Rapid Report

Degradation of extracellular chondroitin sulfate delays recovery of networkactivity after perturbation

Amber E. Hudson,1 Clare Gollnick,2 Jean-Philippe Gourdine,3 and Astrid A. Prinz1

1Department of Biology, Emory University, Atlanta, Georgia; 2Department of Biomedical Engineering, Georgia Institute ofTechnology, Atlanta, Georgia; and 3Department of Biochemistry, Emory University, Atlanta, Georgia

Submitted 11 May 2015; accepted in final form 22 June 2015

Hudson AE, Gollnick C, Gourdine JP, Prinz AA. Degradation ofextracellular chondroitin sulfate delays recovery of network activityafter perturbation. J Neurophysiol 114: 1346–1352, 2015. First pub-lished June 24, 2015; doi:10.1152/jn.00455.2015.—Chondroitin sul-fate proteoglycans (CSPGs) are widely studied in vertebrate systemsand are known to play a key role in development, plasticity, andregulation of cortical circuitry. The mechanistic details of this role arestill elusive, but increasingly central to the investigation is the ho-meostatic balance between network excitation and inhibition. Study-ing a simpler neuronal circuit may prove advantageous for discover-ing the mechanistic details of the cellular effects of CSPGs. In thisstudy we used a well-established model of homeostatic change afterinjury in the crab Cancer borealis to show first evidence that CSPGsare necessary for network activity homeostasis. We degraded CSPGsin the pyloric circuit of the stomatogastric ganglion with the enzymechondroitinase ABC (chABC) and found that removal of CSPGs doesnot influence the ongoing rhythm of the pyloric circuit but does limitits capacity for recovery after a networkwide perturbation. WithoutCSPGs, the postperturbation rhythm is slower than in controls andrhythm recovery is delayed. In addition to providing a new modelsystem for the study of CSPGs, this study suggests a wider role forCSPGs, and perhaps the extracellular matrix in general, beyondsimply plastic reorganization (as observed in mammals) and into afoundational regulatory role of neural circuitry.

central pattern generator; extracellular matrix; homeostasis

HOMEOSTATIC REGULATION of neuronal activity is integral toneural function. Without it, neurons would be in danger ofdrifting toward silence or death via excitotoxicity. However, itis not known how a cell or network defines appropriate activityand regulates accordingly over the long term. Recently, theextracellular neural environment has been recognized as wellplaced to direct long-term cell- or network-level neuronalhomeostasis. One family of extracellular matrix (ECM) mole-cules in particular, chondroitin sulfate proteoglycans (CSPGs),have properties that are ideal for homeostatic feedback (Gal-trey and Fawcett 2007). CSPGs are produced locally by neu-rons and glia (Giamanco and Matthews 2012), with depen-dence on cellular activity patterns (Lander et al. 1997) viaCa2� influx (Dityatev et al. 2007). Structural differences withinthe CSPG family are cell type specific (Lander et al. 1997) andresult in diverse functional roles, such as alternately preventingor encouraging neuronal structural plasticity and pathfinding(Balmer et al. 2009; Fongmoon et al. 2007; Gogolla et al. 2009;Kwok et al. 2012; Snow et al. 1994), binding and presentationof growth factors (Galtrey and Fawcett 2007), and modulation

of ion channel properties (Dityatev et al. 2007; Maroto et al.2013; Snow et al. 1994; Vargas and De-Miguel 2009; Vigettiet al. 2008). The type specificity, activity dependence, anddiverse functional roles of CSPGs together suggest a way bywhich changes in activity output can feed back onto regulatorymechanisms—the hallmark of a homeostatic process. How-ever, all components of this role have not been verified in onesystem, possibly because of technical challenges in mamma-lian and culture-based models. Here, we introduce the use ofinvertebrate systems to study CSPGs and their role in network-level neuronal activity maintenance.

The stomatogastric ganglion (STG) of decapod crustaceans,such as lobsters and crabs, is a model system for studyingactivity homeostasis at cellular and circuit levels. The pyloriccircuit within the STG controls the rhythmic contractions of thepylorus. The component neurons, connectivity, and activity ofthe pyloric circuit are known; it is comprised of �11 neuronsthat generate a triphasic periodic rhythm. The STG is part ofthe stomatogastric nervous system (STNS), which includesconnections to three other ganglia and the brain (Fig. 1A), butthe STG is easily isolated from all projection neurons andneuromodulatory input via a nerve called the stomatogastricnerve (stn). If conduction along the stn is blocked or the stn iscut, known as deafferentation, the pyloric rhythm slows orstops within minutes (Thoby-Brisson and Simmers 1998) (Fig.1B). Over days the activity pattern is spontaneously restored inthe absence of inputs through the stn, and this recovery afterdeafferentation has been a premier model of neuronal activityhomeostasis.

