Low sulfated heparins target multiple proteins for central nervous system repair

Abstract The lack of endogenous repair following spinal cord injury (SCI) accounts for the frequent permanent deficits for which effective treatments are absent. Previously, we demonstrated that low sulfated modified heparin mimetics (LS‐mHeps) attenuate astrocytosis, suggesting they may represent a novel therapeutic approach. mHeps are glycomolecules with structural similarities to resident heparan sulfates (HS), which modulate cell signaling by both sequestering ligands, and acting as cofactors in the formation of ligand–receptor complexes. To explore whether mHeps can affect the myelination and neurite outgrowth necessary for repair after SCI, we created lesioned or demyelinated neural cell co‐cultures and exposed them with a panel of mHeps with varying degrees and positions of their sulfate moieties. LS‐mHep7 enhanced neurite outgrowth and myelination, whereas highly sulfated mHeps (HS‐mHeps) had attenuating effects. LS‐mHeps had no effects on myelination or neurite extension in developing, uninjured myelinating cultures, suggesting they might exert their proregenerating effects by modulating or sequestering inhibitory factors secreted after injury. To investigate this, we examined conditioned media from cultures using chemokine arrays and conducted an unbiased proteomics approach by applying TMT‐LC/MS to mHep7 affinity purified conditioned media from these cultures. Multiple protein factors reported to play a role in damage or repair mechanisms were identified, including amyloid betaA4. Amyloid beta peptide (1–42) was validated as an important candidate by treating myelination cultures and shown to inhibit myelination. Thus, we propose that LS‐mHeps exert multiple beneficial effects on mechanisms supporting enhanced repair, and represent novel candidates as therapeutics for CNS damage.

forming a glial scar rich in reactive astrocytes, myelin-associated inhibitors, microglia, macrophages, and meningeal fibroblasts which play a major role in sequestering damaged tissue in a relatively impermeant environment (Filous & Silver, 2016). Thus, any repair strategies for SCI must consider multifactorial pathways including disruption of the scar, promotion of axonal outgrowth, and remyelination.
Previously, we used a confrontation assay to study the effects of astrocytosis, characterizing boundary formation, expression of GFAP, and astrocyte hypertrophy (Lakatos, Franklin, & Barnett, 2000;Wilby et al., 1999). We found that Schwann cells do not mingle with astrocytes whereas olfactory ensheathing cells (OECs) mingle well (Lakatos et al., 2000). We identified heparin and FGFs as strong inducers of boundary formation between astrocytes and OECs and concluded that highly sulfated heparan sulfates (HS) secreted by Schwann cells induced the astrocyte-Schwann cell boundary (Fairless, Frame, & Barnett, 2005;Santos-Silva et al., 2007).
HS are linear sulfated polysaccharides that play an important role in regulating many mammalian cellular processes. They exist as proteoglycans (HSPG) in which two or three HS chains are attached to core proteins at the cell surface, or extracellular matrix (ECM) proteins. It is thought that the pattern, epimerization, and degree of sulfation is important in facilitating HS-ligand interactions, thereby enabling HS to function as a key regulator of complex cell signaling mechanisms in development, regulation of chemokine function, angiogenesis, and blood coagulation, in both normal and disease states (Bishop, Schuksz, & Esko, 2007;Changyaleket, Deliu, Chignalia, & Feinstein, 2017;Cui, Freeman, Jacobson, & Small, 2013). To investigate how HS sulfation characteristics influence astrocytosis, we applied a panel of modified heparins (mHeps) which have been selectively desulfated to the confrontation assay. We found that mHeps with low sulfated (LS-) moieties could attenuate the astrocytic behavior, whereas highly sulfated (HS-) mHeps induced a strong reactive astrocyte response in culture (Higginson et al., 2012). This led us to hypothesize that modifying the level of HS sulfation in the vicinity of astrocytes might be a therapeutic approach to promoting one pathway involved in CNS repair (O'Neill et al., 2017).
Although the above addresses effects on astrocytosis, if mHeps are to be developed for treatment of the injured CNS, there is a need to investigate possible effects on other neural cells. Therefore, we studied the effect of mHeps on myelinating cultures generated from dissociated spinal cord cells. These basic cultures are referred to as myelinating cultures-development (MC-Dev). Second, an adaptation of MC-Dev can be used to model aspects of CNS injury by cutting mature cultures with a scalpel blade, generating a cell-free area characterized by a persistent lack of neurite outgrowth. Over time, the cell-free area becomes infiltrated with reactive astrocytes, and microglia, as seen in animal models of CNS injury. These cultures are termed MC-Inj (Boomkamp, Riehle, Wood, Olson, & Barnett, 2012). Finally, mature myelinating cultures can be used to investigate demyelination by treating them with anti-MOG antibody plus complement which resulted in demyelination of axons (Elliott et al., 2012;MC-Demy).
When the panel of differentially sulfated mHeps was applied to these three cultures types, we found that LS-mHeps not only attenuate astrocytosis but also exert positive effects on neurite outgrowth and remyelination suggesting these glycomolecules may be an important novel therapeutic for CNS damage. Using tandem mass tag (TMT) labelling and liquid chromatography-mass spectrometry (LC-MS) and chemokine array of secreted factors from the two injury culture models, we identified differences in heparin binding proteins (HBPs) suggesting subtle difference in mechanisms.
The cell suspension was maintained until NS were formed. To generate astrocytes, the NS were triturated to produce smaller cell sphere suspensions, transferred to 13 mm poly-L-lysine (PLL; 13 μg/ml, Sigma) coated coverslips in a 24-well plate (Corning, UK) and incubated for a further 5-7 days in vitro (DIV) at 37 C in an atmosphere of 7% CO 2 /93% air until a confluent monolayer formed. NS-derived astrocytes were maintained in DMEM-1 g/ml glucose (Life Technologies) with 10% fetal bovine serum (Sigma, Poole, Dorset, UK) and 2 mM L-glutamine (Sigma).

