Sensory deafferentation modulates and redistributes neurocan in the rat auditory brainstem

Abstract Introduction Cochlear ablation causing sensory deafferentation (SD) of the cochlear nucleus triggers complex re‐arrangements in the cellular and molecular communication networks of the adult mammalian central auditory system. Participation of the extracellular matrix (ECM) in these processes is not well understood. Methods We investigated consequences of unilateral SD for the expression and distribution of the chondroitin sulfate proteoglycans, neurocan (Ncan) and aggrecan (Agg), alongside various plasticity markers in the auditory brainstem of the adult rat using immunohistochemical techniques. Results In the deafferented ventral cochlear nucleus (VCN), Ncan expression increased massively within 3 postoperative days (POD), but rapidly decreased thereafter. Agg showed a similar but less pronounced progression. Decrease in Ncan was spatially and temporally related to the re‐innervation of VCN documented by the emergence of growth‐associated protein Gap43 contained in nerve fibers and presynaptic boutons. Concurrently, astrocytes grew and expressed matrix metalloproteinase‐2 (MMP2), an enzyme known to emerge only under re‐innervation of VCN. MMP2 is capable of cleaving both Ncan and Agg when released. A transient modulation of the ECM in the central inferior colliculus on the side opposite to SD occurred by POD1. Modulations of glutamatergic synapses and Gap43 expression were detected, reflecting state changes of the surrounding tissue induced by transsynaptic effects of SD. Conclusions The ECM variously participates in adaptive responses to sudden deafness by SD on several levels along the central auditory pathway, with a striking spatial and temporal relationship of Ncan modulation to astrocytic activation and to synaptogenesis.


| S IG NIFIC AN CE OUTCOME S
• The extracellular matrix is variously involved in brain plasticity, but the details of its regulation are still scantly known.
• This study investigates neurocan and aggrecan upon sensory deafferentation in the auditory brainstem, showing a fast rise of their expression followed by a rapid decay.
• These reactive changes are locally and temporally associated with astrocytic growth, astrocytic expression of MMP2, and the emergence and maturation of Gap43 containing presynaptic profiles.
• Understanding the regulation of extracellular matrix in brain plasticity may be useful for future treatments of brain injury, such as introduction of matrix-cleaving enzymes.

| INTRODUC TI ON
Although the extracellular matrix (ECM) in the brain, particularly enriched in so-called perineuronal nets (PNNs), already caught the attention of Camillo Golgi and Santiago Ramón y Cajal in the 19th century (Celio, Spreafico, Biasi, & Vitellaro-Zuccarello, 1998), its significance for nervous function was accepted not before later in the 20th century.
Beginning by about 2 weeks after birth, ECM is restructured. Much of Ncan and versican isoforms V0 and V1 disappear from the PNNs, while Agg expression increases through the following 5 months, reaching a plateau that is maintained throughout adulthood (Milev et al., 1998).
PNNs are present throughout the central auditory system (Sonntag, Blosa, Schmidt, Rübsamen, & Morawski, 2015). CSPG immunoreactivity reaches a peak in the pontine auditory nuclei by about postnatal day 12 (Friauf, 2000), the time of hearing onset in the rat (Blatchley, Cooper, & Coleman, 1987). As sensory deprivation leads to an impairment of PNNs, for example, around neurons of the superior olivary complex (SOC) in deaf rats (Myers, Ray, & Kulesza, 2012) or in the visual cortex of dark-reared animals (Pizzorusso et al., 2002), sensory stimulation appears to be essential for proper PNN maturation. After termination of critical developmental periods, PNNs are thought to serve synaptic stabilization.
The motivation of this study was to investigate the involvement of ECM in the reorganization of the adult auditory brainstem, using a Cochlear ablation inflicts degeneration of the auditory nerve, leading to SD of the cochlear nucleus. This entails a massive reduction of excitatory glutamatergic input to the cochlear nucleus (Hildebrandt, Hoffmann, & Illing, 2011). This wrecking process is quickly accompanied by constructive activity involving functional re-innervation of the cochlear nucleus by axon collaterals originating from medial olivocochlear neurons of the ventral nucleus of the trapezoid body (Kraus & Illing, 2004). These in-growing fibers contain the growth-associated protein (Gap43), a marker for axonal growth and synaptogenesis (Benowitz & Routtenberg, 1997).
We focused on the distribution of CSPGs in PNNs of several auditory brainstem regions, studying their reactive dynamic and their relation to the changing populations of synaptic contacts. Motivated by its abundance in juvenile but not mature brains, we focused on Ncan and compared it to Agg which is abundant in PNNs of adult brains. If the ECM specifically contributes to the dynamic maintenance of neural net function, it was expected to respond in specific local and temporal relation to other cellular and molecular events that are already known to be involved into the constructive reorganization of the auditory brainstem upon SD.