The current understanding of network activity recovery afterdeafferentation in the STG provides a description of cellularand synaptic changes but does not fully address what is drivingthese changes. After the stn is cut, it can no longer transportneuromodulators (NMs) from anterior ganglia. This removes atleast one NM-dependent current (IMI) from several cell types inthe STG, altering their activity (Golowasch and Marder 1992).Changes in intrinsic K� and Ca2� currents and synaptic efficacyoccur after deafferentation to compensate for this loss and restorepyloric activity (Haedo and Golowasch 2006; Khorkova andGolowasch 2007; Thoby-Brisson and Simmers 2002). Thedriving factor behind these changes is unclear; where somestudies suggest activity-dependent or Ca2�-dependent changes(Golowasch et al. 1999), others suggest that the presence orabsence of NMs is the only signal required to induce recovery(Khorkova and Golowasch 2007; Temporal et al. 2012). How-ever, the time to first bout (tbout) (suddenly increased fre-quency) after deafferentation depends on the history of net-work activity before deafferentation or the application of NMs

Address for reprint requests and other correspondence: A. A. Prinz, EmoryUniv., O. Wayne Rollins Research Ctr., Rm. 2105, 1510 Clifton Rd., Atlanta,GA 30322 (e-mail: astrid.prinz@emory.edu).

J Neurophysiol 114: 1346–1352, 2015.First published June 24, 2015; doi:10.1152/jn.00455.2015.

1346 0022-3077/15 Copyright © 2015 the American Physiological Society www.jn.org

that alter network activity, suggesting that recovery requirescoordinated action of both activity- and NM-dependent path-ways (Zhang et al. 2009). Theoretical work has suggested thatactivity-dependent control of ion channel distributions could beworking in concert with NM-dependent intracellular Ca2�

buffering (Zhang and Golowasch 2011), although no experi-mental verification of the involvement of any intracellularregulatory cascades has been published. These models assumea black-box intracellular Ca2� sensor as the integrating com-ponent for observations of activity and feedback to the sys-tem—but this component need not be intracellular if thecondition of feedback on ion channel activity is met. In fact,compartmentalization of this feedback component outside ofthe cell may be an advantage for networkwide regulation orprevention of interference in other intracellular pathways.CSPGs and associated ECM molecules are candidates for thiskind of long-term homeostatic regulation, as shown in recenttheoretical work (Kazantsev et al. 2012).

Chondroitin sulfate (CS) structure is conserved betweeninvertebrates and vertebrates (Fongmoon et al. 2007; Medeiroset al. 2000; Selleck 2001; Sugahara and Yamada 2000), butwhile there has been extensive study of the structural similar-ities, the functional roles of CS in invertebrate nervous systemshave not been described. Fongmoon et al. (2007) demonstratedthat invertebrate CS from cartilage functions similarly to mam-malian neuronal CS, by introducing king crab CS to culturedmouse hippocampal neurons. Invertebrate CS increased neuriteoutgrowth, similar to the effects of mammalian CS with com-parable sulfation patterns. This is evidence that invertebrate CS

has functional similarity to mammalian CS, but no studies havedetermined whether invertebrate neurons also respond to CS ina similar way. Here we describe such an experiment and shedlight on CS as a participant in homeostatic activity regulation.We show evidence of CS in crab neural tissue and then showthat CS degradation in the STG alters recovery of the pyloricrhythm after deafferentation. Surprisingly, ongoing activity inan intact STNS is unaffected by removal of CS, suggesting thatthe influence of CS on activity regulation is specific to homeo-static pathways.

METHODS

All experiments were performed with the STNS of the crab Cancerborealis. The intact STNS was removed from surrounding tissue aspreviously described (Gutierrez and Grashow 2009).

Western blots. Crab brains were removed during STNS dissectionand flash frozen in liquid nitrogen. Brains were pooled and processedas previously described (Deepa et al. 2006) with a series of threeextraction buffers: buffer 1 [in mM: 50 TBS, pH 7.0, 2 EDTA, 10N-ethylmaleimide, and 2 phenylmethylsulfonyl fluoride, with 1 com-plete EDTA-free protease inhibitor cocktail tablet (Roche)], buffer 2(buffer 1 containing 0.5% Triton X-100), and buffer 3 (buffer 2 with6 M urea). A subset of samples were incubated with 2.5 U/mlprotease-free chondroitinase ABC (chABC, Sigma) for 2 h at 37°Cbefore use in a Western blotting protocol using mouse anti-CS IgMCS-56 (1:10,000, Sigma) as primary antibody and goat anti-mouseIgM (1:5,000, KPL) as secondary.

