Comparative Analysis of the Expression of Chondroitin Sulfate Subtypes and Their Inhibitory Effect on Axonal Growth in the Embryonic, Adult, and Injured Rat Brains

Abstract

Background:

Chondroitin sulfate glycosaminoglycans (CS-GAGs) are the primary inhibitory GAGs for neuronal growth after central nervous system (CNS) injury. However, the inhibitory or permissive activity of CS-GAG subtypes is controversial and depends on the physiological needs of CNS tissues. In this study, we investigated the characteristics and effects of CS-GAGs on axonal growth, which was isolated from the brain cortices of normal rat embryo at E18, normal adult rat brain and injured adult rat brain.

Methods:

Isolated CS-GAGs from embryo, normal adult, and injured adult rat brains were used for analyzing their effect on attachment and axonal growth using modified spot assay with dorsal root ganglion (DRG) explants and cerebellar granule neurons (CGNs). CS-GAGs were separated using high performance liquid chromatography (HPLC), and the subtypes of CS-GAGs were analyzed.

Results:

CS-GAGs of all three groups inhibited CGN attachment and axonal growth of DRGs. However, CS-GAGs of normal adult rat brain exhibited higher inhibitory activity than those of the other groups in both assays. When subtypes of CS-GAGs were analyzed using HPLC, CS-A (4S) was the most abundant in all three groups and found in largest amount in normal adult rat brain. In contrast, unsulfated CS (CS0) and CS-C (6S) were more abundant by 3–4-folds in E18 group than in the two adult groups.

Conclusion:

When compared with the normal adult rat brain, injured rat brain showed relatively similar patterns to that of embryonic rat brain at E18 in the expression of CS subtypes and their inhibitory effect on axonal growth. This phenomenon could be due to differential expression of CS-GAGs subtypes causing decrease in the amount of CS-A and mature-type CS proteoglycan core proteins.

Introduction

Chondroitin sulfate proteoglycans (CSPGs) are the most abundant proteoglycans in the central nervous system (CNS). CSPGs are secreted into the extracellular matrix (ECM) and are responsible for various regulatory functions during development, adulthood, and injury [1]. During mice development, juvenile-type CSPGs such as neurocan, versican β, tenascin-C, and a link protein of HAPLN1/Crtl1 are transiently formed in CNS before birth [2]. A deficiency of these CSPGs causes embryonic or perinatal lethality [3, 4]. This phenomenon related with their activities to regulate the growth, differentiation, and migration of multipotent neural precursor cells in vitro and in vivo [5]. From 2 weeks after birth, most of the juvenile-type CSPGs are replaced by the mature type such as aggrecan, brevican, versican α, tenascin-R, and a link protein of HAPLN2/Bral1 [2, 6]. They are involved in forming the perineuronal nets (PNN) and play a role in restricting the experience-dependent plasticity of adult brain [7]. Mature-type CSPGs are also involved in the scar tissue formation after CNS injury and play a major barrier against neuronal regeneration [8, 9]. The expression of neurocan and NG2, known as representative inhibitory CSPGs, is up-regulated in an injured CNS at 1 week [10–12]. Thus, CSPGs in CNS tissue show differential expression patterns and function during the developmental process and upon injury.

The biological activities of CSPGs mentioned above also depend on the characteristics of glycosaminoglycans (GAGs) attached to core proteins. CS-GAGs are linear, unbranched polysaccharides comprising repeating disaccharide units of N-acetylgalactosamine (GalNAc) and glucuronic acid (GlcA). The negative charge on sulfate groups impacts many functions: (1) trapping water molecules to hydrate nearby cells and tissues, (2) trapping cations to maintain a physiological ion balance around cell surface, and (3) interacting with basic residues of proteins such as nerve growth factor (NGF) and fibroblast growth factor (FGF) to regulate their functions [13, 14].

Many studies have shown that CS-GAGs are also classified by the position of sulfate groups into CS-A (GalNAc at C4), CS-C (GalNAc at C6), CS-D (GlcA at C2 and GalNAc at C6), and CS-E (GalNAc at C4 and C6) [15]. Subtypes of CS-GAGs are found in the developing and growth permissive regions of the CNS at different levels and exert differential effects on the proliferation, migration, and differentiation of neural stem/progenitor cells. Dermatan sulfate DS-iB (IdoA at 2S and GalNAc at 4S), and CS-E subtypes promote FGF-2-mediated proliferation of neural stem/progenitor cells [16]. Knockdown of over-sulfation in cortical plate (CP) by in utero electroporation of shRNAs disruptes the migration of neurons from the subventricular zone (SVZ) to CP [17].

Numerous studies have shown that CS-GAGs are the primary inhibitors of axonal growth [18–20]. However, the inhibitory mechanism of CS-GAGs and the roles of CS-GAGs subtypes are not well-defined yet. In this study, we compared the characteristics of CS-GAGs and their inhibitory effect on the axonal growth among three experimental groups: (1) adult rat brain with a surgical injury at acute stage, (2) normal counterpart of the adult rat brain and (3) normal embryonic rat brain at E18 to understand details about the function and mechanism of action of CS-GAGs during development and upon injury.