| Sensory deafferentation
Tympanic membranes were checked for integrity and transparency bilaterally. Unilateral SD caused by cochlear ablation was invoked on the left side as described previously (Illing, Horváth, & Laszig, 1997). In short, the facial nerve was severed upon its exit from the skull. The bulla tympani was opened by removing tympanic membrane and stapes. The opening was widened for full visibility of the cochlea. The bony wall was perforated using a spherical drill, and the interior, including the spiral ganglion, was thoroughly cleared. Cochlea and middle ear cavity were subsequently filled with gel foam, and the skin was surgically closed. In the present study, SD was always done unilaterally. The lesioned and thus deafferented side will henceforth be referred to as ipsilateral.

| Brain preparation
After a predetermined postoperative survival period, rats received a lethal dose of sodium thiopental (Thiopental Inresa; Inresa, i.p. 50 mg/ml per 200 g body weight). At respiratory arrest, they were transcardially perfused first with Tyrode's solution to empty the circulatory system of blood and then with a fixation solution containing 4% paraformaldehyde (PFA) and 0.025% glutaraldehyde in 0.1 M phosphate buffer (PB, pH 7.4) for 75 min. Subsequently, brains were removed from the skull and the operated side was marked by a longitudinal incision. For immunostaining with 3,3′-diaminobenzidine tetrahydrochloride (DAB) and double or triple immunofluorescence staining, tissue was soaked overnight in 30% sucrose for cryoprotection at 4°C before being frozen and cut into 30-µm-thin frontal sections using a cryotome (CM3050 S, Leica, RRID: SCR_016844).  (Schnell, Staines, & Wessendorf, 1999) and mounted on gelatin-coated slides. We proceeded with air-drying overnight, followed by aqueous coverslipping with Mowiol 4-88 (Carl Roth) containing 2.5% triethylenediamine (DABCO, Carl Roth) to stabilize fluorochromes.

| Statistical evaluation
Gray tone ratios were determined, and across-brain comparisons were done of controls and all survival groups using Prism 8.1.0 (GraphPad Software, Inc., RRID: SCR_002798). We tested our data for normal distributions (using the Kolmogorov-Smirnov and Shapiro-Wilk test) and equal variances (using Brown-Forsythe and Bartlett's test). Though our data were normally distributed, the assumption of homogeneity of variances was violated in several cases. Therefore, we turned to Welch's ANOVA to determine significant differences between the means of staining ratios in the respective groups, as this test is robust against heteroscedasticity (Levy, 1978).
Multiple comparisons were submitted to Dunnett's T3 post hoc test.
In the figures, data are presented as scatter dot plot with means connected by line. Significance was set to p < 0.05, and significance levels are indicated as ***p < 0.001, **p < 0.01, or *p < 0.05. Statistical TA B L E 1 Antibodies used in the study