Organ culture and long-term electrophysiology. The intact STNSwas removed from the crab and placed in culture with L-15 medium

Fig. 1. The stomatogastric nervous system (STNS)as a model of network homeostasis and perineuronalchondroitin sulfate (CS). A: the intact STNS asmaintained in culture. Pink arrows show severedconnections to the brain (brain not included inculture). Gray circles indicate Vaseline wells used toisolate nerves and ganglia for administration ofsolutions [around stomatogastric ganglion (STG)] orelectrical recordings [for lateral ventricular nerve(lvn) and pyloric dilator nerve (pdn)]. Blue arrow-head indicates site of deafferentation. stn, stomato-gastric nerve. OG, oesophageal ganglion; CoG,commissural ganglion. B: the pyloric circuit loses,and subsequently recovers, triphasic activity afterdeafferentation. Top: example extracellular record-ings of the lvn (d0, before; d1, within 24 h of; d3,72–96 h after deafferentation) with color-coded py-loric cell types in selected cycles (LP, lateral pylo-ric, red; PY, pyloric, green; PD, pyloric dilator,blue). Scale bar is 1 s. Bottom: pyloric frequency vs.time in culture. Deafferentation occurs at t � 0 (bluearrowhead). Recordings (and treatment, if applica-ble) began 12–18 h earlier (data not shown inentirety). C: Western blots using anti-CS (CS-56)primary antibody show CS immunoreactivity in ho-mogenized Cancer borealis brain. Decorin was usedas a positive control (lane d). Decorin treated withchondroitinase ABC (chABC; lane dc) was used asa negative control. Lanes 1–3 are crab brain samplestreated sequentially with buffers 1, 2, and 3 adjacentto the same samples treated with chABC (1c, 2c, 3c,respectively); n � 2.

Rapid Report

1347CSPGs AND NEURONAL HOMEOSTASIS

J Neurophysiol • doi:10.1152/jn.00455.2015 • www.jn.org

supplemented with 40 U/ml penicillin, 40 �g/ml streptomycin(Sigma), and 1 g/l glucose and adjusted for final salt concentrations (inmM) of 440 NaCl, 11 KCl, 13 CaCl2, 26 MgCl2, 11.2 Trizma base,and 5.1 maleic acid. After verification of a stable rhythm (30–60min), 4 U/ml chABC in medium was added to a Vaseline well aroundthe desheathed STG and allowed to sit for 12–18 h. Alternatively,medium only or denatured chABC (DNchABC) was used as control.After this incubation, the enzyme was removed by rinsing the wellthree times with medium. Some STGs were removed at this stage, andan immunohistochemistry protocol was performed (Hudson 2013). Inother experiments, the stn was cut after rinsing. Sterile medium wasreplaced every 12–24 h, and the activity of the STG was monitored forat least 5 days after deafferentation. Enzyme samples were tested forrelative enzymatic activity with a standard dimethylmethylene blue(DMMB) protocol (Lee et al. 2010).

Data were analyzed with custom MATLAB scripts. In most cases,the burst frequency of the pyloric dilator (PD) neuron as recorded onthe pyloric dilator nerve (pdn) was used as a proxy for pyloricfrequency. If the pdn was not available, burst frequency of the lateralpyloric (LP) neuron or PD neuron was recorded on the lateralventricular nerve (lvn). Student’s t-tests, single-factor ANOVA, andTukey-Kramer tests were used for statistical analysis. Standard errorwas used for reported confidence measures.

RESULTS AND DISCUSSION

Chondroitin sulfate in the STG. CS is present in inverte-brates (Fongmoon et al. 2007; Medeiros et al. 2000; Selleck2001; Sugahara and Yamada 2000), but studies to date haveanalyzed CS from connective (Kozlowski et al. 2011) ratherthan neural tissues. CS from crab cartilage has been success-fully digested by chABC (Fongmoon et al. 2007), but itsimmunoreactivity with common anti-CS antibodies is unclear.In this study we used an antibody that binds to the CS familyof molecules generally (monoclonal CS-56). The CS-56 anti-body displayed immunoreactivity in homogenized crab brainvia Western blotting (Fig. 1C) and in the STG via immuno-histochemistry (data not shown, n � 6 per treatment group).The specificity of this binding, however, has not been previ-ously verified in this system. To test this, we used a Westernblot protocol used previously in the study of perineuronal nets(PNNs) in mammalian systems (Deepa et al. 2006) that in-cludes a series of three extraction buffers that separate diffuseECM (buffers 1 and 2) from more dense and cell membrane-specific types of ECM (buffer 3). We paired this protocol withchABC pretreatment of the extracts to identify specific

Fig. 2. chABC treatment has no effects on pyloric rhythm inlong-term organ culture. A and B, top: untreated, n � 5.Middle: chABC treatment, n � 6. Bottom: denaturedchABC (DNchABC) treatment, n � 4. A: example extra-cellular recordings of the lvn at days 1 (d1) and 2 (d2) andat the times of last triphasic activity (t-Tri) and last stablebursting activity of the PD neurons (t-PD) or LP neuron(t-LP). B: pyloric frequency vs. time in culture. Shadingindicates treatment condition and duration. C–E: gray, un-treated; orange, chABC treatment; blue, DNchABC treat-ment. C: pyloric frequency after 20 h in culture. D: slopebetween pyloric frequency at 20 h and frequency of laststable bursting activity. Example shown in B with purpleline marked d. E: time at last stable bursting activity.

