrating through the glial interface (p < 0.0001). Both mChABC and bChABC caused a similar reduction in CS-GAGs over both the Schwann cell and astrocyte (p = ns) populations shown through a reduction in CS-56 staining (Fig. 4a-d). These data suggest that secretion of mChABC from Schwann cells enables CS-GAG removal that can facilitate the penetration of Schwann cells into typically inhibitory astrocytic populations. mChABC increases neuronal outgrowth over PNS/CNS interfaces. The increased cellular intermingling at the Schwann cell astrocyte boundary caused by secretion of mChABC illustrates that through CS-GAG breakdown, the enzyme can overcome the inhibitory effects of CSPGs. In addition to their effects on cell migration, high concentrations of CSPGs cause growth cones to enter a dystrophic state which, while dynamic 55 , prevents axonal or neuronal growth 6,56-59 . To assess whether mChABC could mediate a return of neurite outgrowth due to CS-GAG digestion, dissociated DRGs were seeded upon confluent monolayers of naïve, mCh-ABC-transduced, or bChABC-treated Schwann cells and left to grow for 48 h (Fig. 5a-d). DRG neurite growth was closely associated with the long processes of Schwann cells in all treatment conditions. No significant differ-Scientific RepoRtS | (2020) 10:11262 | https://doi.org/10.1038/s41598-020-67526-0 www.nature.com/scientificreports/ ence in DRG morphology or the number of neurites produced by each cell body was noted between treatment groups ( Fig. 5a-d). However, secretion of mChABC from Schwann cells did increase the length of the DRG neurites (~ 500 μm increase; Fig. 5b-c). This was not significantly different from the effect caused by bChABC treatment of the Schwann cell monolayer (p = ns). These data confirm that, even on a substrate typically permissive to neurite outgrowth, CS-GAG breakdown can further facilitate the growth of sensory axons. Interestingly, secretion of mChABC from Schwann cells did not alter the myelination of neurites grown in culture (Supplementary Fig. 3). This would suggest that the properties of Schwann cells potentially beneficial to functional recovery following spinal cord injury have not altered despite mChABC transduction. Next, we determined if the mChABC mediated intermingling of Schwann cell and astrocytes would enable primary sensory neurons to grow over the glial cell interface (Fig. 5e-j). We seeded ~ 100 dissociated DRG neurons over the established Schwann cell-astrocyte boundary. Schwann cells were either naïve, transduced, or treated with bChABC. The DRGs were plated close to the boundary so that growing axons could have access to both cell types. Following DRG seeding, control cultures showed no increase in p75 positive cells in astrocytic territory, indicating that satellite p75 + cells did not migrate from the DRG cell body onto the astrocyte layer 53 . Away from the Schwann cell-astrocyte boundary, neurites grew readily on either cell type, albeit typically longer on Schwann cells. However, it was the behaviour of the neurites when forced to choose whether to cross from one cell-type to another at the Schwann cell-astrocyte interface that was interesting. When DRGs were initially plated upon naïve or fGFP expressing Schwann cells, the DRG neurites typically remained growing on this cell type with only ~ 11-13% of axons crossing to astrocytes (Fig. 5e, g). Further, less than 1% continued to grow on astrocytes for distances greater than 100 μm. Indeed, DRG neurites grew along Schwann cell processes, turning tight curves to avoid growth on astrocytic surfaces (Fig. 5f). However, in cultures where mChABC was secreted, a significantly greater number of DRG neurites crossed from Schwann cells to astrocytes (~ 17-19%; Fig. 5f-g). Further, more of these axons grew upon the astrocytic surface (p > 0.05). The effect mediated by the secretion of mChABC on DRG growth over the Schwann cell-astrocyte boundary was not significantly different from cultures treated with bChABC ( Fig. 5g). In contrast to these data, the ability of DRG neurites to cross from astrocytes to Schwann cells was substantial ( Fig. 5h-j). In cultures with naïve or fGFP expressing Schwann cells, ~ 95-98% of DRG neurites crossed from astrocytes to Schwann cells, with ~ 94-96% remaining on the latter population for over 100 μm (Fig. 5h, j). However, in cultures treated with either mChABC secretion or bChABC application there was a reduction in the percentage of neurites crossing from astrocytes to Schwann cells (~ 92-94%; Fig. 5ij). Further, slightly fewer (~ 89-91%) neurites remained growing upon the Schwann cells. These data suggest that the removal of CS-GAGs together with the deep penetration of cells through the Schwann cell-astrocyte boundary can cause a modest, yet significant increase in the number of axons