Materials and methods

Experimental animals

The animal protocol used in this study was approved by the Inha University–Institutional Animal Care and Use Committee (INHA-IACUC) on their ethical procedures and scientific care (INHA-150325-353). Rats aged 8–9 weeks (250–300 g) were used to isolate normal (Nor group) and injured (Inj group) brain cortices (n = 34/group). To induce brain injury, rats were anesthetized by intraperitoneal injection (ip) of 40 mg/kg ketamine and 10 mg/kg xylazine hydrochloride. Heads were shaved and an incision was made into the middle of the scalp; the skulls were exposed. A dental drill was used to remove a rectangular piece of skull bilaterally at the middle of the bregma. A stab wound injury was made using #15 scalpel blades with a depth of 4 mm from the brain surface at 0.5 mm anterior of the bregma, and 0.5 mm, 2.0 mm and 3.5 mm to bilateral of midline. At 1-week post operation, animals were sacrificed by decapitation. Brain tissues of 1.5 mm thickness around the lesion of the injured rats or from the normal counterpart were harvested and stored at − 70 °C before use. To obtain embryonic cortices of rats, E18 fetuses (E18 group) were removed from pregnant female rats (n = 16) and placed in a sterile 100-mm dish containing a cold hank’s balanced saline solution (HBSS) without calcium, magnesium, and phenol red (14175095; Gibco, Grand Island, NY, USA). The fetal rat brains were isolated and cut along the midline. The hippocampus and cerebellum were removed, and cortices were stored at − 70 °C before use.

Extraction and purification of GAGs

Brain cortices obtained from the normal adult rats, injured adult rats, or E18 rat embryos were ground by homogenizer (PT3100; Polytron, Luzern, Switzerland) at 12,000 rpm for 3 min in HBSS. Extraction and purification of GAGs were performed as described by Silva [21] with slight modifications. Briefly, brain tissues were treated with 100 units/mL of DNase I (D5025; Sigma, St. Louis, MO, USA) and 50 μg/mL of proteinase K (V302B; Promega, Madison, WI, USA) at 37 °C for 24 h, and samples were incubated at 95 °C for 10 min to inactivate the enzymes. One volume of samples was mixed with four volumes of methanol/chloroform (2:1) and centrifuged at 10,000 × g for 30 min. The upper methanol layer was removed from the top of the sample, and the interphase GAG layer was harvested. After drying at room temperature, total GAGs were dissolved in phosphate-buffered saline (PBS) and retrieved by precipitation with four volumes of absolute ethanol. GAG subtypes were fractionated by increasing the volumes of absolute ethanol to the sample volume from 0.2 × to 2.0 × by 0.2 × intervals. For each extraction, the mixtures were allowed to equilibrate at 4 °C for 24 h. Afterwards, the precipitate was collected by centrifugation at 3200 rpm for 30 min and completely dried at 60 °C. Another 0.2 volume of ethanol was added to the supernatants, and the procedure was repeated. Precipitates from 0.2 × to 0.8 × were pooled and used as heparan sulfate (HS) and dermatan sulfate (DS) fractions, and precipitates from 1.2 × to 1.6 × as chondroitin sulfate (CS) fraction. All precipitates were stored at − 70 °C before use.

Evaluation of GAG subtypes

The amount of purified GAG fractions was quantified by carbazole assay according to Cesaretti et al. [22]. Commercial CS-C purchased from Sigma was used to plot a standard curve from 0.1 to 10 mg. A serial dilution of the standard or GAG fractions of 50 μL were placed in a 96-well plate, and sodium tetraborate in sulfuric acid was added to a final concentration of 25 mM. Samples were heated for 10 min at 100 °C in an oven, and 50 μL of 0.125% carbazole in absolute ethanol was added. After heating at 100 °C for another 10 min and cooling at room temperature for 15 min, samples were subjected to colorimetric analysis at 550 nm using a microplate reader (Powerwave X; BioTek, Winooski, VT, USA). GAG fractions were also separated on a 0.5% agarose-gel containing 0.04 M barium acetate/1,2-diaminopropane buffer (pH 5.8) [23]. Gel electrophoresis was performed in 0.05 M 1,2-diaminopropane in acetic acid at pH 9.0 for 30 min at 100 V. Then, the agarose gel socked in 0.1% of cetyltrimethylammonium bromide solution for 24 h, dried, and stained with 0.2% toluidine blue in an ethanol:water:acetic acid buffer (50:49:1) for 1 h. Finally, the gel was decolorized using the same solution without toluidine blue.

Cell attachment assay using cerebellar granule neurons (CGNs)

Glass coverslips were first coated with 50 μL of nitrocellulose solution prepared by dissolving 4 cm² nitrocellulose membrane (T20351; Pall Corporation) in 10 mL methanol and coated again with 0.1 mg/mL poly-l-lysine (PLL) solution (p4832; Sigma). One microliter of each GAG solution containing GAGs (1 mg/mL), fluorescein isothiocyanate (FITC, 5 μg/mL; Sigma) and laminin (5 μg/mL; Invitrogen) in HBSS was spotted on coverslips. The laminin solution without GAGs was used as a control. The spots were allowed to air dry completely and humidified in a cell culture incubator for 3 h, and dried again to reduce the formation of the dense rim. CGNs were obtained from 5-day-old rats as described by Kramer and Minichiello [24]. Briefly, the cerebellum of rat brain was harvested and placed in HBSS. A single cell suspension was obtained by triturating the tissue with a 1000 μL pipette tip. Cells were harvested by centrifugation at 210 × g for 5 min and resuspended in neurobasal medium (Gibco) containing B-27 supplement (17504044; Invitrogen, Grand Island, NY, USA), 2 mM glutamine (35050061; GlutaMAX™; Gibco), 6 g/L d-glucose, 1 M KCl, and 0.1% penicillin–streptomysin (Gibco). Cells were plated at 1.9 × 10⁵ cells/well in a 12 well plate and incubated at 37 °C under 5% CO₂. After 24 h, cells were fixed with 4% paraformaldehyde and mounted using VECTASHILD mounting medium with 4′,6-diamidino-2-phenylindole (H1200; Vectorlab, Burlingame, CA, USA). The total number of cells in each spot was counted for analysis.