| Modified heparins
Selectively chemically desulfated mHeps were prepared as described previously (Higginson et al., 2012) and illustrated in Table 1

| Myelinating spinal cord cultures (MC-Dev and MC-Inj)
Generation of rat spinal cord mixed cell cultures (myelinating cultures [MC]) was based on our previously described methods (Boomkamp et al., 2012;Boomkamp, McGrath, Houslay, & Barnett, 2014;Sorensen et al., 2008). The spinal cord of E15.5 SD embryos were enzymatically dissociated and the resulting cell suspension at 300,000 cells per 50 μl, were plated on top of the NS derived astrocytes on coverslips in plating medium (PM) that contained 50% DMEM-1 g/ml glucose, 25% horse serum (Invitrogen), 25% HBSS (with Ca 2+ and Mg 2+ , Life Technologies) and 2 mM L-glutamine. Cells were left to adhere for 2 hr at 37 C, then supplemented with 300 μl PM and 500 μl differentiation medium which contained DMEM (4.5 g/ml glucose, Life Technologies), 10 ng/ml biotin (Sigma), 0.5% hormone mixture (1 mg/ml apotransferrin, 20 mM putrescine, 4 μM progesterone, 6 μM selenium (formulation based on N2 mix of [Bottenstein & Sato, 1979]) 50 nM hydrocortisone, and 0.5 mg/ml insulin known as DM+, or DM − if lacking insulin (all reagents from Sigma). Each 35 mm Petri dish containing three myelinating culture coverslips was fed three times a week by removing 400 μl of medium and adding 500 μl fresh DM+ for 12 DIV then DM− for the proceeding 12 DIV. Cultures were maintained for 24-35 DIV in an atmosphere of 7% CO 2 at 37 C and referred to as MC-Dev. Over time these cultures elaborate myelinated fibers interspaced with organized nodes of Ranvier (Sorensen et al., 2008), with normal periodicity and compaction .
To assess the effect of mHeps on developmental myelination, cultures were treated at 13 DIV and 20 DIV (1 ng/ml). Furthermore, MC-Dev cultures underwent a single treatment at 24 DIV to assess if mHeps (1 ng/ml) had any effects on mature myelin. Following which both sets of cultures were fixed and stained at 28 DIV. MC-Dev cultures were treated with amyloid beta (Aβ)-peptide (1-42) rat (TOCRIS-2425) at 16, 19, and 21 DIV (1 μM) or combined with 100 ng/ml mHep7 during feeding. The cultures were fixed and stained at 18, 20, 22, and 24 DIV for myelin quantification (described below). Additionally, conditioned media was collected for CytoTox 96 ® NonRadioactive Cytotoxicity Assay (Promega G1780).
For generating MC-Inj, at 24 DIV myelinating cultures were cut using a 11 mm single edge razor blade (WPI, Aston, UK) pressed gently across the center of the coverslip. Details can be found in Boomkamp et al., 2012Boomkamp et al., , 2014. The cut created a focal cell-free area (650 μm) with a decrease in neurite density and myelination levels adjacent to it and very low numbers of neurites crossing the cut area site, referred to as the lesion. Using these cultures lesion size, neurite density and outgrowth, and myelination can be assessed using immunocytochemistry.
MC-Inj responds appropriately to treatment with compounds reported to be effective in animal models of CNS injury, corroborating MC-Inj as a moderate throughput screen for CNS injury (Boomkamp et al., 2012(Boomkamp et al., , 2014. The cultures were treated with each mHep at a concentration of 1 ng/ml for a single treatment after 25 DIV and allowed to recover for a further 5 DIV, cultures were then fixed and stained as described below.

