Chondroitin sulfate, a major component of the perineuronal net, elicits inward currents, cell depolarization, and calcium transients by acting on AMPA and kainate receptors of hippocampal neurons


  • Marcos Maroto,

    1. Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • José-Carlos Fernández-Morales,

    1. Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • Juan Fernando Padín,

    1. Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • José C. González,

    1. Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • Jesús M. Hernández-Guijo,

    1. Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • Eulalia Montell,

    1. Pre-Clinical R&D Area, Pharmascience Division, Bioibérica, Barcelona, Spain
    Search for more papers by this author
  • Josep Vergés,

    1. Pre-Clinical R&D Area, Pharmascience Division, Bioibérica, Barcelona, Spain
    Search for more papers by this author
  • Antonio M. G. de Diego,

    1. Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • Antonio G. García

    Corresponding author
    1. Departamento de Farmacología y Terapéutica, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    2. Instituto de Investigación Sanitaria, Servicio de Farmacología Clínica, Hospital Universitario de la Princesa. Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    • Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author

Address correspondence and reprint requests to Antonio G. García, Instituto Teófilo Hernando, Facultad de Medicina, Universidad Autónoma de Madrid, C/ Arzobispo Morcillo, 4, Madrid 28029, Spain. E-mail:


Chondroitin sulfate (CS) proteoglycans (CSPGs) are the most abundant PGs of the brain extracellular matrix (ECM). Free CS could be released during ECM degradation and exert physiological functions; thus, we aimed to investigate the effects of CS on voltage- and current-clamped rat embryo hippocampal neurons in primary cultures. We found that CS elicited a whole-cell Na+-dependent inward current (ICS) that produced drastic cell depolarization, and a cytosolic calcium transient ([Ca2+]c). Those effects were similar to those elicited by α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) and kainate, were completely blocked by NBQX and CNQX, were partially blocked by GYKI, and were unaffected by MK801 and D-APV. Furthermore, ICS and AMPA currents were similarly potentiated by cyclothiazide, a positive allosteric modulator of AMPA receptors. Because CSPGs have been attributed Ca2+ -dependent roles, such as neural network development, axon pathfinding, plasticity and regeneration after CNS injury, CS action after ECM degradation could be contributing to the mediation of these effects through its interaction with AMPA and kainate receptors.

Abbreviations used



chondroitinase ABC


chondroitin sulfate


CS proteoglycan


extracellular matrix


fetal bovine serum




long-term potentiation




perineural net

Chondroitin sulfate (CS) is a glycosaminoglycan (GAG) composed of disaccharide units formed by N-acetylgalactosamine and glucuronic acid whose carbons C4, C6, and C2 may be sulfated in either combination. It attaches to a core protein to form CS proteoglycan (CSPG), a major component of the brain extracellular matrix (ECM) and the perineural net (PNN) which is a specialized net-like matrix surrounding soma and proximal dendrites of a subpopulation of neurons. CSPGs are involved in neural development, axon pathfinding and guidance, plasticity (synaptic maturation) and regeneration failure after injury in the nervous system (Bartus et al. 2012; Kwok et al. 2012). It has been reported that CSPGs expression correlates with ECM maturation and with a post-natal critical period for plasticity. It seems that the formation of mature ECM with CSPGs restricts spine dynamics and motility through the β-1 integrin pathway (Orlando et al. 2012). Curiously, this integrin is linked to α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA)-receptor-mediated transmission, as β-1 integrin KO mice showed a drastic reduction in fEPSP at the CA3–CA1 synapse (Chan et al. 2006).

The molecular mechanisms underlying those effects are unknown. It has been suggested that CS-GAGs are involved in controlling the availability of Ca2+ and other ions to neurons because of its negative charge profile. In fact, removal of CS-GAGs by chondroitinase ABC (ChaseABC) in neocortex and hippocampal slice culture shows an increase in the rate of Ca2+ diffusion (Hrabetova et al. 2009). Furthermore, CS from CSPGs is able to evoke a mild elevation of the [Ca2+]c in dorsal root ganglion neurons (Snow et al. 1994). Recently, the disulfated CS-E (4,6S) has been shown to interact specifically with the receptor contactin-1 to stimulate neurite outgrowth (Mikami et al. 2009).