Rapid Report

1348 CSPGs AND NEURONAL HOMEOSTASIS

J Neurophysiol • doi:10.1152/jn.00455.2015 • www.jn.org

(chABC sensitive) vs. nonspecific (chABC insensitive) bindingof the antibody. We found that a portion of the binding inextracts from detergent (buffer 2) and urea (buffer 3) fractionsis prevented by pretreatment with chABC, demonstrating thepresence of chABC-sensitive CS in these extracts. Extract 2shows bands around 260,000 and at high molecular weights(MWs) that are reduced in the chABC-treated sample. Extract3 has a single large band at high MW that is absent afterchABC treatment. This is particularly interesting becausePNNs are primarily extracted by buffer 3 in mammalian sys-tems (Carulli et al. 2010; Deepa et al. 2006), suggesting thatthe CS found in crab brain could be PNN-like. It is importantto note that the CS-56 binding in the saline-eluted fractions ofbrain tissue (buffer 1) was not changed after treatment withchABC, and we found diffuse staining in the STG immuno-histochemistry results that was similarly unaffected by chABCpretreatment. We interpret these results as either nonspecificbinding of this antibody in this system or the presence of aCS variant not degraded by chABC. The mammalian CSPGdecorin was also used as a positive technical control in ourWestern blotting protocol to ensure that the chABC enzymewas sufficiently active. The decorin control showed ahigh-MW band before chABC treatment and no staining aftertreatment, indicating activity. Our results are the first demon-stration of neural CS in decapod crustaceans and show thepresence of high-MW molecules with chABC susceptibility,particularly in fractions designed to isolate membrane-associated molecules and stable ternary complexes such asthose found in PNNs (Carulli et al. 2010; Deepa et al. 2006).This opens up invertebrate neuroscience—with its advantages

of identified cells and small, defined networks—to the study ofcellular and network effects of perineuronal CSPGs.

Chondroitinase does not interrupt ongoing pyloric rhythm.The pyloric circuit is comprised of four main cell types: theanterior burster neuron (AB), the two PDs, the LP, and thethree to eight pyloric neurons (PYs). These neurons fire in atriphasic rhythm, in which AB and PD fire simultaneously,then LP, then PY (Fig. 1B). This rhythm can be monitoredextracellularly via the lvn, through which all STG neurons sendprojections, or the pdn, which carries only action potentialsfrom the PD neurons (Fig. 1A).

To determine whether CS influences ongoing neuronal ac-tivity, we introduced chABC to cultured intact STNSs. Dis-sected STNSs were placed in culture, and electrical activity ofthe pdn and lvn was recorded with steel extracellular electrodesfor 5–7 days. In the first hour of organ culture, treatment(chABC in medium) or sham treatment (medium only orDNchABC in medium) solutions were added to a well sur-rounding the STG. These solutions were allowed to sit 12–18h before they were rinsed and replaced with medium. Nodifference between treatment groups in the nature or progres-sion of pyloric circuit activity from intact STNSs was seen overthe duration of organ culture (Fig. 2). No significant differ-ences were seen between the three treatment conditions for anyof the investigated activity metrics, and no qualitative differ-ences in activity pattern were observed. For instance, after anaverage of 4 days in culture (105 � 9 h), most preparationsceased rhythmic activity, and this was common to all treatmentgroups. Other metrics that were unchanged between groupsincluded the last cell type to stop bursting (Fig. 2A), the timeand frequency of last stable bursting (defined as continuous

Fig. 3. chABC treatment alters pyloric rhythm recoveryafter removal of neuromodulator via deafferentation. A andB, top: chABC treatment, n � 8. Bottom: DNchABCtreatment, n � 8; untreated (see Fig. 1B), n � 6. A: exampleextracellular recordings of the lvn (d0, before; d1, within 24h of; d3, 72–96 h after deafferentation). B: pyloric fre-quency vs. time in culture. Deafferentation occurs at t � 0.Recordings (and treatment, if applicable) began 12–20 hearlier (data not shown in entirety). Shaded areas indicatetreatment duration and type (orange, chABC; blue, DNch-ABC; gray in C–E indicates no treatment). C: averagepyloric frequency in the 10 h before deafferentation andmaximum average pyloric frequency between 10 and 120 hafter deafferentation as measured by PD burst frequency. D:time to first bout (tbout). E: time to half-maximum pyloricfrequency (t1/2). Error bars represent SE. **P � 0.01; �P �0.075.