| Demyelinated cultures (MC-Demy)
To assess remyelination the myelinating cultures were set up and maintained as described above. At 24 DIV, when many fibers are myelinated, the cultures were demyelinated by overnight incubation with the Z2 antibody (100 ng/ml Hybridoma, a kind gift from Prof C Linington) which recognizes myelin oligodendrocyte glycoprotein (MOG) and rabbit complement (100 μg/ml, Millipore) at 37 C. The demyelinated cultures were then washed with DM− to remove excess complement and treated with the panel of mHeps at 1 ng/ml. At 30 DIV, the cultures were fixed and stained as described below.

| Conditioned medium collection
To obtain medium conditioned by the various cultures, MC-Inj and MC-Demy were set up. Conditioned media was collected after 25 DIV, for the cut conditioned medium (CCM) and demyelination conditioned media (DCM); this was 24 hr after cultures had been cut and demyelinated respectively. Uninjured culture conditioned medium (UCM) was also collected from myelinating cultures at 25 DIV. CM was also taken after treatment at 26 and 28 DIV (corresponding to days 1 and 3 posttreatment). The CMs were added to MC-Dev at 16, 19, and 21 DIV. CM was diluted 1:4 with DM− or combined with 1 ng/ml mHep6. At 24 DIV, the cultures were fixed and stained as described below.

| Cytokine array screen
Conditioned medium from each type of MC was collected as described above and 1.5 ml was assayed in a Proteome Profiler Rat XL Cytokine Array (R&D Systems, ARY030) using the manufacturer's protocol. The array simultaneously detects levels of 79 rat cytokines, chemokines, growth factors, and other soluble proteins (see https:// www.rndsystems.com/products/proteome-profiler-rat-xl-cytokine-array_ary0300. Comparisons of CCM and DCM were made to UCM, using semi-quantitative densitometry analysis on Total Lab Quant Software (TotaLab Limited). Using the circle tool each dot was individually quantified with duplicates averaged and the integrated density values compared.  2.7 | Oligodendrocyte progenitor cell culture Oligodendrocyte progenitor cells (OPC) were isolated from cortical astrocyte monolayers generated from postnatal day 1 (P1) SD pups by differential adhesion as previously described (Miron et al., 2013;Noble & Murray, 1984). OPCs were maintained in serum-free DMEM-BS (adapted from Bottenstein & Sato, 1979) containing 0.5 mg/ml insulin in 10 mM HCL (Sigma, UK), glutamine (100 mM, Sigma), human transferrin (0.1 mg/ml, Sigma) and gentamycin (100 mg/ml, Sigma), supplemented with the growth factors; fibroblast growth factor (FGF-2) at 10 ng/ml and platelet derived growth factor (PDGF) at 2 ng/ml (both Peprotech, UK). The isolated OPCs were plated on poly-L-lysine (PLL, 13 μg/ml, Sigma) coated glass coverslips (VWR) in a 24-well plate at a density of 5,000 cells in 50 μl drop and allowed to attach.
They were maintained in DMEM-BS containing PDGFα and FGF2 for 5 days and then switched to DMEM-BS lacking growth factors and with or without mHeps at a concentration of 1 ng/ml. Cultures were used for proliferation, morphology, and differentiation assays.

| Nanofibres
OPCs were seeded onto PLL coated nanofiber culture inserts
For neurite density, the threshold level pixel value for SMI31 immunoreactivity (IR) was divided by the total number of pixels. The percentage of myelinated axons (PLP) was measured using CellProfiler, which uses pattern recognition software to distinguish between linear myelinated internodes and oligodendrocyte cell bodies. In this manner, we track the co-expression of myelin sheaths (PLP) and axons (SMI31) and calculate this percentage of myelinated fibers. All experiments were carried out at least three times in duplicate. For OL quantification images adjacent to the lesion were used and OL cells were calculated using a Cellprofiler pipeline which counts the presence of PLP+ cell bodies overlapping DAPI nuclei. OPC/OL cultured on nanofibers were analyzed for cell size and myelin internode length. Both were calculated using ImageJ software. Two measurements per cell were taken from the cell body to the outmost myelin internode 20 cells were used per image selected at random. The cell size was defined by the green (PLP+) pixel threshold compared to the total pixel intensity and individual cells were marked with the region of interest (ROI) tool to allow single cell analysis. All CellProfiler pipelines used in this study are available at https://github.com/muecs/cp.