that grow over the Schwann-cell astrocyte boundary. Further, this treatment modestly facilitates the growth of DRG neurites on typically less permissive astrocytic cells. The confrontation assay represents an in vitro model of axonal behaviour at the dorsal root entry zone (DREZ) and the graft-host interface following Schwann cell transplantation into an injured spinal cord. mChABC facilitates in vivo Schwann cell intermingling. To assess the degree to which intraspinal LV delivery of mChABC could facilitate neuronal growth and Schwann cell migration following experimental SCI compared to intraspinal injection of bChABC, we performed a thoracic contusion in adult rats ( Supplementary  Fig. 4). Two weeks following injury, LV-fGFP injected animals showed abundant CS-GAG chains around the lesion epicentre ( Supplementary Fig. 4a). Substantially lower levels of these inhibitory molecules where shown following bChABC and LV-mChABC application ( Supplementary Fig. 4b-c). These data demonstrate that LV-mChABC may be successfully transduced into adult cells in vivo and secreted in an active form which operates effectively. Further, the removal of CS-GAGs extended over large distances from the lesion site, indicative of both the spread of the enzyme in the tissue and the area of viral transduction (shown through fGFP; Supplementary  Fig. 4d). Application of bChABC was shown to reduce the size of the lesion cavity following SCI as compared to controls ( Supplementary Fig. 4d-e). This effect was augmented in mChABC animals who displayed not only a reduction in cavitation after SCI but also substantial axonal staining through the lesion site itself signifying a reduction in neuronal loss and potential sprouting ( Supplementary Fig. 4f, e i -f i ). These observations support our in vitro data, demonstrating that mChABC can facilitate neuronal growth within and across inhibitory environments and has the potential to exhibit superior effects to the bacterial enzyme.
Finally, we assessed the migration of Schwann cells into the astrocyte rich lesion penumbra from the periphery following SCI. Control animals showed no evidence of Schwann cell migration or intermingling around the lesion border and penumbra ( Supplementary Fig. 4g). After administration of bChABC, a small number of Schwann cells were observed in this region ( Supplementary Fig. 4h). However, in mChABC transduced animals a substantial number of Schwann cells were shown to intermingle with astrocytes in and around the lesion penumbra ( Supplementary Fig. 4g). This study confirms our in vitro data, demonstrating that mChABC can facilitate the intermingling and growth of cells and has a potentially more profound effect than a one-time dose of the bacterial enzyme. Further, this demonstrates that mChABC can yield similar effects on adult cells in vivo as those demonstrated by postnatal animals in vitro.

Discussion
Here we demonstrate that Schwann cells transduced with lentiviral vectors encoding ChABC can constitutively secrete high quantities of active mChABC. The activity of this second-generation enzyme causes profound CS-GAG removal in vitro, functionally altering Schwann cell migration, adhesion, and intermingling within inhibitory astrocytic environments whilst enabling neurite outgrowth over the PNS/CNS cellular interface. The mechanisms of mChABC on cellular activity explain the behavioural and regenerative effects of the enzyme  www.nature.com/scientificreports/ in biochemical and in vivo studies. Importantly, we demonstrate that the effects produced from our modified mChABC enzyme are equivalent to, or greater than, the commercially available bChABC. The functional expression and effects of mChABC secretion on the PNS/CNS boundary has not previously been shown. The mChABC mediated removal of inhibitory CS-GAGs from aggrecan, or astrocytic monolayers 41,42,53 increased Schwann cell motility. Through GAG digestion, mChABC facilitated cellular-substrate attachment, resulting in increased numbers and distance of cellular migration 60 . Further, and superior to bChABC [61][62][63] , the action of mChABC facilitated the intermingling of Schwann cells with astrocytes at points of confrontation. Moreover, unlike bChABC 53 , mChABC additionally enabled astrocytes to migrate into a Schwann cell population. This would suggest that the continuous, global activity of mChABC is required to facilitate the migration of CNS glia, perhaps due to modifications to the ECM or cell-to-cell contact. This is echoed in populations of adult cells shown in our in vivo observations where migration and adhesion is prevented in part due to inhibitory CS-GAGs and inactivation of integrins [64][65][66][67][68] . As mChABC can influence these factors regardless of time following development or injury, our data support