Bacterial enzyme ChaseABC catalyzes the cleavage of 1,3 glycosidic bonds of CS, among others GAGs. In vivo treatment with this enzyme restores functional plasticity in various models of CNS pathologies (Kwok et al. 2011). Furthermore, ChaseABC reduces long-term potentiation (LTP) and LTD at pyramidal cell synapses in the CA1 hippocampal area (Bukalo et al. 2001) and elevates the excitability of inhibitory basket interneurons in vitro (Dityatev et al. 2007), thus providing a possible mechanism by which enhanced GABAergic inhibition may affect LTP.

In this study, we explored whether CS could somehow modulate neurotransmission in neuron/glia hippocampal cell cocultures. We discovered that CS elicited an inward Na+-dependent current that caused cell depolarization and the elevation of the [Ca2+]c. All three effects were mainly mediated by glutamate receptors of the AMPA and kainate subtype. To our knowledge, ours is the first report demonstrating that CS, the major GAG forming part of PNNs, tightly mimics the function of the neurotransmitter glutamate on the subtype of AMPA/kainate receptors.

Materials and methods

Preparation of rat hippocampal neurons

All experiments were carried out in accordance with the code of ethics and guidelines established by European Community Directive (2010/63/EU) and Spanish legislation 1201/2005. All animals used in this study were provided by the local university's facilities where they were housed (EX 021-U). Pregnant Sprague–Dawley rats (Rattus norvegicus) were killed by decapitation, and male and female 18-day-old embryos were immediately removed by cesarean section. Hippocampi were rapidly dissected under a stereomicroscope using sterile, cold (4°C), and previously air-bubbled phosphate buffer solution of the following composition (in mM): 137 NaCl, 2.7 KCl, 11.6 NaH2PO4, 1.47 KH2PO4 (pH 7.4, adjusted with NaOH). Tissue was digested with 0.5 mg⁄mL papain and 0.25 mg⁄mL DNAase. Enzymes were dissolved in a Ca2+- and Mg2+-free phosphate buffer solution containing 1 mg⁄mL bovine serum albumin and 10 mM glucose at 37°C for 20 min. The papain solution was replaced with 5 mL of Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS). The digested tissue was then gently triturated by suction using a glass pipette flamed on the tip to avoid cellular damage. The cell suspension was centrifuged for 4 min at 120 g. The supernatant was removed and the cells were resuspended in 5 mL of Dulbecco's modified Eagle's medium and plated at a density of 100 000 cells⁄mL on 25- or 12-mm glass coverslip coated with MaxGel™ (Sigma, Madrid, Spain) over a poly-d-lysine (0.1 mg⁄mL) layer. Cells were plated in Neurobasal supplemented with 1% FBS, 50 μg⁄mL penicillin ⁄ streptomycin, 50 μg⁄mL gentamycin, and 60 mg⁄mL l-glutamine and maintained in a 5% CO2 incubator at 37°C; this medium contained B27 supplement to facilitate hippocampal neuron survival in vitro. For seven-day-old cultures half the volume of the medium was replaced with fresh Neurobasal-B27 medium. Under these conditions, standard cell survival was 3 weeks, but the experiments were performed on neurons that were 8–12 days old. Selection criteria for current recordings were adhesion to the substrate, soma diameters of 10–20 μm, neuronal shape without evident shrinkage or swelling, neurite extensions, and absence of intracellular vacuoles (Baldelli et al. 2005).