Rapid Report

1349CSPGs AND NEURONAL HOMEOSTASIS

J Neurophysiol • doi:10.1152/jn.00455.2015 • www.jn.org

bursting activity with pyloric frequency � 0.2 Hz; Fig. 2E),measurements of pyloric frequency at particular time points(Fig. 2C), and the slope between the frequency at hour 20 inculture and the frequency at the last PD bursting activity (Fig.2D). LP-on phase, as previously defined (Luther et al. 2003),was also calculated for the first hour of culture, at 10 h ofculture, and after 20 h (not shown), and no differences wereseen between groups (ANOVA P values of 0.09, 0.08, and 0.11for 1, 10, and 20 h, respectively). These results demonstratethat degradation of CS in the STG of the intact STNS does notaffect ongoing network activity for up to a week aftertreatment.

This result is reminiscent of experiments in the mammalianvisual system in which chABC treatment does not affectactivity characteristics or ocular dominance in normal animalsbut does alter ocular dominance in monocularly deprivedanimals (Pizzorusso et al. 2002, 2006). In this and othersystems, it appears that CSPGs do not influence normal ongo-ing activity of mature circuits but rather participate in cellular,synaptic, and networkwide changes only after a significantperturbation (Gogolla et al. 2009; Pizzorusso et al. 2002,2006).

Chondroitinase treatment prevents or delays recovery. Wenext asked whether CS in the STG participates in homeostaticfeedback during recovery after deafferentation. To test this,STNSs were cultured as described above; however, in theseexperiments the stn was cut after the treatment solutions werereplaced with medium (12–18 h into culture). This removesneuromodulatory input to the pyloric circuit and causes thepyloric rhythm to slow or stop within minutes. In some cases,bursting activity goes through a period of bouting beginning5–25 h after the cut. Bouting is defined as alternating periodsof time, on the order of minutes, in which the rhythm getsfaster and slower; specifically, a bout is at least three pyloric

cycles with an instantaneous frequency 40% above the baselinefrequency, as previously described (Zhang et al. 2009). Base-line frequency was defined as the average frequency of theprevious 7 min or a minimum of 50 pyloric cycles. Stablebursting activity returns an average of 3 days after deafferen-tation in control preparations (72.8 � 14.8 h) at a postdeaffer-entation maximum average frequency (Fmax) approximatelyhalf of the original value (0.66 � 0.09 Hz) (Fig. 3). Fmax wasdefined as the maximum value of a 5-h moving average thatscanned instantaneous frequency values from 10 h after deaf-ferentation to the last recording.

Chondroitinase treatment before deafferentation delayed thetime to maximum average burst frequency after deafferentation(Figs. 3 and 4) but did not significantly alter other activitymetrics (Fig. 3, D and E). The Fmax of all control pyloriccircuits occurred within the first 100 h after deafferentation.During this time frame, chABC-treated preparations recoveredbursting activity; however, their Fmax values were lower (0.23 � 0.03Hz) compared with untreated activity and those treated withdenatured chondroitinase (P � 0.01) (Fig. 3C). Denaturedchondroitinase and control preparations were not significantlydifferent in any metric investigated, suggesting that it is theenzymatic activity of chABC that is mediating its effects.There was no difference between groups in the incidence oftriphasic recovery patterns—50% of pyloric circuits demon-strated triphasic activity after deafferentation regardless oftreatment conditions—or the qualitative appearance of burst-ing activity after recovery. These results show that extracellu-lar CS is involved in the process of recovery of pyloric activityafter a perturbation and that removal of CS via chABC pre-vents recovery in the first 5 days after deafferentation.

Previous work on recovery mechanisms of the pyloric circuithas shown that cessation of pyloric activity before deafferen-tation via PD hyperpolarization or application of low-Na�

Fig. 4. Recovery profiles after chondroitin sulfate proteoglycan (CSPG) degradation fall into 2 categories. A: example extracellular recordings of the lvn (d1,before; d2, within 24 h of; dX, day X after deafferentation). Scale bar is 1 s. B: pyloric frequency vs. time in culture. Deafferentation occurs at t � 0. All culturedata shown. Shaded areas indicate duration of treatment with chABC. C: average time to maximum average frequency (Fmax). D: average pyloric frequency inthe 10 h before deafferentation and maximum average pyloric frequency from 10 h after deafferentation to the last activity recorded in each experiment, asmeasured by PD burst frequency. *P � 0.05. Error bars represent SE. Colors as defined in Fig. 3.