>0.999
Note: The upper half of the table shows statistic data for the ipsilateral-to-contralateral ratio in VCN for Ncan, Agg, Gap43, Gad65, and vGluT1. The lower half shows data of gray value comparisons in the right/contralateral VCN. Welch's ANOVA was used to determine whether means among groups differed significantly. Dunnett's T3 post hoc tests were applied. For each group, number of animals, mean, and SEM are indicated.
Abbreviations: W, W ratio of the Welch's ANOVA; DFn, degrees of freedom for numerator; DFd, degrees of freedom for denominator. Significance levels are indicated as ***p < 0.001, **p < 0.01, or *p < 0.05.  Table 2 for VCN and in Table 3 for CIC. In the figures, significance bars were only shown for the most relevant differences, with a complete listing provided in Tables 2 and 3.

| Electron microscopy
Following perfusion and brain preparation as described above, brains were postfixed in 4% PFA for 20 hr at 4°C. They were cut into 30-µm-thin frontal sections on a vibratome (VT 1000S, Leica, RRID: SCR_016495) and collected in 50 mM glycine in PBS. Freefloating sections were permeabilized for 10 min with 5% and 10% dimethyl sulfoxide and for 20 min with 20% and 40%. Following each step, sections were rinsed in PBS. Endogenous peroxidase activity was blocked with 0.045% H₂O₂ in PBS, and nonspecific binding sites were blocked with 5% goat serum in PBS each for 30 min. The first primary antibody (

| SD-dependent Ncan expression in the auditory brainstem
Upon unilateral SD, Ncan expression strongly increased in ipsilateral VCN and contralateral CIC (Figure 1).

| Modulation of Ncan Expression in VCN
Ncan expression was determined before and after SD (Figure 2). In control animals, Ncan staining in VCN was predominantly found in neuronal cell bodies (Figure 2a′ and b′). This situation changed drastically as a consequence of SD. By postoperative day 3  localized in PNN surrounding cell bodies (Figure 2c′, arrowheads).
Moreover, the pericellular space was enriched in Ncan. By POD7, Ncan expression was reduced and fragmented, leaving a darkly stained region medially adjacent to the vestibulocochlear nerve surrounded by tissue that was less intensely stained (Figure 2d).
At higher magnification, Ncan seemed largely removed from  (Table 2). Ncan ratio increased significantly by POD3, followed by a rapid decrease toward POD7. However, the staining ratios by POD7 and POD14 were still above controls.
To determine whether the changes of ratios were due to an ipsilateral increase or a contralateral decrease in staining intensity, gray tones were compared between controls to those in contralateral VCN of SD rats (Table 2, across-brain comparison). Dunnett's T3 test revealed no statistically significant changes between the groups for Ncan and all other markers that will be introduced in the following.
Obviously then, SD-dependent changes of expression in the VCN occurred ipsilaterally rather than contralaterally. Changes in staining ratios can therefore be read as changes in ipsilateral staining for all molecular markers.

| Temporal relation to plastic processes
These changes of ECM volume and composition were related to the concurrent SD-dependent synaptogenesis shown by Gap43 staining (Hildebrandt et al., 2011;Illing et al., 1997). In controls, Ncan and Agg, there was no significant difference in staining ratio F I G U R E 3 Agg and Gap43 in VCN. Agg(+) PNNs were present in controls (a, arrowheads). Their staining intensity rose by POD3 (b) and decreased toward POD7 (c). Gap43 immunoreactivity was hardly present in untreated adult brains (d) but rose toward POD3 (e). By POD3 and POD7, Gap43 mostly stained presynaptic boutons (arrows). Neuronal cell bodies are indicated by asterisks. Scale bar for a-f: 20 µm. Staining ratio of Agg (g) rose toward POD3 and declined thereafter. Gap43 staining ratio (h) was increased by POD3 and remained high by POD14 between POD3 and POD7 and between POD3 and POD14, indicating that the processes of axon growth and synaptogenesis were ongoing by 14 days after SD.