Patch-clamp recordings, data acquisition, and analysis

CS-induced currents, action potentials, and membrane potential changes were recorded using the perforated-patch configuration (Horn and Marty 1988) of the patch-clamp technique (Hamill et al. 1981) under either voltage or current clamp mode. Perforated patch was performed following the method used in Arnaiz-Cot et al. (2008) and Gonzalez et al. (2011). Briefly, we used an intracellular solution containing (in mM): 135 KCl, 10 NaCl, 2 MgCl2, 10 HEPES and 5 EGTA (pH 7.3 with KOH), in addition we dissolved 50–100 μg / mL of amphotericin B as permeabilizing agent (Watsky and Rae 1991). Recordings were made with slightly fire-polished borosilicate pipettes of resistance ranging from 3 to 5 MΩ mounted on the headstage of an EPC-10 patch-clamp amplifier (HEKA Electronic, Lambrecht, Germany). Sampling rate was 20 kHz. Recordings started when series resistance reached less than 20 MΩ. For voltage-clamp recording, only cells with less than 100-pA leak current were included in the study, and for current clamp recordings, a maximum of 10-pA current injection was applied; 75% of cells recordings were obtained in 0-pA current mode. The cells were focused on the stage of a Nikon Eclipse T2000 (Tokio, Japan) inverted microscope and was locally superfused with a Tyrode solution containing (in mM) 2 CaCl2, 137 NaCl, 1 MgCl2, 10 glucose, 5 KCl, and 10 HEPES⁄NaOH (pH 7.4). Solutions were exchanged using a five-ways multibarrelled rapid perifusion system controlled by solenoid valves. The exchange solution flux of the perifusion system was placed within 100 μm of the cell under study. Data acquisition was performed using PULSE programs (HEKA Elektronic). Data analysis was performed using the GraphPad Prism version 5.00 for Window (GraphPad Prism Software, San Diego, CA, USA). All experiments were performed at 22–24°C.

Statistical analysis

Data are expressed as the means ± SEM of the number of cells (n) studied, from at least three different cell cultures. Membrane potential and current peak were measured at the plateau phase of the CS pulse. To calculate EC50, log-agonist versus response was analyzed by non-linear regression. Comparison with an F-test, between log-agonist versus response and log-agonist versus response with a variable slope, gave a p value of 0.3773 and resulted in the best-fitted equation of the first one (non-variable slope analysis). Data were fitted to a sigmoid curve defined by the equation:

display math

where Y is the response measure as ICS peak (pA), X is the Log of CS concentration (M), and EC50 is the molar concentration where the effect is the 50% of the maximum.

For the I-V ICS analysis, we performed linear regression analysis between −20 and +20 mV clamped membrane potential and interpolated the X value when Y = 0 (which is the reverse potential for ICS).

Student's t-test or one-way Anova followed by Tukey or Dunnett post hoc tests was used to determine statistical significance between means. Statistical significance was established at p < 0.05.

Measurements of changes in the cytosolic Ca2+ concentrations ([Ca2+]c)

Hippocampal cells were incubated for 1 h at 37°C in Neurobasal medium containing the calcium probe Fura-2 AM (10 μM). After this incubation period, the coverslips were mounted in a chamber, and cells were washed and covered with Tyrode's solution. Data acquisition and setup for fluorescence recordings were the same as previously reported (Padin et al. 2012). Briefly, a Leica DMI 4000 B inverted light microscope (Leica Microsystems; Barcelona, Spain) equipped with an oil-immersion objective (Leica 40x Plan Apo; numerical aperture 1.25) was used. Once the cells were placed under the microscope, they were continuously superfused by the same system as described in patch-clamp recordings method. Fluorescence images were generated at 1-s intervals. Images were digitally stored and analyzed using LAS AF software (Leica).

Calcium data analysis and statistics

Data analysis was performed using the Graphpad Prism software, version 5.01 (GraphPad Software Inc.,). Maximum peak was calculated by integrating the Ca2+ transient by means of Origin Pro 8 SR2 software, version 8.0891 (OriginLab Corporation, Northampton, MA, USA). One-way anova test and Dunnett's post test was performed using GraphPad Prism version 5.01.

Chondroitin sulfate digestion with chondroitinase ABC

Chondroitinase ABC from Proteus vulgaris;[ EC] (catalog number SigmaAldrich c2905) was used to catalyze the elimination of disaccharides units from polysaccharides containing (1-4)-β-D-hexosaminyl and (1-3)-β-D-glucuronosyl or (1-3)-α-L-iduronosyl linkages, that is, it acts on chondroitin 4-sulfate, chondroitin 6-sulfate, and dermatan sulfate.