Rapid Report

1350 CSPGs AND NEURONAL HOMEOSTASIS

J Neurophysiol • doi:10.1152/jn.00455.2015 • www.jn.org

saline causes an advance in the tbout, suggesting that tbout is anindicator of an activity-dependent regulatory mechanism(Zhang et al. 2009). tbout was defined as the time from deaf-ferentation to time of the first recorded bout that occurred afterthe time to half-maximal activity (t1/2), which was defined asthe time from deafferentation to the time at which median-filtered (order 3) pyloric frequency first dipped below half ofthe predeafferentation pyloric frequency. tbout was not signifi-cantly different between the three conditions when a 95%confidence interval was used; however, the P value was nota-bly low at P � 0.075 (Fig. 3D). The suggestive (P � 0.1),though not significant, delay in tbout for chABC-treated STGssuggests that CS may be synthesized and secreted locally overtime such that treated networks eventually go through periodsof bouting and recovery. To determine whether treated net-works eventually recover, chABC-treated preparations wereallowed to persist in culture beyond 120 h (Fig. 4). In three ofeight chABC-treated STGs, the pyloric activity did not in-crease in frequency past 120 h after deafferentation. Thesepreparations show a “flat” recovery profile, meaning that py-loric activity remains consistently very slow after deafferenta-tion and eventually stops. In the remaining five cases, activityeventually returned to control recovery values of up to half theoriginal frequency. These preparations show a “slow ramp”recovery profile; the pyloric frequency starts slow after deaf-ferentation and gradually increases. The average Fmax valueafter deafferentation for chABC-treated pyloric circuits (0.48 � 0.08Hz) was not significantly different from values for untreated(0.66 � 0.09 Hz) or DNchABC-treated (0.62 � 0.09 Hz)controls when no time constraints were placed on the analysisperiod. However, the time needed for chABC-treated prepara-tions to reach these values was significantly greater (139 � 23h) than both untreated and DNchABC-treated circuits (P �0.05). The significant delay to reach control Fmax values sug-gests that some component of the recovery mechanism is beinginterfered with temporarily; from this data we hypothesize thatdegradation of CS by chABC treatment is followed by synthe-sis and secretion of CS by STG neurons and glia to replacewhat was lost. CSPGs in mammalian systems are replenishedon the order of days to weeks after chABC treatment (Lin et al.2008), so secretion by STG neurons and glia would fit thetimeline of this experiment.

Across model systems, most of the neuronal effects ofCSPGs fall into two categories—structural plasticity and neu-ronal excitability. CSPGs inhibit structural plasticity via theirrole in the glial scar after injury, axonal pathfinding, andsynapse formation during development (Kwok et al. 2011).There are changes in synaptic strength during normal recoveryof the pyloric rhythm, but these changes do not appear to affectthe overall wiring pattern of the circuit, and are therefore notthought to be an important factor in recovery (Thoby-Brissonand Simmers 2002). For that reason, any interference in orexacerbation of these changes due to the absence of CSPGswould likely not prevent or delay recovery, unless they alsoresulted in large changes in structural plasticity or networkwiring. That scenario would likely result in nonfunctionalnetwork activity in both the intact and decentralized pyloriccircuit, which we did not observe. We take this as evidence thatstructural or synaptic changes due to chABC treatment are nota factor in the pyloric network, but further experiments mea-

suring the synaptic strengths of identified synapses before andafter treatment with chABC would inform this hypothesis.

We suggest that CSPG-mediated effects on neuronal excit-ability present a more likely mechanism of action in the pyloriccircuit. Although it is unclear how CSPGs mediate their effectson neuronal excitability, there are three categories of evidencefor potential mechanisms in other systems: cation buffering(Hrabetová et al. 2009; Vigetti et al. 2008), interaction withcell-surface receptors or channels (Maroto et al. 2013; Snow etal. 1994), and binding of growth factors for later presentationto cellular receptors (Deepa et al. 2002, 2004; Kantor et al.2004). The lack of a clear change in the ongoing pyloricrhythm in the intact STNS after chABC treatment suggests thation buffering is an unlikely mechanism in our system. Theother two possibilities are consistent with the lack of change inongoing activity and the delay in recovery metrics in thetreated pyloric circuit, if we assume that CSPGs are slowlyreplenished after chABC treatment, and will be an area forfuture experiments.

We have demonstrated the presence of CS in the crabnervous system and its involvement in activity maintenanceafter deafferentation. As described here, the STG is ideallysuited for further study of the mechanisms by which CSPGsmight be involved in activity maintenance and homeostasis.

ACKNOWLEDGMENTS

We thank Drs. Richard Cummings and Ravi Bellamkonda for assistanceand advice regarding the molecular biology experiments.

GRANTS

This work was supported by a National Science Foundation GraduateResearch Fellowship (A. E. Hudson) and by National Institute of NeurologicalDisorders and Stroke Grant R01 NS-054911 (A. A. Prinz).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: A.E.H. conception and design of research; A.E.H.,C.G., and J.-P.G. performed experiments; A.E.H. analyzed data; A.E.H., C.G.,J.-P.G., and A.A.P. interpreted results of experiments; A.E.H. prepared figures;A.E.H. drafted manuscript; A.E.H., C.G., and A.A.P. edited and revisedmanuscript; A.E.H., C.G., J.-P.G., and A.A.P. approved final version ofmanuscript.

REFERENCES

Balmer TS, Carels VM, Frisch JL, Nick TA. Modulation of perineuronalnets and parvalbumin with developmental song learning. J Neurosci 29:12878–12885, 2009.