| Neurite outgrowth and lesion size quantification
Ten images of the cut in the MC-Inj were collected using random sampling. Neurite outgrowth was defined as a SMI31 positive projection which enters and crosses the lesion site. In each image, the number of neurites which cross the lesion site was counted. Any area around the lesion that appeared uninjured was excluded from analysis. The number of neurites per image was averaged across the lesion and termed neurite outgrowth per field of view. Using the same images, the width of the lesion was calculated using ImageJ at 10 fields of view per lesion, averaged and termed lesion width (μm).
2.14 | HBPs pull down and mass spectroscopy analysis CCM, DCM, and UCM were affinity purified on an mHep7 column. To make the column a commercial HiTrap NHS-activated HP column (1 ml, GE Healthcare) was washed extensively with ddH 2 O (20 ml) and 10 mg (1 ml) of mHep7 was introduced onto the column and allowed to react at 15 C for 2 hr. Any unbound mHep7 was washed off the column with ddH 2 O (20 ml) and a small soluble amine (5% ethanolamine in ddH 2 O, Sigma) was added to react with any remaining N-hydroxysuccinimide (NHS) groups at 15 C for 2 hr. The column was washed with PBS (pH 7.5, 20 ml) and ready for use to capture mHep7 binding proteins. Six milliliters of the different CM was run down the column with a peristaltic pump, following the manufacturer's instructions. Binding buffer and elution buffer was 10 mM sodium phosphate, pH 7 and 10 mM sodium phosphate, 1 M NaCl pH 7, respectively (Both from VWR). The eluted HBPs were desalted and concentrated using Amicon Ultra-15 centrifugal device (3K, 15 ml; Millipore, UK). The proteins were digested with trypsin, using the Peptide separation was performed on a Pepmap C18 reversed phase column (LC Packings). The output from the LC-MS/MS was deconvoluted using ProteomeDiscoverer software, with advice from bioinformaticians at Glasgow Polyomics, and the relative change in abundance of proteins between samples for comparison was determined, with statistical significance assessed by analysis of variance between replicates using Minitab. Any value with a Mascot score of greater than 70 was considered significant and those with Mascot scores below 20 were discounted.

| Amyloid beta (Aβ) (1-42) and (1-40) ELISAs
MC-Demy cultures were treated with mHep7 (1 ng/ml) and media collected at 26 and 28 DIV (equating to days 1 and 3 posttreatment, respectively). This CM and mHep7 eluate described above were tested using amyloid beta peptide 1-42 and 1-40 ELISAs (Thermofisher, KHB3441/ KHB3481) as per manufacturer's description.   from 5% in untreated control (p = .0017, <.0001, and .0001 for mHep6, 7, and 8, respectively), but treatment with HS-mHep1 resulted in a decreasing trend to 2.5% although this appears nonsignificant (Figure 1d, FIGURE 1 LS-mHep promote myelination and neurite outgrowth (a,b) High power images of demyelination adjacent to the lesion after injury and the subsequent myelination occurring after 5 days. Representative images of MC-Inj of a control cut (c) and after a single treatment with LS-mHep6-8 (e-g) showing a promotion of myelination adjacent to the lesion which is adjacent and just out of view in the images. In contrast, the HS-mHep1 showed a nonsignificant decrease in the levels of myelination (d). Quantification of the images are shown in h (p = .0017, p < .0001 and p = .0001 for mHep6, 7 and 8 respectively). Only mHep6 significantly promoted neurite density adjacent to the lesion when compared to control cultures (p = .0127) (i). Representative images of the lesion after cutting with a scalpel blade in MC-Inj (j), and after HS-mHep1 treatment showing no effect on neurite outgrowth (k). Significant neurite outgrowth across the lesion was seen after treatment with LS-mHep6-8 compared to control untreated cultures (p = .0172, p < .0001 and p = .0029 for mHep6, 7, and 8, respectively) (l-o). Representative image of lesion size is shown in (j-n) identified by dashed line. An increase in lesion width following treatment with HS-mHep1 (564.5 μm) was seen, compared to the control lesion (400 μm p = .0026). LS-mHep6-8 had no effect on lesion width (p). Statistical test used was one-way ANOVA with post hoc Dunnett's multi-comparison correction. Scale bar, 25 μm and 50 μm, error bars SEM, SMI31-red, PLP-green (n = 6; technical replicates = 3) g, quantification in panel h). Neurite density was analyzed adjacent to the injury site and a significant increase (p = .0127) was observed after treatment with LS-mHep6 compared to the untreated control, whereas the other mHeps1, 7, and 8 had no significant effect on neurite density ( Figure 1i). Figure 1j-n show representative images of neurites crossing the lesion and the lesion size in control and treated cultures. Quantification of the average number of neurites crossing the lesion is shown in Figure 1o. The LS-mHeps 6-8, all significantly promoted neurite outgrowth across the lesion compared to the untreated control (p = .0172, <.0001, and .0029 for mHep6, 7, and 8, respectively). The HS-mHep1 did not promote neurite outgrowth after treatment, with neurite numbers being similar to the untreated control (~3 neurites/field of view).