| Spatial relation of Ncan to synaptic populations
Aiming to define the effect of the temporal modulation of PNNs

| Upregulation of MMP2 in reactive astrocytes
Asking for the source of Ncan for its SD-induced emergence and the reason of quick Ncan degradation shortly thereafter, the spatial and temporal expression of Ncan, matrix metalloproteinase-2 (MMP2), and glial fibrillary acidic protein (GFAP) were put into relation ( Figure 5). In VCN of control brains, MMP2 staining was weak and largely confined to neuronal cell bodies but could also be found in the neuropil. Astrocytic fibers were rare and only faintly stained for MMP2. By POD3, although astrocytes were now massively activated and their processes thick and numerous (Fredrich et al., 2013), levels of MMP2 were not obviously altered. GFAP often appeared in close vicinity to the now prominently present Ncan(+) PNNs. Frequently, we found Ncan staining within GFAP(+) profiles ( Figure 5b inset; Supporting information). By POD7, MMP2 was seen to be enriched and redistributed, now often localized within GFAP(+) profiles (Figure 5c, arrows). This indicated an upregulation of MMP2 in astrocytes. By that time, Ncan has almost disappeared.

| Glial scar in facial nucleus
Before lesioning the spiral ganglion, the facial nerve was severed during the procedure of cochleotomy. Similar to ipsilateral VCN, enhanced GFAP expression was seen by POD3 in the affected facial nucleus compared to the unoperated side (see Supporting information). Importantly, no changes were detected in the expression of Ncan or Agg due to the lesion of the facial nerve.
Concurrently to the decrease in Ncan toward POD7, MMP2 has risen in thick astrocytic fibers (c'''), while it was mainly found in neuronal somata and the neuropil in controls (a''') and by POD3 (b'''). Neuronal cell bodies are indicated by asterisks. Scale bar for a-c: 20 µm; scale bar for inset: 5 µm  with the neuronal soma. This synaptic bouton neighbored to an astrocyte (purple) which contained noticeable amounts of Ncan. By POD4, MMP2 expression was mostly found in neurons, astrocytes, and neuropil ( Figure 6g) where it was prominently present in neuropil regions rich in Ncan (asterisks), thought to begin Ncan degradation.

| Changes in CIC
When VCN was deafferented by cochlear ablation, transsynaptic effects developed in CIC. A significant change in Ncan staining was detected by POD1 (Figure 7a). By that time, modulations of expression were also observed for other molecular markers here employed (Figure 7b-d). Statistical data are shown in Table 3. DAB images of the ipsilateral CICs were taken from sections of control brains and all survival groups, normalized for background brightness, and submitted to statistical analysis ( comparison). Dunnett's T3 post hoc test revealed no significant differences, indicating that the changes observed are due to the modulation of staining contralaterally, that is, on the side primarily affected by SD due to decussation of auditory pathways, rather than ipsilaterally.
Contralateral-to-ipsilateral staining ratios for Ncan (Figure 7 a)  Looking closer into the brief increase in staining intensity of Ncan in CIC contralateral to SD by POD1 (Figure 7f-i), we found that well-defined PNNs have emerged by POD1 that was absent before.
Bilateral differences in staining for vGluT1 and Gap43 by POD1 detected by averaging of gray values (Figure 7c,d) are too subtle in photographs of fluorescent staining to be appreciable. Conspicuously, Gap43 is present in a population of well-defined boutons in CIC of the adult rat independent of SD. Gap43 was also found in CIC of control animals, which was suggested to imply conserved readiness for plastic dynamics rather than active axonal growth.
Gad65 expression (Figure 7 e) was modulated differently. There was no significant change in staining ratio between control and POD1, but a decrease has occurred by POD3. This asymmetry continued to POD7 and POD14. It seems to indicate that GABAergic inhibition in signaling networks of the contralateral CIC is persistently weakened due to SD-induced deafness.