Carulli D, Pizzorusso T, Kwok JC, Putignano E, Poli A, Forostyak S,Andrews MR, Deepa SS, Glant TT, Fawcett JW. Animals lacking linkprotein have attenuated perineuronal nets and persistent plasticity. Brain133: 2331–2347, 2010.

Deepa SS, Carulli D, Galtrey C, Rhodes K, Fukuda J, Mikami T, Suga-hara K, Fawcett JW. Composition of perineuronal net extracellular matrixin rat brain: a different disaccharide composition for the net-associatedproteoglycans. J Biol Chem 281: 17789–17800, 2006.

Deepa SS, Umehara Y, Higashiyama S, Itoh N, Sugahara K. Specificmolecular interactions of oversulfated chondroitin sulfate E with variousheparin-binding growth factors. Implications as a physiological bindingpartner in the brain and other tissues. J Biol Chem 277: 43707–43716, 2002.

Deepa SS, Yamada S, Zako M, Goldberger O, Sugahara K. Chondroitinsulfate chains on syndecan-1 and syndecan-4 from normal murine mammary

Rapid Report

1351CSPGs AND NEURONAL HOMEOSTASIS

J Neurophysiol • doi:10.1152/jn.00455.2015 • www.jn.org

gland epithelial cells are structurally and functionally distinct and cooperatewith heparan sulfate chains to bind growth factors. A novel function tocontrol binding of midkine, pleiotrophin, and basic fibroblast growth factor.J Biol Chem 279: 37368–37376, 2004.

Dityatev A, Brückner G, Dityateva G, Grosche J, Kleene R, Schachner M.Activity-dependent formation and functions of chondroitin sulfate-rich ex-tracellular matrix of perineuronal nets. Dev Neurobiol 67: 570–588, 2007.

Fongmoon D, Shetty AK, Basappa, Yamada S, Sugiura M, KongtawelertP, Sugahara K. Chondroitinase-mediated degradation of rare 3-O-sulfatedglucuronic acid in functional oversulfated chondroitin sulfate K and E. JBiol Chem 282: 36895–36904, 2007.

Galtrey CM, Fawcett JW. The role of chondroitin sulfate proteoglycans inregeneration and plasticity in the central nervous system. Brain Res Rev 54:1–18, 2007.

Giamanco KA, Matthews RT. Deconstructing the perineuronal net: cellularcontributions and molecular composition of the neuronal extracellular ma-trix. Neuroscience 218: 367–384, 2012.

Gogolla N, Caroni P, Luthi A, Herry C. Perineuronal nets protect fearmemories from erasure. Science 325: 1258–1261, 2009.

Golowasch J, Abbott LF, Marder E. Activity-dependent regulation ofpotassium currents in an identified neuron of the stomatogastric ganglion ofthe crab Cancer borealis. J Neurosci 19: RC33, 1999.

Golowasch J, Marder E. Proctolin activates an inward current whose voltagedependence is modified by extracellular Ca2�. J Neurosci 12: 810–817,1992.

Gutierrez GJ, Grashow RG. Cancer borealis stomatogastric nervous systemdissection. J Vis Exp 25: e1207, 2009.

Haedo RJ, Golowasch J. Ionic mechanism underlying recovery of rhythmicactivity in adult isolated neurons. J Neurophysiol 96: 1860–1876, 2006.

Hrabetová S, Masri D, Tao L, Xiao F, Nicholson C. Calcium diffusionenhanced after cleavage of negatively charged components of brain extra-cellular matrix by chondroitinase ABC. J Physiol 587: 4029–4049, 2009.

Hudson AE. Neuronal Mechanisms for the Maintenance of Consistent Behav-ior in the Stomatogastric Ganglion of Cancer borealis (PhD dissertation).Atlanta, GA: Georgia Institute of Technology, 2013.

Kantor DB, Chivatakarn O, Peer KL, Oster SF, Inatani M, Hansen MJ,Flanagan JG, Yamaguchi Y, Sretavan DW, Giger RJ, Kolodkin AL.Semaphorin 5A is a bifunctional axon guidance cue regulated by heparanand chondroitin sulfate proteoglycans. Neuron 44: 961–975, 2004.

Kazantsev V, Gordleeva S, Stasenko S, Dityatev A. A homeostatic model ofneuronal firing governed by feedback signals from the extracellular matrix.PLoS One 7: e41646, 2012.

Khorkova O, Golowasch J. Neuromodulators, not activity, control coordi-nated expression of ionic currents. J Neurosci 27: 8709–8718, 2007.

Kozlowski EO, Gomes AM, Silva CS, Pereira MS, Silva AC, Pavão MS.Structure and biological activities of glycosaminoglycan analogs frommarine invertebrates: new therapeutic agents? In: Glycans in Diseases andTherapeutics, edited by Pavão MS. Berlin: Springer, 2011, p. 159–184.