| Statistical analysis
Finally, the width of the lesion was quantified after mHep treatment ( Figure 1p). HS-mHep1 treatment resulted in a significant increase in lesion size (564 AE 27.5 μm, p = .0026) compared to the untreated control (400 AE 20.0 μm, Figure 1j,k, and p). In contrast, the LS-mHeps6-8, did not significantly affect the lesion size, with average lesion sizes being,

| LS-mHeps promote oligodendrocyte process extension
To determine the effect of the mHeps directly on OPCs, we treated

| LS-mHep treatment enhances CNS remyelination
Because the developmental myelination in MC-Dev was essentially unaffected by treatment with mHeps, we hypothesized that the injury environment produced by MC-Inj may be essential to their effectiveness in promoting myelination. We propose that similar effects would be seen in a different CNS injury environment. To address this, we

| Secreted chemokine/cytokine profile is altered after MC-Inj and MC-Demy
As a first step to identifying molecules involved in the effects of mHeps described above, we took a candidate approach, reasoning that chemokines and cytokines which bind heparin/HS could be implicated. To identify specific chemokines and cytokines released in CCM and DCM, we conducted a protein array of conditioned media (see methods for details of array). Figure 6b-d illustrates the changes in the secreted cytokines standardized to the UCM. Green and red circles indicate only these factors that had a >2-fold increase or < 0.8-fold decrease, respectively. Several factors were upregulated in both CCM and DCM including CXCL2, CXCL5, and CCL5 (average fold increase of 8.5/16.4, 4.9/6.4, and 3.5/6.6 for CCM and DCM, respectively). Striking differences in secreted cytokine profiles were also seen between the two injuries. Many immune-associated factors including CCL3, IL-1α, IL-1β, IL-6, and TNF-α were increased in DCM by an average fold change of 9.1, 6.3, 8.9, 3.3, and 4.5, respectively.
Interestingly, the same proteins showed no change in CCM, apart from IL-1α and IL-1β, for which a decrease was observed (0.4 and 0.6, FIGURE 4 LS-mHeps have significant effects on remyelination in MC-Demy. Myelinating cultures were allowed to mature until 24 DIV at which point they were demyelinated via overnight incubation with the anti-MOG specific antibody Z2 (100 ng/ml) and rabbit complement (100 μg/ml). Treatment occurred at 25 DIV (1 ng/ml) and then cultures were allowed to recover till 30 DIV, at which point they were immunolabeled with SMI31 (red, neurites) and AA3 (green, myelin).

