Neuroscience 238 (2013) 168–184

PERINEURONAL AND PERISYNAPTIC EXTRACELLULAR MATRIX

IN THE HUMAN SPINAL CORD
C. JÄGER, a,b D. LENDVAI, c G. SEEGER, a containing link-protein and brevican in all regions except
G. BRÜCKNER, a R. T. MATTHEWS, d T. ARENDT, a of the Lissauer’s zone. Intersegmental differences were
A. ALPÁR c,e* AND M. MORAWSKI a,d* reflected in the appearance of segment-specific nuclei but
a
Paul Flechsig Institute of Brain Research, University of Leipzig, not in overall matrix distribution pattern or chemical hetero-
Jahnallee 59, 04109 Leipzig, Germany geneity. Perineuronal nets were typically associated with
b long-range projection neurons including cholinergic ventral
Fraunhofer Institute for Cell Therapy and Immunology,
Perlickstraße 1, 04103 Leipzig, Germany horn motorneurons or dorsal spinocerebellar tract neurons
c
of the Clarke–Stilling nuclei. Multiple immunolabelling
Department of Anatomy, Histology and Embryology,
revealed that nociceptive afferents were devoid of individual
Semmelweis University, Budapest, Hungary
matrix assemblies unlike glycinergic or GABAergic syn-
d
Department of Neuroscience and Physiology, State University apses. The detailed description of ECM distribution in the
of New York Upstate Medical University, Syracuse, NY 13210, USA human spinal cord shall support clinical approaches in
e
Division of Molecular Neurobiology, Department of Medical injury and regenerative therapy. Ó 2013 IBRO. Published
Biochemistry & Biophysics, Karolinska Institutet, Stockholm, Sweden by Elsevier Ltd. All rights reserved.

Abstract—Extracellular matrix (ECM) forms an active inter- Key words: aggrecan, brevican, axonal coats, motorneurons,
face around neurons of the central nervous system (CNS). pain fibres, extracellular matrix.
Whilst the components, chemical heterogeneity and cellular
recruitment of this intercellular assembly in various parts of
the brain have been discussed in detail, the spinal cord
received limited attention in this context. This is in sharp INTRODUCTION
contrast to its clinical relevance since the overall role of
ECM especially that of its chondroitin sulphate-based pro- Extracellular matrix (ECM) is an essential part of the
teoglycan components (CSPGs) was repeatedly addressed central nervous system (CNS) assembly that fills the
in neuropathology, regeneration, CNS repair and therapy intercellular space as an amorphous substance. A
models. Based on two post-mortem human specimen, this specialised part of the ECM forms dense and distinct
study gives the first and detailed description of major ECM structures around certain populations of neurons or
components of the human spinal cord. Immunohistochemi- synapses, called perineuronal nets (PNs; for reviews
cal investigations were restricted to the systematic mapping see Celio and Blümcke, 1994; Celio et al., 1998;
of aggrecan, brevican, proteoglycan link-protein as well as
Dityatev et al., 2010; Kwok et al., 2011; Morawski et al.,
tenascin-R and hyaluronan containing matrices in the whole
cranio-caudal dimension of the human spinal cord. Other 2012a,b) or the recently described axonal coats (ACs;
proteoglycans like versican, neurocan and NG2 were Brückner et al., 2008; Morawski et al., 2012b; Lendvai
exemplarily investigated in restricted areas. We show the et al., 2012, 2013), respectively.
overall presence of tenascin-R and hyaluronan in both white PNs were first described at the end of the 19th century
and grey matters whereas aggrecan, proteoglycan link- (Golgi, 1882, 1893, 1898; Lugaro, 1895; Ramón y Cajal,
protein and brevican were restricted to the grey matter. In 1897; Donaggio, 1898) with earliest illustrations from the
the grey matter, the ECM formed aggrecan-based perineuro- ventral horn of the spinal cord of the cat (Golgi, 1898)
nal nets in the ventral and lateral horns but established and dog (Donaggio, 1898). They form around the
single perisynaptic assemblies, axonal coats (ACs), somatic, proximal dendritic parts and axon initial
segment of the neurons and occur in different vertebrate
*Correspondence to: M. Morawski, Paul Flechsig Institute of Brain species (Ohyama and Ojima, 1997; Brückner et al.,
Research, Universität Leipzig, Jahnallee 59, 04109 Leipzig, 1998a, 2006; Szigeti et al., 2006; Morawski et al., 2009,
Germany. A. Alpár, Department of Anatomy, Histology and Embry-
ology, Semmelweis University, Budapest, Hungary. 2010b) including human (Brückner et al., 1993, 1996;
E-mail addresses: alan.alpar@ki.se (A. Alpár), morm@medizin. 2008; Morawski et al., 2010c). Their exact role remains
uni-leipzig.de (M. Morawski). enigmatic with possible impacts upon formation
These authors contributed equally to this work. (Bandtlow and Zimmermann, 2000; Dino et al., 2006)
Abbreviations: AC, axonal coat; B-HABP1, biotinylated hyaluronan-
binding protein; ChAT, choline acetyl transferase; CNS, central
and stabilisation (Hockfield and McKay, 1983;
nervous system; CSPGs, chondroitin sulphate proteoglycans; CS- Pizzorusso et al., 2002; Berardi et al., 2003; Dityatev
GAG, chondroitin sulphated glycosaminoglycan; ECM, extracellular and Schachner, 2003; Rhodes and Fawcett, 2004) of
matrix; GABA, gamma-aminobutyric acid; HAS, hyaluronan synthase; synaptic contacts, regulation of local ion homoeostasis
NeuN, neuronal nuclear protein; PBS, phosphate-buffered saline;
PMD, post-mortem delay; PN, perineuronal net; SCI, spinal cord (Brückner et al., 1993, 1996; Härtig et al., 1999; Reinert
injury; WFA, Wisteria floribunda lectin; VVA, Vicia villosa lectin. et al., 2003; Morawski et al., 2004) or neuroprotection