Axonal growth assay using dorsal root ganglion (DRG) explants

For inhibitory guidance cue assay, DRGs were obtained from E15 rat embryos and cultured in neurobasal medium containing 50 ng/mL NGF (R&D Systems). DRG explants in culture media were pipetted out and placed on coverslips in 12-well plates and incubated for 24 h, DRG cultures were treated with GAGs from each experimental group prepared in HBSS at final concentrations of 0.01–5 μg/mL. After 9 h, explants were immunostained for βIII-tubulin. The explants were observed under a fluorescence microscope, and images were analyzed by Neurite-J software (http://rsbweb.nih.gov/ij/index.html) with a modification of Sholl’s method [25] that provides a semi-automatic tool for the quantification of neurite outgrowth in explant cultures. The analysis was performed as previously described by Torres-Espín et al. [26]. Briefly, original fluorescent images for βIII-tubulin were transformed to a gray scale 8-bit images, and the explant at the center was marked automatically or manually. The threshold of fluorescent intensity was selected by a ROI selector, and noises were eliminated. The plug-in algorithm quantified the number of axons that crossed every concentric circle with a 25-μm interval. Mathematical analysis of the intersection profile measured the maximum number of neurites (Nmax) in a concentric circle, the distance at Nmax, and the total area occupied by axons (axon area) in each group.

For modified spot assay, 2-well chamber slides (177380; Lab-Tek, Rochester, NY, USA) were coated with 0.1 mg/mL PLL solution and 50 μg/mL laminin. Each of these was spotted with either 0.2 μL each of 5 μg/mL laminin alone or those with three different GAGs at 1 mg/mL in HBSS at four cardinal positions, 500 μm away from the center. In one experiment, purified GAGs from three experimental groups were used, and in another experiment, commercial GAG subtypes of CS-A (C9819; Sigma), CS-C (C4384; Sigma) and CS-E (CSR-NACS-E2[SQC]3; Cosmo bio, Tokyo, Japan) were used. The spots were allowed to air dry completely and humidified in a cell culture incubator for 3 h and dried again. A DRG explant was placed at the center of the spots in a chamber slide and cultured in neurobasal medium containing 50 ng/mL NGF. After 3–4 days, explants were immunostained for βIII-tubulin as below.

Immunocytochemistry for βIII-tubulin

DRG explants were fixed with 4% paraformaldehyde at room temperature for 30 min. Explants were rinsed with PBS and pre-incubated for 1 h in blocking buffer (2% normal donkey serum and 0.1% Triton-X 100 in PBS). Explants were then incubated with rabbit anti-βIII-tubulin antibody (T2200; Sigma) and subsequently with donkey anti-rabbit antibody conjugated with rhodamine (Jackson ImmunoResearch, West Grove, PA, USA) both in the blocking buffer overnight at 4 °C. After three rinses with PBS, explants were stained with Alexa Fluor^® 488 Phalloidin (A12379; Life Technologies, Carlsbad, CA, USA) and mounted.

Analysis of disaccharide GAGs using high-performance liquid chromatography (HPLC)

Purified GAGs and standards (Iduron, Macclesfield, UK) were labeled with a reducing agent, 2-aminoacridone (AMAC). HPLC separation of GAGs was performed on a Waters Spherisorb^® 5 μm ODS2 column (PSS831915; Waters, Milford, MA, USA). For the elution, 50 mM NaCl (pH 5.6) (Eluent A) and increasing concentrations of acetonitrile (Eluent B) at a flow rate of 1.0 μL/min were used. The gradient program of eluent B was as follows: 5% for 0–15 min, 5–23% for 15–45 min, 23–50% for 45–65 min, 50–5% for 65–66 min, and 5% for 66–75 min. The eluent was monitored using the on-line FD detector at excitation λ = 428 nm and emission λ = 525 nm.

Western blot analysis

Four brain cortices isolated from the three experimental groups were homogenized in RIPA buffer (Sigma) containing protease inhibitor cocktail (Sigma) and untreated or treated with chondroitinase ABC (ChABC) (C2905; Sigma) at 37 °C for 3 h. A total of 10–30-μg proteins was fractionated on a sodium dodecyl sulfate polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane (1620177; Bio-rad, Hercules, CA, USA). Membranes were blocked in 5% non-fat milk (Bio-rad) in tris-buffered saline with 0.1% Tween 20 (TBST) for 1 h and incubated overnight at 4 °C with mouse antibodies against neurocan (1:500, 1F6; DSHB, Iowa City, IA, USA), brevican (1:200, MABN491; Merck, Darmstadt, Germany), aggrecan (1:200, Cat-301; Merck), phosphacan (1:500, 3F8; DSHB), and β-actin (1:1000, SC-47778; Santa Cruz, Dallas, TX, USA) diluted in the blocking solution. Membranes were washed with TBST and incubated with horseradish peroxidase-conjugated goat anti-mouse antibody (1:10,000, 31,430; Pierce, Rockford, IL, USA). Finally, membranes were incubated in an enhanced peroxidase solution (EBP-1073; ELPIS-Biotech, Daejeon, Korea) for the detection of specific bands. The signal intensity of immunoreactive bands was quantified using Image J program.