| Mass spectrometry analysis of conditioned medium from MC-Inj and MC-Demy
As an unbiased approach to identify molecules involved in the effects of mHeps, we used affinity proteomics to explore directly bound proteins that might be mediating the biological responses. To assess which proteins present in the UCM, CCM, and DCM interact with LS-mHep7, we performed a protein pull-down using an mHep7 affinity column, followed by mass spectroscopy ( (Table 2c). There were nine proteins which demonstrated an elevated abundance uniquely in the CCM (Table 2a)  3.9 | A focus on amyloid beta (Aβ) 1-42 which is present in the DCM-mHep7 eluate   1 and 3 posttreatment). (e) Graph shows a statistically significant increase in Aβ 1-42 present in DCM 24 hr after mHep7 treatment (D1) compared to untreated control (p = .0003). However, by day 3 posttreatment (D3), there was a significant reduction in Aβ 1-42 present in the CM following mHep7 treatment (p = .0096). The same pattern was observed with Aβ 1-40 with treatment leading to an increase in concentration shortly after treatment at D1 (p = .0010), but by D3 there seems to be a significant decrease in the level of Aβ 1-40 following mHep treatment compared to untreated control (p = .0001). (g-k) Cultures were treated with Aβ 1-42 (1 μM) at 16, 19, and 21 DIV (co-treatments with mHep7 at 100 ng/ml). (g-i) Representative images of control, Aβ treated and Aβ + mHep7 treated cultures at 24 DIV, immunolabeled with SMI31-red and PLP-green. (j) Quantification of myelination over time showing a significant decrease in developmental myelination at 24 DIV after Aβ treatment (p = .0082) this is overcome by mHep7 co-treatment (p = .0321). Quantification of neurite density over time suggests no effect of either treatment on neurite density (data not shown). (k) LDH release cytotoxicity assay demonstrating no statistically significant effect on cell death at 24 DIV following Aβ treatment or mHep7 co-treatment. Scale bar, 100 μm, error bars SEM (ELISAs n = 3; technical replicates = 2, cell culture experiments n = 4; technical replicates = 3) To investigate the presence of these peptides and validate the TMT-LC/MS analysis, an ELISA for the 1-42 peptide and 1-40 peptide was carried out on the eluate from the mHep7 column ( Figure 8a,b). This demonstrated that there was a significant increase in concentration for both Aβ peptides in the DCM eluate compared to UCM. However, this increase was not observed in the CCM eluate, suggesting that the increased abundance for the APP hit in CCM was due to a different cleavage product. This suggests that the APP cleavage product in the CCM mHep eluate may be from the nonamyloidogenic pathway. Moreover, the sequence of peptide fragment identified in the mass spectrometry analysis is a 12 residue peptide corresponding to residues 439-450 of the precursor protein and represents the soluble product of the initial cleavage by either α-secretase (nonamyloidogenic pathway) or β-secretase (amyloidogenic pathway); therefore, the peptide can be from either pathway.
3.10 | mHep7 treatment modulates Aβ peptide concentration following demyelination The absence of any effect on neurite density following treatment along with the cytotoxicity assay demonstrated that this was not due to generic cell death (Figure 8k). This inhibitory effect could be rescued by co-treatment with mHep7 (p = .032  (Barnett & Linington, 2013;Hara et al., 2017;O'Shea, Burda, & Sofroniew, 2017;Williams, Piaton, & Lubetzki, 2007). Moreover, after an injury to the adult rat brain, there is an overall increase in the quantity of HSPG around the injury site, as well as an increase in mRNA for heparan 2-O-sulfotransferase (HS2ST) and subsequently the level of 2-O-sulfated HS (Properzi et al., 2008). Nevertheless, we found little evidence of reactivity induction following mHep treatment in scratch astrocyte assays or after Western blotting with GFAP (data not shown). In addition, changes in sulfation patterns of HSPGs have many effects on axon growth and guidance. For example, HS2ST and heparan 6-O-sulfotransferase (HS6ST) knockout mice have shown multiple navigational errors in their axons, probably due to disturbance of the guidance effects of slit proteins (Pratt, Conway, Tian, Price, & Mason, 2006), and genetic manipulations of syndecan expression in Drosophila showed similar guidance defects (Johnson et al., 2004). Although information is known about the role of HSPGs in the development of the CNS, their function in nerve injury is not yet fully understood (Murakami, Tanaka, Bando, & Yoshida, 2015). There have been some reports of upregulation of syndecan-1 and glypicans after injury (Leadbeater et al., 2006) and evidence that this promotes neurite outgrowth but the mechanism is not known. It is likely that the HS chains carried by these HSPGs could be targets for the LS-mHeps.
Our data suggest that the sulfation level or its position on the HS disaccharide is crucial in regulating cellular function. Using our panel of mimetics, we found that only the monosulfated forms at the 2-Oand N-sulfated positions promoted neurite outgrowth. This is in agreement with a study that used mutants of the Hst-2 gene thereby reducing 2-O-sulfation on the HS leading to axonal patterning defects (Kinnunen, Townsend, & Turnbull, 2004). These experiments suggest that the 2-O-sulfate moiety is involved in neurite outgrowth and pathfinding. However, LS-mHep8 which lacks the 2-O-and 6-O-sulfate also promotes neurite outgrowth, implying that the reduced sulfation level of these mHeps may be contributing to the observed outgrowth.
Others have shown that applying HS with different sulfation modifications disrupts axons guidance in the Xenopus visual system, with 2-O-and 6-O-sulfated HS having the most marked effects (Irie, Yates, Turnbull, & Holt, 2002). Therefore, it has been postulated that there is a sulfation code that regulates axon guidance (Holt & Dickson, 2005). The LS-mHeps could be directly affecting neurite outgrowth, and astrocytosis through artificially mimicking their interaction with the growing neurite (Lander, Fujii, Gospodarowicz, & Reichardt, 1982), or indirectly by altering surrounding cellular behavior, creating a permissive environment for outgrowth and re/myelination.