a universal process of mChABC-mediated GAG removal to aid the intermingling and growth of cells. Within our assay, the action of mChABC facilitated neuronal growth through CS-GAG removal and, potentially, by increasing neuroprotection 14 . The reduced inhibition and border irregularity enabled neurites to cross the PNS/CNS cellular boundary. This mechanism of mChABC action explains the similar in vitro and in vivo 29,53,69 data and was supported by observations collected from our own in vivo work. mChABC gene therapy following cervical and thoracic contusion injury has been shown to remove CS-GAGs, increasing neuroprotection through modulation of macrophage phenotype, and facilitate significant locomotor and sensorimotor function as demonstrated through ladder walking, axonal conduction, and grip strength 21,22 . Recently, it has been shown that use of an inducible mChABC vector facilitated sensory conduction and gross movement following short-term treatment, however, long-term treatment was required to evoke significant effects on skilled reaching and grasping 23 . We demonstrate that the means for this recovery is likely to be based upon the growth or sprouting of neurites through previously inhibitory substrates. Further, our study validates an in vitro and in vivo study where an autologous population of lentiviral mChABC transduced Schwann cells were implanted into the spinal cord following thoracic contusion injury 24 . Similar to our work, Kanno et al. 24 www.nature.com/scientificreports/ demonstrated that mChABC removes CS-GAGs within the graft and surrounding host tissue and facilitates the integration of Schwann cells with host astrocytes. The degree and distance to which these cells intermingle is accurately reflected in our in vitro and in vivo findings. Indeed the echoing of the in vitro data with our in vivo observations and other studies suggests that, although our in vitro functional assays are simplistic models of the complex network, matrices, and cellular, chemical and electrical interactions which occur in the animal, they yield fundamental and translational information. Our data suggest that Schwann cells are able to integrate into this environment as they can myelinate neurites. However, a complete demonstration of integration in inhibitory environments would also illustrate local trophic support and formation of cellular scaffolds produced by the cells. Nevertheless, here we provide evidence of a cellular mechanism governing the intermingling of cells at inhibitory boundaries caused by the changes in cellular migration and adhesion of the Schwann cells. Further, we have identified a mechanism through which Schwann cell migration, adhesion and intermingling can be affected by ChABC. However, there are many additional processes that could be affected by use of the enzyme, for example, effects on all ECM components, signalling pathways, ligands, receptor binding and autophagy 2,64,70 . These need to be addressed to determine all the mechanisms through which mChABC may mediate effects at the CNS/PNS boundary. bChABC has been used to great effect in combination with peripheral nerve [71][72][73][74] and Schwann cell 61,75 grafts to yield functional recovery following severe SCIs. However, for the first time we demonstrate that secretion www.nature.com/scientificreports/ of mChABC is more effective than a single application of bChABC at producing these effects. These data have substantial weight in the development of genetic modification through mChABC as a treatment for SCI. However, a full biochemical analysis of mChABC activity, as compared to the commercially available bChABC, is still required. Accounting for number of cells and vector particles used, the transduced Schwann cells in this study produce similar yields of active mChABC to that shown in vivo 24 , demonstrating the robust application of this genetic modification treatment. Moreover, we uniquely demonstrate that these cell populations produce constant amounts of enzyme over time. Single aliquots of enzyme have also been shown to retain activity for longer at 37 °C than bChABC, suggesting stability. These are similar data to forms of the enzyme stabilised by trehalose 17 . This modified stability at biological temperatures means that any amount of enzyme produced from transduced cells would remain active for longer, increasing applicability of the treatment approach. This is important as in vivo studies suggest that, while the functional and anatomical effects of the enzyme are present at 8-12 weeks, lentiviral mChABC has reduced cellular production 3 months following transduction 21,24 . These data suggest that purified conditioned medium from mChABC transduced cells might be a viable alternative to treatment with commercially available purified bChABC.