Stock solutions of 1 mL CS 10 mM were prepared in a modified Tyrode buffer containing (in mM) 87 NaCl, 1 MgCl2, 10 HEPES, 10 glucose, and 50 sodium acetate as reaction activator. To each stock, 50 mU/mL of enzyme was added and the stocks were then incubated at 37°C for 24 h, after which they were kept at −20°C. For the experiments, 10-μM working solutions of CS from those predigested stocks were used.


MK801, CNQX, NBQX, D-AP-5, AMPA, and kainate were purchased from Ascent Scientific Ltd (UK). GYKI-53566, TTX, and heparin were purchased from Tocris bioscience (Bristol, UK). Other chemical components for solutions were obtained from Sigma-Aldrich. Neurobasal, B-27, FBS, penicillin⁄streptomycin, and gentamycin were purchased from GIBCO-Invitrogen (Barcelona, Spain). Chondroitin sulfate and hyaluronic acid used in this study was provided by Bioibérica (Barcelona, Spain). CS is highly purified chondroitins 4&6 sulfate of bovine origin in a concentration not less than 98% (measured by CPC titration assay (the official assay method of the USP CS monograph and European Pharmacopeia to ensure a correct measure of CS purity and potency). This product from Bioibérica consists of a mixture of CS sulfated at position 4 (62%), 6 (32%), or unsulfated (6%) on the N-acetyl-D galactosamine group. The full range of its molecular weight is ~ 13–16 kDa with an intrinsic viscosity of ~ 0.02–0.06 m3/Kg.


The effects of CS on cell excitability were first explored under voltage clamp, at −80 mV holding potential. The example cell of Fig. 1a exhibited spontaneous post-synaptic currents (sPSCs) similar to those previously recorded in these rat hippocampal cultures (Arnaiz-Cot et al. 2008). sPSCs immediately disappeared on cell perifusion with 10 μM CS (135 μg/mL), that also elicited an inward current of about 450 pA (Fig. 1a), that did not undergo inactivation along the 10-s pulse (ICS). The current quickly relaxed to near zero baseline on CS washout; only small sPSCs appeared during the 10-s period that followed. The morphology of ICS was quite similar to the currents elicited by 30 μM AMPA (Fig. 1b), and 100 μM kainate (Fig. 1c). Large unsulfated glycosaminoglycan such as hyaluronic acid (160 KDa) was tested to discard the possibility that the effect of CS could be because of a non-selective effect associated to molecular size or a disaccharide polymer structure. In the cell shown in Fig. 1d, 2.24 mg/mL hyaluronic acid did not alter baseline; however, 10 μM CS elicited its typical current in the same cell (not shown). We also tested heparin (up to 1 mg/mL) in this regard, and found no effect (not shown).

Figure 1.

Chondroitin sulfate (CS) elicits inward non-inactivating currents (ICS) in hippocampal neurons. (a), example current elicited by CS (top horizontal bar); (b and c), example currents elicited by α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) and kainate, respectively. (d), lack of effect of hyaluronic acid (HA) on spontaneous post-synaptic currents (sPSCs). (e), concentration–response curve for CS to elicit ICS. Inward and outward currents depended on the holding potential (f) and an I–V curve shows that reversal potential was −14 mV (f). Three consecutive CS pulses were applied to the same cell; the substitution of Na+ in second pulse by N-methylglucamine blocked ICS by over 80% (g). ICS was nearly fully blocked by NBQX and CNQX; however, the AMPA receptor selective antagonist GYKI-53655 blocked ICS by 75%, suggesting the involvement not only of AMPA but also kainate receptors. Furthemore, CS pre-treated with enzyme ChaseABC decreased its potency by 60%. Nevertheless D-APV and MK801 did not affect ICS. Data in (g and h) are presented as means ± SEM. **p = 0.0046 with respect to control buffer (g), or ***p < 0.0001 with respect to normalized control (h). One-way anova of variance and Dunnett post hoc test were performed.