| HS role in re/myelination
We originally considered that mHeps might regulate growth factors which act in concert to drive efficient myelination of OPCs. However, there was no effect of their treatment in developmental myelination occurring only when cultures were injured. We hypothesise this is because endogenous HS are sufficient for developmental myelination but during injury, there are dramatic changes in the extracellular environment suggesting that the LS-mHeps may elicit their effects through interacting with secreted factors present in this aberrant injury environment.

| Identification of mHep binding-proteins in CM from MC-Inj and MC-Demy
To examine the molecular basis for the pro-repair effects in postinjury myelinating cultures we aimed to identify heparin-binding proteins in two ways. First, we examined chemokine/cytokine candidates, and secondly conducted TMT-LC/MS. The chemokine/ cytokine array illustrated major differences in the secretome between the MC-Demy and MC-Inj. In the former immune-mediated MC-Demy, inflammatory factors in DCM were more prominent than in CCM. This suggests that these two different culture conditions affect remyelination by distinct mechanisms. For example, IL1α, TNF-α, and C1q (found in DCM, see Figure 6) induce the neurotoxic A1 astrocyte phenotype (Liddelow et al., 2017) suggesting that in MC-Demy this astrocyte phenotype may be affecting remyelination, as previously reported (Nash et al., 2011). Thus, mHeps may be removing or inhibiting A1 astrocyte inducing factors, therefore promoting remyelination.
Our initial MS analysis was nonquantitative but strongly indicated pull down heparin-binding proteins as the most significant group proteins (GO enrichment). GO term analysis (data not shown) confirmed CCM contained many factors involved in neurite outgrowth, guidance migration, and changes in astrocyte development. In contrast, DCM contained factors that were related to chemokine signaling and  Candidates that are secreted and of interest in these cultures are highlighted in bold and reflect those indicated in Figure 7b 4.4 | Factors found in CCM A2M, a broad spectrum proteinase inhibitor and a carrier of growth factors was increased in CCM. It has neuromodulatory activities (Wolf & Gonias, 1994) and has been demonstrated as a marker of neuronal injury and associated with preclinical AD (Varma et al., 2017). Apolipoprotein D (ApoD) is a secreted glycoprotein with many roles within lipid transport, detected in neurons, astrocytes, and oligodendrocytes (Ong et al., 1999). It has been associated with neurological disorders (multiple sclerosis [MS] and AD), other inflammatory diseases of the CNS and nerve injury (Li et al., 2015;Reindl et al., 2001). Interestingly levels were not above threshold in MC-Demy, suggesting a nerve damage component to its upregulated expression.
Clusterin (CLU) a small heat shock protein that can act as a molecular chaperone protein was also upregulated. CLU like ApoD, has been implicated in ameliorating oxidative stress in neurodegenerative diseases and may be involved in the death of damaged neurons (Törnqvist, Liu, Aldskogius, Holst, & Svensson, 1996). It has been iden- thought to act as a carrier of several proteins across the BBB and CSF barrier including amyloid-β (Aβ) (Ghiso et al., 1993;Zlokovic et al., 1996). It has been suggested that CLU directly interacts with Aβ, thereby regulating its clearance from the brain (Bell et al., 2007). Amyloid precursor protein (APP), which was also secreted in CCM and DCM, is known to be upregulated during axonal injury in MS (Ferguson, Matyszak, Esiri, & Perry, 1997), and CLU may clear it and thereby prevent aggregation of Aβ. It was also interesting that RAGE was upregulated in the cytokine array as it is thought to play a role in the peripheral re-entry of Aβ into the brain (Deane et al., 2003). Table 2b shows there were more candidates pulled down exclusively in DCM compared to CCM. Several have already been implicated in CNS injury, for example tenascin C (TnC), a glycoprotein synthesized by astrocytes and secreted into the ECM. Increased expression of TnC has been implicated after demyelination in vivo (Zhao, Fancy, Franklin, & ffrench-Constant, 2009) and been shown to inhibit OPC differentiation in vitro both directly and indirectly through astrocytes (Czopka, von Holst, ffrench-Constant, & Faissner, 2010;Nash et al., 2011). Moreover, knockout of TnC resulted in a favorable outcome on ADs pathology in vivo (Xie et al., 2013). Another ECM glycoprotein thrombospondin 1 (TSP-1) also appeared to have increased abundance in DCM. TSP-1 interacts with Neuroligin 1 to accelerate synaptogenesis of hippocampal neurons (Xu, Xiao, & Xia, 2010) but also reported to promote OPC migration (Scott-Drew & ffrench-Constant, 1997).