Statistical analysis

Statistical analysis was carried out with Prism (GraphPad Software). For the comparison of multiple groups, one-way analysis of variance was used with Tukey’s test and least significance difference test (*p < 0.05 and **p < 0.01). Graphical data are presented as the mean ± standard deviation (SD).

Results

Isolation and purification of GAGs

The amount of total GAGs isolated from brain cortices was 3.91 ± 0.12 mg/g, 2.91 ± 0.32 mg/g, and 3.13 ± 0.15 mg/g (all tissues) in E18, Nor, and Inj groups, respectively. GAGs from the three experimental groups were purified by ethanol fractionation from 0.2 × to 2 × volumes with 0.2 × intervals, and the amount in each fraction was quantified through carbazole assay. The representative data of Nor group showed the peak value of 0.22 ± 0.077 mg/g tissues at 1.2 × fraction (Fig. 1A). The other two groups also showed similar elution profiles. GAG fractions from Nor group were analyzed by 0.5% agarose gel electrophoresis with some factions pooled together (Fig. 1B). The 0.2–0.4 × fractions showed small amount of GAGs corresponding to HS-GAGs, and the 0.6–0.8 × fractions contained large amount of both HS-GAGs and DS-GAGs. The 1.0 × fraction showed trace amount of CS-GAGs, and the 1.2 × fraction showed the majority of CS-GAGs. The 1.2–1.6 × fractions were pooled as CS-GAGs and 0.2–0.8 × fractions as HS/DS-GAGs, and their amounts were determined in the three experimental groups. As shown in Fig. 1C, the amount of CS-GAGs was more in E18 group (0.16 ± 0.001 mg/g tissue) than in Nor group (0.109 ± 0.021 mg/g), but with no statistical significance, and it was the least in Inj group (0.066 ± 0.016 mg/g, p < 0.05 vs. E18 group). The amount of HD/DS-GAGs was significantly more in E18 group (0.152 ± 0.029 mg/g) than in Nor (0.046 ± 0.002 mg/g) and Inj (0.083 ± 0.026 mg/g) groups, but the statistical significance was shown only between E18 and Nor groups (p < 0.05) (Fig. 1D). When the ratio of CS-GAGs and HS/DS-GAGs was calculated, it was significantly larger in Nor group (2.347 ± 0.345 mg/g, p < 0.05 vs. E18 and Inj groups) than in E18 (1.067 ± 0.208 mg/g) or Inj (0.806 ± 0.063 mg/g) groups (Fig. 1E).

Fig. 1.

Isolation and purification of GAGs. A A representative profile showing the amount of GAGs purified from normal rat brain tissue (Nor group) by ethanol fractionation from 0.2 × to 2 × ethanol to the initial sample volume. The amount of GAGs was measured by carbazole assay. B A representative image of harvested fractions (0.2–0.4 × , 0.6–0.8 × , 1.0 × , and 1.2 ×) of Nor sample separated on a 0.5% agaroses gel. CS-GAGs purchased from Sigma (C4384) were used as a reference in the first lane. Arrows indicate expected positions of CS-GAGs, DS-GAGs, and HS-GAGs. C, D Histogram shows the amount of CS-GAGs (1.2–1.6 × fractions) and sum of HS-GAGs and DS-GAGs (HD/DS; 0.2–0.8 × fractions) in experimental groups, E18, Nor, and Inj, measured by carbazole assay. E The ratio in the amount of CS-GAGs (CS) to that of HS/DS-GAGs (HD/DS) is presented. All data are mean values with SD from two independent experiments (n = 2). Statistical analysis was performed among the experimental groups. *p < 0.05

Effect of total and purified GAGs on CGNs attachment

Total GAGs or purified CS-GAGs and HS/DS-GAGs obtained from the three experimental groups were coated on the coverslips, and CGNs were cultured on them for 24 h. The attachment of cells was observed by fluorescence images (Fig. 2A), and quantitative data were obtained by the average number of attached cells in a microscopic field (Fig. 2B–D). When compared with the control group (385 ± 26.2), total GAGs from the experimental groups showed slightly reduced CGN attachment with the Nor group having the least number of cells (222.3 ± 32.3, p < 0.05 vs. the control group) (Fig. 2B). In contrast, CS-GAGs showed a significant reduction of CGN attachment in all groups (Fig. 2C). The numbers of attached CGNs were only 58.0 ± 21.9, 1.7 ± 1.5, and 11.0 ± 3.6 in E18, Nor, and Inj groups, respectively, whereas it was 278.7 ± 91.4 in the control group (p < 0.01 vs. experimental groups). HS/DS-GAGs also showed slight reduction in CGN attachment in the experimental groups but with no statistical significance (Fig. 2D). The numbers of attached CGNs were 353.5 ± 89.7 in the control, 171.5 ± 102.7 in E18, 178.3 ± 88.5 in Nor, and 289.6 ± 270.0 in Inj groups. These results suggest that CS-GAGs not HS/DS-GAGs have an inhibitory effect on the attachment of CGNs.