The action of bChABC and mChABC within this study has revealed further insight into the mechanisms of CSPG mediated inhibition of cellular migration and growth. While known to exert cellular effects due to their strong negative charge 76 , CSPGs have been shown to alter function through receptor binding [77][78][79][80] . CSPGs also mediate effects through integrin-dependent mechanisms 31,41,44,51 . Here we have shown that when grown on an aggrecan substrate, Schwann cell phospho-FAK is reduced, while total FAK levels within the cells remains constant, suggesting a reduction in intracellular integrin signalling. Interestingly, we further demonstrate that the action of mChABC and bChABC can rescue this effect suggesting that it is the aggrecan GAG chains, and not the core protein, that interferes with integrin signalling. This demonstrates that, despite genetic modification, both mChABC and bChABC operate through similar mechanistic pathways. Utilising this in vitro model, determining the complete downstream intracellular pathways affected by CSPG binding on cytoskeletal dynamics and how they are affected by mChABC activity would provide significant insight.
Our data support the continued development of second-generation ChABC enzymes for the treatment of SCI. There have been a number of viral chondroitinase AC and ABC vectors developed [81][82][83] . However, we have shown that mChABC uniquely demonstrates a high and robust yield and substantial functional effects in vitro and in vivo. However, for additional effects on neuronal growth our study suggests that mChABC should be combined with treatments known to modulate Schwann cell and axonal behaviour such as integrins, N-cadherins, and Ephs 31,42,84,85 . The effectiveness of lentiviral mChABC treatment both in vitro and in vivo shows the significance of modifying the ECM in the treatment of SCI. Here we have explained a cellular mechanism underlying the functional and anatomical recovery shown in mChABC use. However, to further develop this as a clinical treatment, the construct should be optimised, perhaps by placing it in an increasingly stable vector such as an adeno-associated virus 86 , or assessing the risk of insertional mutagenesis in host cells 87 .
In summary, we provide evidence demonstrating the potent effects that lentiviral mChABC has upon glia cell dynamics and behaviour, facilitating cellular intermingling and neurite growth. These data highlight the importance of matrix modification in the treatment and recovery following SCI and provides a cellular mechanism for mChABC induced functional and behavioural recovery shown in in vivo studies. Furthermore, we demonstrate that this modified enzyme is stably secreted in high yield with functional activity similar to, or greater than, the commercially available bChABC. These findings have significant implications for the continued development of second-generation enzymes for the treatment of SCI and other CNS diseases. Tables 1 and 2   www.nature.com/scientificreports/ were performed with the individual LV-fGFP and LV-mChABC vectors. The following day, the solution was washed off the cells.

Antibodies and reagents. Antibodies used are detailed in
neonatal astrocytes. The cerebral cortices were excised from neonatal rats, the meninges removed, the tissue diced and incubated with 0.1% trypsin for 30 min. The solution was triturated, centrifuged, the cells suspended in DMEM with 10% FCS, and plated on PDL coated flasks. Contaminating oligodendrocyte precursor cells and microglia were removed by shaking the culture overnight at 500 rpm in a 37 °C incubator shaker (Luckham R300). The astrocytes were maintained in DMEM supplemented with 10% FCS and 2% PSF. Astrocytes displayed a mature/reactive phenotype demonstrated through GFAP staining and production of laminin and CSPGs demonstrated through CS56 and aggrecan staining (Fig. 3).
neonatal dorsal root ganglion (DRG) neurons. The 30 neonatal DRG pairs were extracted and dissociated. The suspension was centrifuged through a 15% BSA density gradient at 100 × g and the cells suspended in DMEM supplemented with ITS + (1:100), PSF (1:100) and the growth supplement NGF (10 ng/mL). Cells were plated on to PDL and laminin (1 μg/mL) coated flasks and incubated at 37 °C with 7% CO 2 .