The concentration–effect curve for CS showed a threshold current at 1 μM (13.5 μg/mL). A sigmoid curve with an EC50 of 10.9 μM is graphed in Fig. 1e. ICS showed time dependence, but was insensitive to 1 μM tetrodotoxin (not shown). It also was dependent on the holding voltage applied to the cell; thus, ICS showed a reversal potential of −14 mV (Fig. 1f). We also studied the ion dependence of the current and found that the main ion carrying the current was Na+, as the current was blocked over 70% by Na+ substitution for N-methyl glucamine (Fig. 1g).

ICS was highly reproducible on repeated application of 10-s pulses with 10 μM CS, given within the same cell. We took advantage of this protocol to explore the effects of distinct blockers of glutamate receptors on ICS (Fig. 1h). Thus, 40 μM of non-selective blockers for AMPA and kainate receptors NBQX and CNQX perifused few seconds before and during the CS pulse caused 97 ± 1.3% and 96 ± 0.7% of ICS peak blockade, respectively. Also, pre-incubation with 50 μM of the selective AMPA receptor blocker GYKI 53655 caused 79 ± 7.3% blockade. In contrast, 10 μM of the selective NMDA blockers, D-APV and MK-801, did not affect ICS peak in a statistically significant manner (Fig. 1h). Chondroitin sulfate pre-digested with 50 mU/mL of ChaseABC for 24 h produced a current of smaller amplitude (60% reduction) than that evoked by undigested CS (Fig. 1h). We can speculate that there is a minimum size required for CS to have the reported effect below which the molecule becomes inactive.

The large inward Na+-dependent ICS could obviously be associated with cell depolarization; this possibility was explored under current clamp. Thus, in the cell of Fig. 2a, 10 μM CS shifted the membrane potential (Em) from the basal value of −67 mV to −32 mV. This pronounced depolarization was initially accompanied by a brief burst of evanescent action potentials (Fig. 2b). The depolarizing effect of CS mimicked that produced by 30 μM AMPA (Fig. 2c) and 100 μM kainate (Fig. 2d). CS-elicited depolarization was concentration dependent and had an EC50 similar to that of ICS (Fig. 2e).

Figure 2.

Chondroitin sulfate (CS) causes depolarization of current-clamped hippocampal neurons. At 10 μM, CS produced rapid, sustained, and fast reversible depolarization (a) with initial evanescent action potentials (b) and concentration dependency (e). Example depolarization traces elicited by α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) and kainate are shown in (c and d). This effect was nearly fully blocked by CNQX (f) and NBQX (g), suggesting the involvement of AMPA/kainate receptors. AMPA selective blocker GYKI prevented by 50% this depolarization (h). I, normalized pooled results (% over basal change in each cell), of blocker effects. Data in (e and i) are means ± SEM of the number of cells shown in parentheses from at least three different cultures. ***p < 0.0001 with respect to basal (e, i). One-way anova of variance and Dunnett post hoc tests were performed.

The effect of blockers of glutamate receptor subtypes on CS-elicited depolarization was explored following ‘sandwich’-type experiments, as in the case of ICS. Thus, 40 μM CNQX caused a full and reversible blockade of CS depolarization (Fig. 2f); in this cell, spontaneous action potentials were suppressed in the presence of CS, but they recovered on compound washout. Also, at 40 μM, NBQX caused a reversible full blockade of the depolarizing effects of CS, as shown in the example cell of Fig. 2g. However, 50 μM of the selective blocker of AMPA receptors GYKI only caused a partial blockade, about 60% (Fig. 2h). Pooled results from cells of various cultures are shown in Fig. 2i. Thus, blockade of CS depolarizing effects was 51 ± 3.7%, 79 ± 5.9%, and 93 ± 2.4% for GYKI, NBQX, and CNQX, respectively; however, NMDA receptor blockers MK801 and D-APV had no effect.