Fig. 2.

Cell attachment assay of CGNs on total GAGs, CS-GAGs (CS), and HS/DS-GAGs (HS/DS) isolated from the experimental groups. CGNs (1.5 × 10⁵ cells/well) were cultured for 24 h on 5 μg/mL laminin alone (Control) or on laminin (5 μg/mL) containing 1 mg/mL of total GAGs, CS-GAGs, or HS/DS-GAGs isolated from brain cortices of E18 embryo (E18), normal adult (Nor), and injured adult (Inj) rats. FITC was mixed with each ECM coating to indicate the spot area (green), and DAPI staining was used to recognize CGNs (blue). A Representative images of cell attachment assay. Scale bar = 100 μm. B–D The number of attached CGNs in each experimental group was counted from DAPI images and presented in the histogram. Data are mean values with SD from three independent experiments (n = 3 or n = 4 in HS/DS-GAG). Statistical analysis was performed among the experimental groups. *p < 0.05 and **p < 0.01

Inhibitory effects of soluble CS-GAGs on axonal growths of DRG neurons

CS-GAGs from the experimental groups were used to treat the culture of DRG explants at 0.5 μg/mL for 9 h under axonal growth condition. Axons formed were immunostained for βIII-tubulin, and the number of axons crossing concentric circles with 25-μm intervals from the explant was quantified (Fig. 3A). The fluorescence images of βIII-tubulin showed that DRG axons grew significantly for 9 h in the control group (Fig. 3B). When compared with control 9 h group, the axonal growth was similar or slightly better in E18 group, whereas it was significantly reduced in Nor and Inj groups. In the quantitative analysis, the experimental groups showed clearly different patterns in the number of axonal intersections at each concentric circle along the distance (Fig. 3C). When compared with control 0 h group, the maximum number (Nmax) of axonal intersections at concentric circles increased but with no statistical significances in control 9 h group. When compared with control 9 h group, it showed significant decrease in E18 and Nor groups (252.7 ± 27.1 and 174.7 ± 29.6, respectively, p < 0.01 vs. control 9 h for both groups) but not in Inj group (241.0 ± 20.2) (Fig. 3D). The distance at Nmax of control 9 h (991.7 ± 57.7) group was also slightly higher than that in control 0 h (725.0 ± 109.0), and it was decreased significantly only in Nor group (716.7 ± 76.4) (p < 0.05 vs. control 9 h group) (Fig. 3E). The total axonal area of each group showed similar results to those of Nmax and the distance at Nmax (484.3 ± 66.3 in control 9 h, 412.3 ± 31.4 in E18, and 253.4 ± 88.6 in Nor groups) (Fig. 3F). These results show that the inhibitory effect on axonal growth of CS-GAGs from the rat brain tissue varies depending on the developmental stages and upon brain injury, and it was observed only in normal adult brain (Nor group).

Fig. 3.

Axonal growth assay of DRG explants on CS-GAGs of the experimental groups. A A DRG explant each was cultured on a PLL/laminin-coated well of 12-well plate, containing neurobasal medium and 50 ng/mL of NGF, for 24 h. Samples were then untreated (Control) or treated for 9 h with 0.5 μg/mL CS-GAGs of E18, Nor or Inj group to examine their inhibitory effect on axonal growth. The length and fasciculation of axons grown from DRG explants were analyzed by the number of intersections between axons and virtual circles with 25-μm intervals around the explants. B The DRG explants were stained with βIII-tubulin to visualize axons in each group. Control 0 h indicates a group before the treatment with CS-GAGs and Control 9 h indicates a mock-treated group. The number of intersections was counted starting at 600 μm from the DRG explant that shows the maximum intersection number (Nmax) in Control 0 h group. Scale bar = 500 μm. C The histogram shows the number of axonal intersections (Y axis) along the distance of virtual circles (X axis). D–F Histograms show the maximum number of intersections (Nmax), the distance at Nmax and the area occupied by axons in each group. Data are presented by mean values with SD from three independent experiments (n = 3). Statistical analyses were performed among all groups except Control 0 h group. *p < 0.05 and **p < 0.01

Inhibitory effects of CS-GAG coatings on axonal growths of DRG neurons

The axonal inhibitory effect of CS-GAGs from the experimental groups was investigated by spot assay. The control and experimental CS-GAGs were spotted at four cardinal positions from a DRG explant at the center (Fig. 4A). Axonal growth was induced for 3–4 days before immunostaining for βIII-tubulin. As shown in Fig. 4B, axonal growth occurred normally at the spot area in the control group, whereas it was inhibited at the boundary of and went around the spot forming thick axon bundles (fasciculation) in the experimental groups. Quantitative analysis confirmed the inhibitory effect of CS-GAGs from the experimental groups (59.0 ± 9.6% in E18, 35.7 ± 9.7% in Nor, and 48.1 ± 10.0% in Inj) with the most significant effect shown in Nor group (p < 0.05 vs. E18 group) (Fig. 4C).

Fig. 4.