Functional assay design and quantification. Schwann cell inverted coverslip migration assay. Schwann cells (1 × 10 6 cells/mL) were plated on PDL and laminin coated coverslip fragments (~ 1 mm 2 ) in a 20 μL droplet and incubated until confluent. Test substrates were coated with PDL and either aggrecan (20 μg/mL), laminin (1 μg/mL), or laminin and aggrecan. Coverslip fragments (with cells) were inverted on the test substrates, and cells were allowed to migrate for 24 h in DMEM with 10% FCS before live imaging, analysis, and fixation in 4% paraformaldehyde (PFA). When required, the ECM or astrocyte monolayer was treated with 0.2 U/mL bChABC for 1 h at 37 °C. In assays requiring a mitotic inhibitor, aphidicolin (1 μg/mL; or equivalent volume of DMSO) was supplemented to the sDMEM during migration. When migrating Schwann cells upon a confluent monolayer of astrocytes, the PNS glia were labelled with vibrant DiI (Thermo Fisher Scientific). The confluent astrocyte monolayer was confirmed through microscopy and coverslip fragments only plated on areas know to be confluent. All data was normalised to control conditions (naïve Schwann cells migrated upon a PDL). Schwann cell migration was assessed live under phase contrast (10 × magnification, Nikon) or fluorescent microscope (10 × magnification, Leica). Quantification using an eyepiece grid reticule (Nikon) included: the maximum distance of cell body migration from the edge of a coverslip fragment (assessed at 90° to the edge of the coverslip), and the number of cells migrating at various distances (assessed in 100μm 2 bins; Supplementary Fig. 5a).
Preparation of astrocytic monolayers and matrix. Purified astrocytes (2 × 10 5 cells/mL) were plated on PDL coated coverslips and left for 48 h to form dense monolayers. For immunocytochemistry, medium was replaced with that collected and concentrated from transduced primary Schwann cells or cDMEM supplemented with 0.2 U/mL bChABC for 1 h at 37 °C, before fixation with 4% PFA. For the preparation of astrocytic matrix, the confluent astrocyte monolayers were lysed by hypotonicity through 45-50 min exposure to sterile water. The matrices were washed 3 times to remove debris and stored at 4 °C in PBS for maximum of 24 h until use.
Confrontation assay. Sterile silicone gaskets with 9 mm 2 wells (Sigma) were placed on Permanox™ slides (Nunc, LabTEK). 2-hole silicone cell culture inserts (Ibidi) were placed into the PDL coated wells. Purified primary Schwann cells (7 × 10 5 cells/mL) and astrocytes (4.6 × 10 5 cells/mL) were each plated into one of the insert holes, 0.5 mm apart. After 12 h, the insert was removed and cells washed. Cultures were grown for 5 days in sDMEM until a clear boundary was established. If required, bChABC (100 mU) was placed in the cell medium for 1 h at 37 °C daily. For experiments assessing neurite outgrowth, dissociated DRGs (1 × 10 4 cells/mL) were plated over the boundary. Cells were grown at 37° for 2 days in DMEM supplemented with 2% PSF, 10% FCS, BPE (10 μg/ mL), and NGF (10 ng/mL) with medium changed every 24 h. All experiments were terminated through cell fixation with 4% PFA. All data was normalised to control conditions (naïve Schwann cells and astrocytes). Experiments were analysed using fluorescent microscopy (Leica6000; following ICC) using Adobe Illustrator (CS5) via a methodology described previously 31 (Supplementary Fig. 5b). The number of Schwann cells migrating into the glial cell population was counted at distances of 200, 400 and greater than 400 μm (63 × magnification). Neurite growth on Schwann cell-astrocyte confrontation assays was quantified using previously specified criteria 29 (Supplementary Fig. 5c). A minimum of 10 neurites were counted per image.