Kwok JC, Dick G, Wang D, Fawcett JW. Extracellular matrix and perineu-ronal nets in CNS repair. Dev Neurobiol 71: 1073–1089, 2011.

Kwok JC, Warren P, Fawcett JW. Chondroitin sulfate: a key molecule in thebrain matrix. Int J Biochem Cell Biol 44: 582–586, 2012.

Lander C, Kind P, Maleski M, Hockfield S. A family of activity-dependentneuronal cell-surface chondroitin sulfate proteoglycans in cat visual cortex.J Neurosci 17: 1928–1939, 1997.

Lee H, McKeon RJ, Bellamkonda RV. Sustained delivery of thermostabi-lized chABC enhances axonal sprouting and functional recovery after spinalcord injury. Proc Natl Acad Sci USA 107: 3340–3345, 2010.

Lin R, Kwok JC, Crespo D, Fawcett JW. Chondroitinase ABC has along-lasting effect on chondroitin sulphate glycosaminoglycan content in theinjured rat brain. J Neurochem 104: 400–408, 2008.

Luther JA, Robie AA, Yarotsky J, Reina C, Marder E, Golowasch J.Episodic bouts of activity accompany recovery of rhythmic output by aneuromodulator- and activity-deprived adult neural network. J Neurophysiol90: 2720–2730, 2003.

Maroto M, Fernández-Morales JC, Padín JF, González JC, Hernández-Guijo JM, Montell E, Vergés J, de Diego AM, García AG. Chondroitinsulfate, a major component of the perineuronal net, elicits inward currents,cell depolarization and calcium transients by acting on AMPA and kainatereceptors of hippocampal neurons. J Neurochem 125: 205–213, 2013.

Medeiros GF, Mendes A, Castro RA, Baú EC, Nader HB, Dietrich CP.Distribution of sulfated glycosaminoglycans in the animal kingdom: wide-spread occurrence of heparin-like compounds in invertebrates. BiochimBiophys Acta 1475: 287–294, 2000.

Pizzorusso T, Medini P, Berardi N, Chierzi S, Fawcett JW, Maffei L.Reactivation of ocular dominance plasticity in the adult visual cortex.Science 298: 1248–1251, 2002.

Pizzorusso T, Medini P, Landi S, Baldini S, Berardi N, Maffei L. Structuraland functional recovery from early monocular deprivation in adult rats. ProcNatl Acad Sci USA 103: 8517–8522, 2006.

Selleck SB. Genetic dissection of proteoglycan function in Drosophila and C.elegans. Semin Cell Dev Biol 12: 127–134, 2001.

Snow DM, Atkinson PB, Hassinger TD, Letourneau PC, Kater SB.Chondroitin sulfate proteoglycan elevates cytoplasmic calcium in DRGneurons. Dev Biol 166: 87–100, 1994.

Sugahara K, Yamada S. Structure and function of oversulfated chondroitinsulfate variants: unique sulfation patterns and neuroregulatory activities.Trends Glycosci Glycotechnol 12: 321–349, 2000.

Temporal S, Desai M, Khorkova O, Varghese G, Dai AH, Schulz DJ,Golowasch J. Neuromodulation independently determines correlated chan-nel expression and conductance levels in motor neurons of the stomatogas-tric ganglion. J Neurophysiol 107: 718–727, 2012.

Thoby-Brisson M, Simmers J. Neuromodulatory inputs maintain expressionof a lobster motor pattern-generating network in a modulation-dependentstate: evidence from long-term decentralization in vitro. J Neurosci 18:2212–2225, 1998.

Thoby-Brisson M, Simmers J. Long-term neuromodulatory regulation of amotor pattern-generating network: maintenance of synaptic efficacy andoscillatory properties. J Neurophysiol 88: 2942–2953, 2002.

Vargas J, De-Miguel FF. Growth-inhibiting extracellular matrix proteins alsoinhibit electrical activity by reducing calcium and increasing potassiumconductances. Neuroscience 158: 592–601, 2009.

Vigetti D, Andrini O, Clerici M, Negrini D, Passi A, Moriondo A.Chondroitin sulfates act as extracellular gating modifiers on voltage-depen-dent ion channels. Cell Physiol Biochem 22: 137–146, 2008.

Zhang Y, Golowasch J. Recovery of rhythmic activity in a central patterngenerator: analysis of the role of neuromodulator and activity-dependentmechanisms. J Comput Neurosci 31: 685–699, 2011.

Zhang Y, Khorkova O, Rodriguez R, Golowasch J. Activity and neuro-modulatory input contribute to the recovery of rhythmic output after decen-tralization in a central pattern generator. J Neurophysiol 101: 372–386,2009.

Rapid Report

1352 CSPGs AND NEURONAL HOMEOSTASIS

J Neurophysiol • doi:10.1152/jn.00455.2015 • www.jn.org