Spot assay with DRG explants on CS-GAGs. A Laminin alone (C) and CS-GAGs (1 mg/mL each) of E18 (E), Nor (N), and Inj (I) groups mixed with 50 μg/mL laminin were spotted at four cardinal points on a chamber slide pre-coated with PLL/laminin. A DRG explant was placed at the center and cultured in neurobasal medium with 50 ng/mL of NGF for axonal growth for 3–4 days. The representative image shows a DRG explant with axonal growth after βIII-tubulin staining and position of four cardinal ECM spots. Scale bar = 500 μm. B High magnification images show the extent of axonal growth in four experimental groups. The spot area is marked with a dotted circle in the Control group as an example. Scale bar = 100 μm. C The amount of penetrated axons into the spots in each group was measured by fluorescent intensity. Data are presented by mean intensities with SD from three independent experiments (n = 3). Statistical analyses were performed among the experimental groups. *p < 0.05 and **p < 0.01

Analysis of CS-GAG subtypes in the experimental groups

CS-GAG subtypes in E18, Nor and Inj groups were analyzed using HPLC in conjunction with commercial standards (Fig. 5A). The amounts of CS-GAG subtypes were determined in each group by the area of peaks for 0S (unsulfated CS), 4S (CS-A), 6S (CS-C) and 4,6S (CS-E). As shown in Fig. 5B, the amount of unsulfated CS and CS-C subtypes was relatively larger in E18 group (Measured values were 3,312,712 in E18, 1,095,288 in Nor, and 871,472 in Inj groups and 2,874,769 in E18, 414,523 in Nor, and 487,860 in Inj groups, respectively), whereas that of CS-A subtype was relatively larger in adult groups (5,381,904 in E18, 8,021,999 in Nor, and 7,584,627 in Inj groups). The amount of CS-E was relatively less in all groups, with Nor group showing slightly larger amount than that of other groups (180,547 in E18, 476,791 in Nor, and 267,587 in Inj groups). This result suggests that CS-GAG subtypes can be classified into the juvenile (unsulfated CS and CS-C) and mature (CS-A and CS-E) subtypes like the case of CSPGs [6, 27]. When the ratio in the amount of mature types to that of juvenile type was calculated, it was significantly higher in Nor group than in E18 group, which decreased clearly in the Inj group (1.872 in E18, 19.352 in Nor, and 10.95 in Inj of CS-A/CS-C ratio, and 0.63 in E18, 1.15 in Nor, and 0.548 in Inj of CS-E/CS-C ratio) (Fig. 5C).

Fig. 5.

Analysis of CS-GAG subtypes purified from E18, Nor, and Inj groups. CS-GAGs of each group were digested with chondrotinase ABC and tagged with 2-AMAC. Tagged GAGs were separated by HPLC, and elution profile was monitored at 428 nm. A Representative elution profiles of CS disaccharide purchased from Iduron (Standard) and CS-GAGs of the three experimental groups. 0S; GlcA-β1, 3-GalNAc, 6S; GlcA-β1, 3-GalNAc (6S), 4S; GlcA-β1, 3-GalNAc (4S), 2, 6S; GlcA (2S)-β1, 3-GalNAc (6S), 4, 6S; GlcA-β1, 3-GalNAc (4S, 6S), and 2, 4, 6S; GlcA (2S)-β1, 3-GalNAc (4, 6S). B The peak area was measured to quantify CS-GAG subtypes of non-sulfated CS (0S), CS-A (4S), CS-C (6S), and CS-E (4,6S) in the experimental groups. C The ratio of CS-A (4S) or CS-E (4,6S) to CS-C (6S) is presented in the histogram

CS-A was more inhibitory on axonal growth than CS-C and CS-E

To compare the inhibitory effects of CS-GAG subtypes on axonal growth directly, modified spot assay was performed using commercially available CS-A, CS-C, and CS-E with more than 80% purity (Fig. 6A). As shown in the fluorescence images of βIII-tubulin, all CS-GAG subtypes inhibited axonal growth but CS-A (13.5 ± 7.8%) showed the strongest inhibitory effect (Fig. 6B). The quantitative analysis of the fluorescence intensity within the spot area elicited the difference in the inhibitory effect of CS-A from that of CS-C (40.6 ± 10.7%) or CS-E (29.3 ± 9.9%, p < 0.05 vs. CS-C) (Fig. 6C). These results indicate that the mature subtype of CS-A has a stronger inhibitory activity on axonal growth than that of the juvenile subtypes.

Fig. 6.

Spot assay with DRG explants on CS-GAG subtypes. A DRG explants were cultured at the center of four cardinal spots of laminin or three commercially available CS-GAG subtypes (CS-A, CS-C, and CS-E) as described in Fig. 4. After 3–4 days, axons were stained for βIII-tubulin (red). Spot area is shown in green by FITC fluorescence as above. Scale bar = 500 μm. B High magnification images of βIII-tubulin show the extent of axonal growth in four experimental groups (top). The spot area is shown by FITC images (bottom) and marked with a dotted line in βIII-tubulin image of the Control group as an example. Scale bar = 100 μm. C The amount of penetrated axons into the spots was measured by fluorescent intensity of βIII-tubulin. Data are presented by mean intensities with SD from three independent experiments (n = 3). Statistical analyses were performed among the experimental groups. *p < 0.05 and **p < 0.01

The amount of CSPGs core proteins in 3 experimental groups

The inhibitory or permissive effect of CSPGs on axonal growth depends not only on the CS-GAG subtypes but also the types of core proteins [28]. Therefore, the amount of CSPG core proteins was examined in the experimental groups by western blot analysis with or without ChABC treatment. As shown in Fig. 7A, Neurocan was observed in large amount in all groups. Interestingly, Nor group showed only the C-terminal truncated form of neurocan at 140 kDa, whereas the other groups showed both the full length (250 kDa) and truncated forms. Similar amounts of phosphacan and aggrecan were observed in E18 and Inj groups, but a larger amount was observed in Nor group. In contrast, very less amount of brevican was observed in E18 group, but significant amounts were found in Nor and Inj groups. The quantitative data confirmed the statistical differences in the amount of each core protein among experimental groups as observed in the image analysis (Fig. 7B–E).