| NCAN in DCN, LSO, and MGB
Compared to VCN, the texture of Ncan staining in the dorsal cochlear nucleus (DCN) was different (Figure 8a,b). However, the SD-induced temporal progression with an ipsilateral upregulation of Ncan in PNNs (arrowheads) by POD3 and a subsequent decline by POD7 was similar, although much less pronounced.
F I G U R E 7 Effects in CIC. Graphs (a-e) show analysis of contralateral(c)-toipsilateral(i) staining ratio of mean gray values in CIC for all survival groups and 5 molecular markers. Means are connected by line. Dashed lines indicate symmetry between ipsilateral and contralateral sides. Ncan (a), Agg (b), Gap43 (c), and vGluT1 (d) all showed transient peaks of ratios by POD1. Only Gad65 (e) failed to show an effect by POD1 but decreased by POD3 to remain low. In fluorescence double labeling at POD1 in the CIC (f-i), Ncan staining dominated contralaterally and was mostly localized in PNNs (g, f, arrowheads). Staining differences for the synaptic markers vGluT1 and Gap43 were too subtle to be discernible in selected photographs but emerged by sampling from DAB-stained sections (c, d). Synaptic boutons are indicated by arrows. Neuronal cell bodies are indicated by asterisk. Scale bar for f-i: 20 µm Within the superior olivary complex, differences in Ncan staining between the ipsilateral and contralateral side were not noted at any survival time (Figure 8c,d). Images at higher magnification (insets) show the presence of Ncan in PNNs (arrowheads) independent of

SD.
We failed to find definite staining for Ncan in the MGB in control brains, and this situation did not change at any time after SD on either side (Figure 8e,f). However, PNNs were recognizable by their contents of Agg (Sonntag et al., 2015). Dense Agg staining was present in ventral MGB, but PNNs were hardly prominent dorsally.
As for Ncan, Agg expression in the MGB remained unaffected by SD.

| D ISCUSS I ON
The major findings of this study are as follows: (a) Following SD, Ncan

| Lesion-dependent NCAN expression
Upregulation of Ncan expression in the adult brain as a consequence of lesions to the CNS has been reported before. It was observed after traumatic injuries of neocortex (Asher et al., 2000;McKeon et al., 1999), entorhinal cortex (Haas et al., 1999), the nigrostriatal pathway (Moon, Asher, Rhodes, & Fawcett, 2002), the spinal cord (Jones, Margolis, & Tuszynski, 2003;Tang, Davies, & Davies, 2003), and after kainateinduced seizures (Matsui et al., 2002) or focal ischemia (Deguchi et al., 2005). Full-length Ncan is found in the injured tissue (Asher et al., 2000), which is also recognized by the antibody used in the present study (Bekku & Oohashi, 2010). Whereas this form is abundant during development, only fragments of Ncan are found in the healthy adult brain Meyer-Puttlitz et al., 1995;Rauch et al., 1991). Apparently, neuro-and glioplastic changes induced by brain injuries like the ones seen in VCN after SD re-invoke Ncan expression reminiscent of ontogenetic dynamics.

| Reactive astrocytes synthetized NCAN
During maturation of the mammalian CNS, Ncan is predominantly of neuronal origin (Engel, Maurel, Margolis, & Margolis 1996). By contrast, a lesion-dependent rise of Ncan-mRNA and Ncan-protein is almost exclusively generated by reactive astrocytes (Asher et al., 2000;Deguchi et al., 2005;Haas et al., 1999;Jones et al., 2003;Matsui et al., 2002;McKeon et al., 1999). Our results are compatible with these reports as Ncan was found inside as well as immediately around astrocytes recognized by GFAP (light microscopy: Figure 5b inset; electron microscopy: Figure 6f; Supporting information). Following SD, GFAP intensity rises quickly and massively in VCN (Fredrich et al., 2013;Janz & Illing, 2014), indicating activation of these cells. Colocalization of Ncan and GFAP was most prominent by POD3, but still recognizable by POD7. We found no evidence for other than astrocytic origin of Ncan.