The use of blockers for the various subtypes of ionotropic glutamate receptors suggested a major role for AMPA receptors in the CS-elicited inward currents and cell depolarization. Thus, the question arose as to whether cyclothiazide, a positive allosteric modulator of AMPA currents (Yamada and Tang 1993), also exerted this action on ICS. The experiment of Fig. 3a proved that this was so. A first CS pulse (10 μM) caused an initial peak ICS of 429 pA. When the CS pulse was repeated in the presence of 100 μM cyclothiazide, a rapid activation of an inward current (485 pA) was followed by a slow component of activation that peaked at 1116 pA. This current profile was very close to that elicited by combined cyclothiazide plus 30 μM AMPA: small rapid activation (with some initial inactivation) was followed by a gradual current activation (Fig. 3b). Pooled results from cells of various cultures indicated that cyclothiazide augmented peak ICS by 2.7-fold and the peak IAMPA by 2.2-fold.

Figure 3.

Positive allosteric modulation by cyclothiazide of inward currents elicited by chondroitin sulfate (CS) and α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA), and generation of cytosolic Ca2+ transients ([Ca2+]c). The Icurrent elicited by CS (a, left trace) was drastically enhanced if simultaneously given with cyclothiazide (a, right part of the trace). A similar current facilitation was produced when using AMPA (b). Such facilitation developed slowly in both cases; ICS augmented nearly threefold and IAMPA over twofold (c). At 10 μM, CS caused a [Ca2+]c transient elevation that was similar to that produced by 70 mM K+ (left part of trace in d). CNQX fully blocks the transient that recovered after washout (middle and right part of the trace in d, and e). Data in c and e are means ± SEM of the number of cells shown in parentheses from at least three different cultures. One-tailed paired t-test analysis showed a ***p = 0.0003 with respect to control current or $$$p = 0.0009 with respect to AMPA current (c). Data in e showed a p < 0.0001 with respect to control [Ca2+]c elicited by CS (One-way anova and Dunnett post hoc).

AMPA receptors are Na+ and Ca2+ permeable (Hollmann et al. 1991) and thus, it seemed appropriate to test whether CS, that causes cell depolarization, could elicit a [Ca2+]c elevation. In the fura2-loaded cell of Fig. 3d, an initial K+ pulse (70 mM K+, low Na+) elicited a [Ca2+]c transient that was cleared following a rapid initial phase (probably because of rapid mitochondrial Ca2+ buffering) and a slower phase (probably associated with subsequent mitochondrial Ca2+ release into the cytosol). Of interest was the fact that a 10 μM CS pulse elicited a [Ca2+]c transient very close to that of K+. Such transient was abolished in the presence of 40 μM CNQX along the three sequential pulses of CS. On blocker washout, the response fully recovered. Pooled results from 21 cells indicated a blockade by CNQX of the CS [Ca2+]c transient of 95.4 ± 1.2%.


In this study, we have shown that CS produces a Na+-dependent inward whole-cell current; such current causes cell depolarization and a [Ca2+]c transient in primary cultures of rat embryo cocultures of glia and hippocampal neurons. These effects are probably because of direct activation by CS of AMPA and kainate receptors for glutamate because (1) ICS had a kinetic profile similar to IAMPA and IKainate; (2) non-selective blockers of AMPA/kainate receptors as NBQX and CNQX fully inhibited the CS effects; and selective blockers of NMDA receptors such as MK801 and D-APV did not modify those effects. AMPA receptors could play a dominant role in the CS effects because: (i) the selective AMPA receptor blocker, GYKI 53655, inhibited by 79% ICS (but the depolarizing effect was only halved), and (ii) the positive allosteric modulator cyclothiazide augmented ICS following a kinetics and amplitude similar to IAMPA.

To our knowledge, there is only one report showing that CSPG immobilized on beads, produced a mild, slowly developing elevation of [Ca2+]c in dorsal root ganglion neurons; this effect was attributed to the CS-GAG chain of the proteoglycan, as the enzymatic removal of the carbohydrate moiety canceled the response (Snow et al. 1994). Our study shows in a more direct way that free CS in solution causes a very rapid, sharp [Ca2+]c transient, which closely mimics what is elicited by a depolarizing K+-based solution. In the light of these results, the interesting question emerges as to whether a fraction of CS could be found free in the brain under physiological or pathological conditions, in addition to its more abundant state of attachment to a core protein in the form of proteoglycan.