Fig. 7.

Changes in the expression levels of CSPG core proteins in rat brain cortices. A Protein extracts were isolated from rat brain cortices of E18, Nor, and Inj groups. Samples were untreated (−) or treated (+) with chondroitinase ABC before western blot analysis for neurocan, brevican, phosphacan, and aggrecan. β-actin was used as an internal control. B–E The amount of core proteins was quantified by relative density and compared among the experimental groups in histograms. Data are presented by mean values with SD from three independent experiments (n = 3). Statistical analyses were performed among the experimental groups. *p < 0.05 and **p < 0.01

Discussion

This study investigated the in vitro inhibitory effects of CS-GAG subtypes in the brain on axonal growth by comparing the three experimental groups: the normal adult, injured adult, and normal E18 embryo rats. Isolated CS-GAGs from the groups showed inhibitory effect on CGN attachment and axonal growth of DRG neurons, but those from the normal adult group had the strongest effect. Subtype analysis showed that relatively more amounts of CS-A and CS-E were observed in the two adult groups, whereas significantly more amounts of unsulfated CS and CS-C were observed in E18 group than the other groups. This result suggests that CS-A and CS-E can be mature types and CS-C a juvenile type of CS-GAGs. Among the mature-type CS-GAGs, significantly more amount of CS-A was observed than CS-E, and it showed the strongest inhibitory effect on axonal growth of DRG neurons suggesting its important role. Unexpectedly, however, the amount of CS-A did not increase at all but rather decreased slightly after brain injury in this study. Therefore, the effects of CS-GAG subtypes on axonal growth appear to be very complicated and depend on the interplay between different GAG types (HS and DS) and CSPG core proteins during normal development and pathophysiological process after injury. In this study, the amount of CS-E, another mature-type CS-GAG, in rat brain was very less in all experimental groups but relatively higher in Nor group. CS-E is also known as a strong inhibitory molecule of axonal growth [20, 29]. CS-E is synthesized from CS-A via the addition of sulfate groups by GalNAc 4-sulfate 6-O-sulfotransferase. Accordingly, the amount of CS-E and its role in CNS might depend on the amount of CS-A, which suggests the importance of CS-A in the inhibition of axonal growth. The inhibitory mechanism of mature-type CS-GAGs on axonal growth is known to be mediated by CSPG receptors such as Ngr1, NgR3, and PTPσ [30–32]. The inhibition of axonal growth by CS-GAGs is closely related with the formation of perineural nets (PNNs) in adult brain and normal development of embryonic brain [33]. Negative charges of CS-GAGs are known to play important roles in PNN functions such as modulation of ionic homeostasis, synaptic stabilization and plasticity, and protection of neurons from oxidative stress and neurotoxins [33, 34]. However, its action mechanism is not well understood. We think the results of this study can provide helpful information to solve the question in the future.

According to a study by Gibert et al., the amount of adult types of 4S decreased from 91 to 25%, whereas that of the juvenile type of 6S increased from 4 to 50% at 4 week after brain injury in rats [29]. Their result is similar to ours showing the decrease of CS-A/CS-C ratio upon brain injury in adult rats, except for the extent of changes. However, there are other studies with opposite results, showing the increase of mature-type CS-GAGs after brain injury [19, 20]. The conflicting results from different studies could be caused by many factors such as disease or injury models, sampling times and methods, and assay protocols. We speculate that it is worth considering the sampling times and the changes in the pathophysiology through the acute, subacute, and chronic phases after injury as important contributing factors. In this study, brain samples were harvested at a subacute phase of 1 week after injury when glial limitans is not formed. At this time, reactive astrocytes highly proliferate and secrete various cytokines and CSPGs to induce inflammation astrogliosis [35]. In contrast, a large amount of glial limitans surrounds the wound area and blocks the leakage of blood–brain barrier (BBB) and leukocyte infiltration at 4 weeks [35, 36]. We speculate that the lesion epicenter could be in a more unstable and dynamic environment at the subacute phase. The results of this study suggest that CS-GAG subtypes might play different roles in the cytotoxic and reparative events and the possible role of CS-GAG subtypes in the cytotoxic or reparative events at this transition phase.