| Ncan modulation corresponded spatially and temporally to re-innervation
An outstanding aspect of SD by cochlear ablation is the fast, synchronous, and complete loss of sensory axons (Gentschev & Sotelo, 1973), followed by a broad synaptogenesis in the deafferented VCN (Hildebrandt et al., 2011;Illing et al., 1997). This re-innervation, visualized by Gap43 immunoreactivity, becomes prominent by POD3. A previous study (Kraus & Illing, 2004)  Since all Gap43 synapses are cholinergic (Meidinger, Hildebrandt-Schoenfeld, & Illing, 2006), Gad65 synapses cannot be part of the process of synaptogenesis and must have been present before SD.
Under the electron microscope, Gap43(+) profiles were found close to Ncan-rich PNNs by POD3, mostly failing to form contact with the neuron (Figure 6b). By contrast, morphologically mature presynaptic endings containing Gap43 were often attached to the plasma membrane toward POD7, now morphologically indistinguishable from Gad65(+) presynaptic boutons. Ncan has been described as growth impermissive (Asher et al., 2000;Friedlander, 1994;Katoh-Semba, Matsuda, Watanabe, Maeda, & Oohira, 1998). Our observations are suggestive of an Ncan degradation in VCN as a precondition for synaptic maturation to occur.

| Rebalancing of synaptic networks following SD
Confirming a previous study (Hildebrandt et al., 2011), we noted an SD-related loss of glutamatergic endings in VCN, which is significant by POD3 and later (Figure 4j). Accompanying the loss of glutamatergic endings upon SD, the inhibitory synaptic contacts identified by gephyrin decline in number by 30%. However, the fraction of inhibitory synapses of all synaptic contact zones in VCN after SD dominates excitatory contacts (Hildebrandt et al., 2011). Here, we report a rise of Gad65 staining intensity on the affected side which we attribute to an upregulation of Gad65 in pre-existing, stable synaptic contacts. The fact that Gad65(+) presynaptic endings seem embedded and stabilized by PNNs rich in Agg and transiently also in Ncan seems consistent with their metabolic rather than structural modulation.
Moreover, changes of MMP2 expression after SD are conditional upon an effective re-innervation revealed by Gap43 immunoreactivity (Fredrich & Illing, 2011). A release of MMP2 from astrocytes upon the approach of growing neurites and synaptogenesis locally and temporally coincides with the degradation of Ncan from PNNs in VCN.

| AGG changes were moderate
Although Ncan and Agg are differently regulated in early ontogeny, they changed in a similar direction in our experimental model. Unlike Ncan, Agg is abundant in VCN of adult control brains, making Agg a widely present component of PNNs. Following SD, the expression of Agg in VCN changed only mildly but in the same vein as Ncan. This suggests an involvement of Agg in plastic processes as has been previously reported (Harris, Carmichael, Hovda, & Sutton, 2009;Lemons, Sandy, Anderson, & Howland, 2001;McRae, Rocco, Kelly, Brumberg, & Matthews, 2007;Ueno et al., 2018;Ye & Miao, 2013). As Ncan, Agg is susceptible to dismounting by MMP2 (Overall, 2002).

| Putative functions of CSPGS in reorganization of the VCN
We suggest that Ncan and Agg are crucial components in the regulation of SD-dependent re-innervation in VCN. While several studies point to the growth impermissive properties of CSPGs (Asher et al., 2000;Friedlander, 1994;Katoh-Semba et al., 1998), others suggest supporting effects of CSPGs on axon outgrowth (Haas et al., 1999;Maeda & Noda, 1996). Recently, it has been shown that the potential of CSPGs to support neuroplasticity depends on their sulfation pattern and decreases with age (Foscarin, Raha-Chowdhury, Fawcett, & Kwok, 2017). In our experimental model, the rise and fall of Ncan and Agg might indicate an effort for emergency stabilization of synaptic contacts before neurite outgrowth triggers astrocytes to express and release MMP2 that degrades CSPGs in order to provide access for growth cones to form new synapses. Cutting efferent axons originating in the facial nucleus nerve led to glial scar formation but did not change levels of Ncan or Agg. In contrast to SD of the VCN, retrograde degeneration of the facial nerve did neither cause loss of synaptic input to the facial nucleus nor the formation of new synaptic contacts.