Adhesion assay. Primary Schwann cells (2 × 10 5 cells/mL) were plated on each coverslip and left to adhere at 37 °C for 1 h. The coverslips were then shaken with randomized rotation (100 rpm; Eppendorf) for 4 h at 37 °C, then washed to remove any non-attached cells. If required, 0.2U/mL bChABC was applied to the coverslip for 1 h at 37 °C. Assays conducted on pure substrate were analysed live using a phase contrast microscope (5 × magnification, Nikon) while those conducted on astrocyte monolayer were fixed with 4% PFA and then stained before fluorescent imaging (5 × magnification, Leica) and analysis. All data was normalised to control conditions (naïve Schwann cells adhering to PDL or astrocyte monolayer). Adhesion was assessed by scoring the number of cells in a defined 100 μm x 1000 μm area. A minimum of 15 images at random locations on each coverslip were taken at time points both before and after shaking. in vivo assessments. Animals. All animal surgeries were conducted in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986. Adult female Sprague Dawley rats (200-220 g; Harlan Laboratories) were housed in groups of three or four, exposed to a normal 12-h dark-light cycle at 21 °C with free access to food, water and environmental enrichment ad libitum. The health and welfare of all animals was monitored daily by veterinary staff and the study investigators at Kings College London. Experimenters were blinded to treatment conditions during testing and histological assessment and sterile procedures were observed during all surgical techniques.

Contusion injury and intraspinal injections.
Animals were anaesthetised with ketamine (60 mg/kg) and medetomidine (0.25 mg/ml). Upon reaching a surgical plane of anaesthesia, animals received a subcutaneous (s.c.) injection of Carprofen (5 mg/kg). Body temperature was maintained throughout the surgery at 37 ± 1 °C using a homothermic heat pad (Harvard Apparatus). Animals received a 150 kdyne midline contusion at T10 (Infinite Horizon impactor; Precision Systems Instrumentation). Following contusion, rats received intraspinal injections (2 × 0.5 μl; 1 mm rostral and caudal to the injury site) of protease-free bChABC (10 U/ml, n = 3, Seikagaku); LV-mChABC (n = 3) or LV-fGFP (n = 3). Animals were given s.c. injections of saline and recovered in a heated environment before transfer to their home cage. Analgesia and hydration were maintained for 5 days post-surgery in addition to nutritional support if the animal's weight fell 5% below that determined pre-injury.
No animal showed any adverse effects to the surgery or procedures performed.
Tissue processing. Two weeks following injury, animals were anesthetised (Euthatal, 80 mg/kg, i.p.) and transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer and the spinal cord removed. Following 2 h post-fixation, tissue was cryoprotected in 20% sucrose for 48 h. Following embedding in Optimum Cutting Temperature Compound, the spinal cord was cut into serial 20 μm sagittal sections.
immunocytochemistry. Cells/tissue were fixed for 10 min in 4% PFA, washed, and blocked in 0.3% Triton-X 100 (Sigma) and either 3% bovine serum albumin (BSA; Sigma) or 10% donkey serum (Stratech). Cells/tissue were incubated with primary antibody overnight at 4 °C and secondary antibody for 2 h at 25 °C. If required, Hoechst-33342 applied to the cells for 10 min at 25 °C. Coverslips were mounted onto slides with Fluorosave (Calbiochem) while tissue was mounted using ProLong Gold (ThermoFisher Scientific). For experiments with a biotinylated secondary, cells were additionally incubated in streptavidin for 2 h at 25 °C. Cells were analysed under fluorescent microscopy (Leica6000; 20, 40 and 63 × magnification). A minimum of 20 images were taken at random locations on each image. Images were analysed using ImageJ (NIH) software with the integrated pixel density assessed in a minimum of 10 areas positive for the epitope. Signal intensity readings were standardised for area and background. To determine the rate of cell division and transduction efficiency, the total number of cells was determined through Hoechst-33342 staining. Tissue was assessed under confocal microscopy (Zeiss; 20 × magnification).
cetylpyridinium chloride (cpc) turbidity assay. Cell culture medium (DMEM without phenol red, supplemented with sodium pyruvate and ITS +) was collected from naïve and transduced Schwann cells over 24-168 h at 37 °C. EDTA-free protease inhibitor cocktail (Roche) was added and the supernatant and concentrated using a 50 K centricon (Millipore). 5μL media samples were incubated with 50μL chondroitin sulphate A (CS-A; 20 μg; Sigma) at 37 °C for 30 min and denatured at 95 °C. 20μL of each sample was incubated with an equal volume of CPC reagent (containing 1:1 of 0.2% (w/v) CPC and 133 mM magnesium chloride; Fluka). Absorbance was measured at 405 nm using a μQuant™ Microplate Spectrophotometer (Biotek Instruments) and data adjusted for baseline based on the negative controls 89 .