In the axonal growth assay of this study, the number of axonal intersections decreased by treatment with soluble CS-GAGs of all experimental groups. The number of axonal intersections indicates the extent of axonal fasciculations. During CNS development, axons form large bundles or fascicles, a process known as axonal fasciculation, to travel long distances and reach their targets [37–39]. Two possible mechanisms are known for axonal fasciculations. Axon–axon interactions via cell adhesion molecules such as L1 and NCAM can induce robust axonal bundling [40, 41]. Interactions between axons and ECM as guidance cues also form fasciculations [37, 38]. For example, neurocan is responsible for the transient formation of densely packed thalamocortical axons on E16 subplate during development [42]. In our spot assay, axon bundles were dissociated when passing across the boundary of CS-GAG spots, indicating defasciculation of axons. Accordingly, CS-GAGs alone or in conjunction with GSPG core proteins such as neurocan, as the case above, could be involved in the axon fasciculation. Decrease of axonal fasciculation was also observed in E18 group, which implicates that CS-GAGs in the embryonic brain can also have an inhibitory effect on axonal growth only at a lesser extent than that of the adult counterpart. We speculate that it might be a necessary event in guiding neuronal growth and the formation of developing brain tissues.

The compound findings of this study might be also related with the roles of HS-GAGs. Heparan sulfate proteoglycans (HSPGs) serve as CSPGs in the proliferation and differentiation of neurons during CNS development and glioma formation [43, 44]. Coles et al. [31] have suggested that axon outgrowth is determined by the CSPG/HSPG ratio. Many receptors involved in axonal growth are known to recognize both CSPGs and HSPGs [31, 32, 45, 46]. For example, Sema5A is a bifunctional guidance cue that induces axonal growth and is regulated by both CSPGs and HSPGs [46]. The competing effects of CS- and HS-GAGs on PTPσ oligomerization serve as a molecular switch to determine a go or stop for the axonal growth and cell motility [31]. CS-GAG is known to inhibit the formation of PTPσ dimers or oligomers and its function in axonal outgrowth, whereas HS-GAG is known to bind IgG1 domain of PTPσ and induce its multimerization. In this study, the amount of HS-GAGs was significantly higher in the embryonic brain than in the normal adult brain. Consequently, the ratio of CS-GAGs to HS-GAGs became much higher in the normal adult brain than in the embryonic brain although the amount of CS-GAGs decreased. In the case of injured brain, the amount of CS-GAGs decreased, whereas that of HS-GAGs increased slightly, resulting in a much decrease in the CS-GAGs to HS-GAGs ratio. This result suggests that the composition of GAGs after brain injury could be more permissive to axonal growth and neuronal regeneration at least at 1 week after injury.

The full length neurocan of 250 kDa, a representative inhibitor of glia scar, was observed in large amount at 1 week after brain injury in this study, which implicates a possibility for it to form physical barriers. However, the expression of neurocan was previously shown to gradually decrease again after 1 week [47], and we have observed similar results (data not shown). Therefore, neurocan might be expressed transiently until the subacute phase after brain injury. This study also showed that the amount of mature type CSPGs, aggrecan, and phosphacan decreased after injury. Harris et al. has also reported similar results and speculated that it could increase synaptic plasticity around PNN [48]. There are other studies with similar results showing a decrease in the amount of mature-type CSPGs after spinal cord injury [47, 49]. Therefore, it appears that the expression of inhibitory CSPG core proteins decrease after brain injury. It can also be speculated again that both GAGs and CSPG core proteins play roles together to exert permissive or inhibitory effect on axonal growth.

Axonal growth depends on the action of the growth cone, one of the sensory and motile organelles located at the tip of projecting axons [50]. Dynamic behavior of the growth cone is determined by the interaction with both adhesion molecules and soluble guidance cues. Adhesion molecules like integrins and CAMs form focal adhesions with ECM molecules such as laminin and collagen and deliver external or intercellular signals to maintain the adherent growth cone [51]. Chemotropic cues as a ligand determine “go” and “stop” of the growth cone by attractive or repulsive signals [52]. It is known that the expression of several repulsive guidance cues such as semaphorin and slit is induced in lesion epicenter after CNS injury to serve an inhibitory function on axonal growth [53, 54]. Attractive guidance cues such as NGF, NT-3, and BDNF are also expressed after CNS injury but they appear to be insufficient to guide axonal growth into the lesion epicenter. Fibrotic scar formed rapidly after CNS injury is another critical hurdle for axonal growth determining the injury size [55]. Many strategies have been investigated to inhibit the formation of fibrotic scar or remove it once formed. The suppression of fibrotic scar formation using pharmacological reagents reduces defect size and promotes neuronal regeneration in several CNS lesions [36, 56, 57]. Hydrogels containing NT3 and BDNF injected into the lesion center also induce axon regrowth across the compact astrocytic scar [58, 59]. Recently, human bone marrow-derived clonal mesenchymal stem cells (hcMSCs) reduced fibrotic scar in lesion center and protected neural tissues from fluid-filled cavity after spinal cord injury (SCI) in rats [60]. The therapeutic effect is known to depend on paracrine factors secreted from hcMSCs, but the exact mechanism is not clearly understood. Despite many promising studies as above, current therapeutic approaches are not effective to provide a clinical benefit to patients with CNS injury. Accordingly, novel approaches or combination therapies are required dealing with various controlling factors including CS-GAGs, adhesion molecules, growth factors, repulsive guidance cues, and fibrotic scar. The results of this study provide few findings about the composition of fibrotic scar, but their therapeutic implications are yet to be defined.

Compliance with ethical standards

Conflict of interest

Authors declare there is no conflict of interest with the manuscript.

Ethical statement

The animal studies were performed after receiving approval of the Institutional Animal Care and Use Committee (IACUC) of Inha University. (IACUC approval No. INHA-150325-353).