| Astrocytes are key players in the SDdependent reorganization of VCN
Apart from their involvement in the management of tissue degeneration and glial scar formation, astrocytes also participate in recovery processes after brain lesions. Preventing astrocyte activation has negative effects on tissue reorganization. Unilateral cerebral stroke of the motor cortex in mice normally induces axons of the contralesional corticospinal tract to sprout into the denervated spinal cord, contributing to motor functional recovery (Liu, Li, Zhang, Savant-Bhonsale, & Chopp, 2008). When astrogliosis is attenuated by double knockout of GFAP and vimentin in mice that then underwent stroke of motor cortex, corticospinal axons only rarely crossed the midline and their length was significantly reduced compared to the axons in wild-type mice. The reduced astrocytic reactivity leads to impaired neurological recovery (Liu et al., 2014). In our SD model, astrocytes in VCN are activated, grow, branch, and express PSA-NCAM, MMP2, and MMP9 (Fredrich et al., 2013). Their expression of PSA-NCAM and MMP2 is conditional upon the arrival of Gap43(+) nerve fibers (Fredrich et al., 2013). Here, we show that, shortly after SD, they also synthetize Ncan and then engage in MMP2 expression. This succession suggests being causal to the degradation of Ncan through the released MMP2, thus providing access for neurites to form new synaptic contacts.

| Transient transsynaptic reaction of the contralateral IC on POD1
By POD1, expression of Ncan, Agg, Gap43, and vGluT1 was found to be increased in the contralateral CIC that does not receive auditoryborn signals directly by cochlear axons. The quick responses were remarkable given their remoteness to degenerational processes.
The molecular modulations seen in CIC were transient. Compared to VCN, changes were mild but significant. A quick but brief effect was the same for all molecular markers employed except for Gad65. The level of Gad65 staining decreased in contralateral CIC and remained low, indicating a lasting effect on network activity (Mossop, Wilson, Caspary, & Moore, 2000;Vale, Juíz, Moore, & Sanes, 2004). Taken together, fast molecular adaptations occurred in CIC as a consequence to cochlear lesioning, some transient and some sustained, which together indicate remarkable quick and specific transsynaptic modifications of signal processing in the auditory midbrain. In chinchillas, about 50% of collicular neurons significantly increase their spiking rate upon acoustic trauma, an effect that the authors associate with loss of inhibition (Wang, Ding, & Salvi, 2002). This enhancement shows up as quickly as by 8h post-trauma. Indeed, a short-term rise of excitability in CIC neurons upon hearing loss has been demonstrated before (McAlpine, Martin, Mossop, & Moore, 1997;Popelár, Erre, Aran, & Cazals, 1994;Salvi, Saunders, Gratton, Arehole, & Powers, 1990;Szczepaniak & Møller, 1996;Willott & Lu, 1982). These electrophysiological studies give no clue about underlying molecular correlates. CSPGs were reported to protect cortical and hippocampal neurons from excitotoxicity (Okamoto, Mori, Ichimura, & Endo, 1994). In the rat hippocampus, neuronal hyperactivity induced by electrical stimulation induces astrocytic expression of Ncan (Schwarzacher et al., 2006). A protection of collicular neurons against overexcitation could be reflected by the growth of Ncan-enriched PNNs as reported here.

ACK N OWLED G M ENTS
We thank Katrin Kaehlke for technical assistance, Anna Schädler for participation in initial stages of this project, Günther Schlunck for access to the confocal microscope, and Roland Laszig for continuous support.

CO N FLI C T O F I NTE R E S T
The authors disclose no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The materials used in this study are made fully explicit and are available from public suppliers. Brain sections, protocols, and measurement data on which this study is based are fully archived and accessible in the laboratory library of our department.