CS in form of proteoglycan is present in CN tissues. CSPGs are widely expressed in the developing and the adult central and peripheral nervous system and have a potent inhibitory influence on regeneration within the CNS. Few hours after injury to the CNS, a glial scar is formed primarily by reactive astrocytes, which, as well as oligodendrocyte precursor cell (OPC), become hypertrophic and increase expression of scar-associated inhibitory CSPGs (Silver and Miller 2004). This up-regulation in CSPGs create an inhibitory gradient (Asher et al. 2001) that is highest at the center of the lesion and diminishes gradually into the penumbra area. The regeneration failure (Zuo et al. 1998a) that occurs in this scar is overcome with ChaseABC treatment, an enzyme that removes GAG chains from the protein core (Prabhakar et al. 2005), which eliminates or reduces CSPG-associated inhibition (Zuo et al. 1998b) (Barritt et al. 2006). In vivo treatment with ChaseABC enhances the regeneration of the axons of dopaminergic neurons (Moon et al. 2002) and promotes axonal regeneration and functional recovery after spinal cord injury (Moon et al. 2001; Bradbury et al. 2002; Yick et al. 2003; Caggiano et al. 2005). CSPGs are constituents of the specialized form of the extracellular matrix in certain types of neurons called perineural net. This structure stabilizes receptor diffusion and modulates and directs synapse formation.

Although forming an important part of the brain ECM and PNN, it is unlikely that bound CS (CSPG) could bind to AMPA receptors to exert the actions found in this study on cultured hippocampal neurons. However, CS-GAG degradation with bacterial ChaseABC opens up spaces for axonal growth and the formation of new neuronal contacts, as shown in various experimental models (Moon et al. 2001; Bradbury et al. 2002) and it is in this experimental approach where free CS-GAG molecules could be found in nervous system. Physiologically, though, the counterpart of bacterial ChaseABC are matrix metalloproteinases (MMPs), which are capable of degrading components of the ECM and are essential for glial scar formation (Hsu et al. 2008). If we focus on CS, MMP-2 and MMP-3 are the only two enzymes that can digest CS-GAGs from CSPGs, and both can individually activate MMP-9. This protease cascade is up-regulated after traumatic brain injury and focal cerebral ischemia (Park et al. 2009). Degradation of extracellular molecules by MMPs also facilitates cell migration and neurite outgrowth; interestingly, axonal outgrowth is facilitated when MMP-2 is localized to axonal growth cones (Hayashita-Kinoh et al. 2001; Pizzi and Crowe 2007). The concentration of MMP2 to the leading edge of an extending axon facilitates the degradation of inhibitory CSPGs in the sciatic nerve (Zuo et al. 1998b). Moreover, some studies provide explicit evidence that MMP-9 is integral to the formation of an inhibitory glial scar and cytoskeleton-mediated astrocyte migration. Matrix metalloproteases are expressed at significant levels in regions of neural plasticity in the adult CNS, such as the cerebellum (Vaillant et al. 1999; Hayashita-Kinoh et al. 2001) and may also be concerned with the structural plasticity in the hypothalamus–neurohypophysial system (Miyata et al. 2005). Matrix metalloproteases also are rapidly up-regulated after nearly all types of CNS insult, including spinal cord injury (Xu et al. 2001), Alzheimer′s disease (Yoshiyama et al. 2000), and stroke (Sole et al. 2004). Thus, degradation of the CSPGs facilitates cell migration and axonal cones growth (Hayashita-Kinoh et al. 2001). But, in addition to this structural role for CSPGs, a more active physio-pharmacological role of these degradation products has also been suggested (Rolls et al. 2004). This may be the case for CS that, as shown in this study, caused a surprising direct activation of an AMPA/kainate receptor current in the absence of the physiological excitatory neurotransmitter glutamate, with concomitant cell depolarization and an augmentation of [Ca2+]c.