| Factors found in DCM
Gelsolin, an actin regulatory factor was identified in DCM. Mice lacking gelsolin, display a delayed remyelination after PNS crush injuries, presumed due to gelsolin recruiting macrophages to the injury site (Gonçalves et al., 2010). Moreover, gelsolin knockout mice had wrapping defects in the CNS, (Zuchero et al., 2015). This implies gelsolin might have a dual role after injury, firstly through the recruitment of immune cells to clear debris and secondly facilitating oligodendrocyte axon ensheathment.

| Candidate factors identified in both DCM and CCM
In Table 2c, it can be seen that a few candidates were upregulated in both CCM and DCM and therefore may relate to promoting myelination. For example, growth hormone releasing hormone receptor has been detected in the CNS, and its ligand activity promotes the secretion of insulin-like growth factor (IGF-1, Zhao et al., 2008), a known mitogen for OPCs and an important regulator of brain development, maintenance and neurogenesis (Aberg, 2010). Moreover, APP was identified in both CMs. HS has been shown to interact with Aβ peptides and thus been implicated in facilitating Aβ cytotoxicity and accelerating amyloid fibril formation (Castillo, Ngo, Cummings, Wight, & Snow, 1997;Sandwall et al., 2010). Additionally, infusion with Aβ and HS into rat brains resulted in increased amyloidosis compared to Aβ alone (Snow, Sekiguchi, Nochlin, Kalaria, & Kimata, 1994). This suggests that endogenous HS somewhat aids Aβ aggregation pathology.
Aβ has been shown to be secreted in excess following traumatic injury (Gentleman, Nash, Sweeting, Graham, & Roberts, 1993) and MS lesions (Pajoohesh-Ganji et al., 2014). Although not understood in these pathological conditions Aβ has been reported to have negative correlation with functional outcome and induce microglia activation, inflammation, and neuronal cell death (Matsuoka et al., 2001).
Due to the detection of Aβ in MC-Inj and MC-Demy and its known role in CNS pathology we decided to focus more on its function and interaction with mHeps in these cultures. Since APP has numerous different fragment peptides, it was important to determine the peptides identified in the TMT/LC-MS. ELISAs of the mHep7 eluates and CMs suggests that after demyelination in MC-Demy Aβ peptides (1-40/1-42) are secreted, and that the degradation of these peptides is modulated directly by mHep treatment. Previously, it has been demonstrated that treatment with Aβ 1-42 inhibits OPC differentiation (Horiuchi et al., 2012), in this study, we developed this further showing that the effect was present at the level of myelination and could be rescued through mHep co-treatment. However, it appears that the APP detected in the CCM mHep7 eluate TMT-LC analysis was not Aβ 1-40 or 1-42 peptides, suggesting it was the P3 peptide from the nonamyloidgenic pathway. This peptide is the equivalent of Aβ 17-40/42, although this fragment does not assemble into soluble oligomers, it does possess cellular toxic properties (Dulin et al., 2008;Wei, Norton, Wang, & Kusiak, 2002). Hence the P3 peptide could be a mHep7 modulated negative injury factor secreted after MC-Inj. As Aβ peptides have been reported in both neurodegeneration and traumatic injury, with studies demonstrating the therapeutic benefits of targeting BACE-1 in AD (Scholefield et al., 2003). This Aβ peptides could be a valuable target for mHep7 and the action of sequestering or inhibiting aggregation, could promote neurite outgrowth and myelination in MC-Inj and MC-Demy.
In summary, the present study has demonstrated beneficial effects of LS-mHeps on repair in models of CNS damage specifically promoting neurite outgrowth and myelination by modifying the properties of secreted factors generated after injury. Our data has identified multiple protein candidates for mediating these effects and thus plausible underlying mechanisms. Moreover, this illustrates the complexity of mediating repair and highlights that therapeutics need to target many factors, as seen by LS-mHeps. Furthermore, we show Aβ peptides can play a role in demyelination. Finally, based on the protein hits and the relation of some of these proteins to other neurological disease such as AD, these novel compounds could also have therapeutic potential in other neurological disorders.