At inhibitory hippocampal neurons, Ca2+-permeable AMPA receptors dominate; in contrast, pyramidal cells mostly express GluR2-containing Ca2+- impermeable AMPA receptors (Geiger et al. 1995; Isa et al. 1996). These receptors mediate excitatory synaptic transmission and play a key role in hippocampal LTP and depression (LTD). We provide here solid evidences favoring a role of CS-GAG in binding to, and activating AMPA receptors very much as AMPA (and glutamate) do.

Although a physiological role for this action at intrasynaptic AMPA receptors is hard to envisage, an action on extrasynaptic AMPA (or kainate) receptors could be plausible. We however favor the hypothesis that under pathological conditions (i.e. CNS trauma or stroke), MMP elicited degradation of ECM and PNN CSPGs, could liberate free CS to elicit [Ca2+]c signal mediated by AMPA receptors, to facilitate Ca2+-dependent processes such as tissue repair, cell migration, or axon regrowth.

Several authors have described the relevance of sulfated groups as well as their position, regarding the various biological functions attributed to CSPGs. For instance, CS-A is monosulfated at carbon 4 and is the most prominent form within normal CNS. However, after brain injury, CS-C (monosulfated at carbon 6) and its synthetic enzyme chondroitin 6-sulfotransferase 1 (C6ST1) are up-regulated in most glia cell types around a cortical lesion (Properzi et al. 2005). On the other hand, the 4,6-sulfated CS (CS-E) is the most prominent GAG disaccharide in the scar of injured cortex (Gilbert et al. 2005). Regarding axon outgrowth, it seems to be regulated by a ‘sulfation code’ (Gama et al. 2006); thus, whereas CS-C is particularly inhibitory, CS-D (two and six sulfated) and CS-E (two and six sulfated) promote embryonic axon elongation in vitro (Clement et al. 1998, 1999). The CS used in this study contains 6% CS-0S, 62% CS-4S, and 32% CS-6S (Bioiberica). A form of CS-4S from Sigma (catalog number C-9819) generated ICS with a potency 30 times lower to that found in this study (data not shown). It seems, therefore, that the composition of CS with sulfation at different carbons and with different degrees of sulfation (net negative charge), as well as the molecular size could exert different effects on cultured hippocampal neurons.

In conclusion, we believe this is the first report showing a direct role of the glycosaminoglycan CS to open AMPA receptors causing cell depolarization and rapid [Ca2+]c signals. Other ECM molecules have been shown to modulate neurotransmitter effects on brain receptors and ion channels (Xiao et al. 1999; Strekalova et al. 2002), polisialic acid being one of them (Senkov et al. 2012). Even sulfated polysaccharides as dextrans have been studied in this regard (Suppiramaniam et al. 2006); however, a direct agonist effect of those compounds have been unrecognized until now. Being the most abundant at the brain ECM and PNN, CSPGs contribute to various physiological roles such as serving as a releasable pool of neurotrophic factors or to exert neuroprotection against excitotoxic/free radicals stress, or to promote/inhibit neurite outgrowth actions depending on sulfation pattern and size (Gama et al. 2006). This unique direct action of CS on AMPA and kainate receptors opens new pathways to understand the complex physiological and pathological roles of ECM CSPGs.


The authors declare no competing financial interest. Authorship contributions: Conception of the work: A.G.G., J.V., and E.M. Design of research: A.G.G., M.M., J.M.H-G., and Diego. Acquisition and data analysis: M.M., J.C.F-M., J.F.P., J.C.G. Interpretation of the results: A.G.G., J.M.H-G., M.M., and J.C.F-M. Drafting and writing: A.G.G., M.M. Revision: J.V, E.M., Diego, and A.G.G. This work was supported by the following grants to AGG: (1) SAF 2010-21795, Ministerio de Economía y Competitividad, Spain; (2) RENEVAS-RETICS-RD06/0026, Instituto de Salud Carlos III, Spain; (Gama et al.) CABICYC, UAM/Bioibérica, Spain. We thank the continued support of Fundación Teófilo Hernando, Madrid Spain.