0306-4522/13 $36.00 Ó 2013 IBRO. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.neuroscience.2013.02.014
168
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 169

(Morawski et al., 2004, 2010a,c, 2012a,b; Wu et al., cord. The present study gives a systematic description of
2005). Molecular mechanisms corroborating these the aggrecan- and brevican-based ECM of the human
functions include the blockade of transmitter spillover spinal cord. We show at representative cervical, thoracic,
(Vargová and Syková, 2008; Dityatev et al., 2010) or lumbar and sacral segments the typical matrix patterns,
lateral receptor diffusion (Frischknecht et al., 2009). In the distribution of PNs and ACs and their relation to
contrast to the massive assembly of PNs, ACs are different neuronal and synaptic subtypes. The detailed
small, round or oval structures which occasionally line description of the ECM properties and distribution in the
up to pearl-lace-like structures and enwrap individual human spinal cord may support future therapies.
synapses (Brückner et al., 2008; Morawski et al., 2012b).
Depending on age, region or type of the surrounded EXPERIMENTAL PROCEDURES
neurons and synapses, components of the matrix show
considerable diversity and specificity due to the wide
Removal, perfusion, tissue preparation and profile of
array of molecules which load the intercellular space.
cases
Chondroitin sulphate proteoglycans (CSPGs) are
eminent representatives of the adult matrix assembly. A Human tissue was collected at standard necropsy from patients
pivotal role is taken by aggrecan, the major component at the Second Department of Pathology of Semmelweis
of the ECM of PNs (Köppe et al., 1997; Brückner et al., University (Budapest, Hungary), with family consent or as
1998b, 2000; Yamaguchi, 2000; Matthews et al., 2002; medico legal cases. Removal of human tissue and subsequent
Rauch, 2007) and brevican, the major component of the preparation were in accordance with the ethical guidelines of
ECM of ACs (Brückner et al., 2008; Morawski et al., Semmelweis University. Two human spinal cords (SC) were
removed from male patients without any sign of CNS-related
2012a,b). CSPGs are attached to a hyaluronan degeneration, age below 60 years (55 and 58 years) and a
backbone which is continuously secreted by a neuronal post-mortem delay (PMD) of 12 and 15 h. The perfusion was
hyaluronan synthase (HAS3; Kwok et al., 2010); and performed in situ via the internal carotid, vertebral arteries and
their connections are stabilised via link proteins (LP1–4; via intradural space. The perfusion was started with saline
Neame and Barry, 1993; Bekku et al., 2003; Spicer (0.9% NaCl), followed by a fixative containing 2%
et al., 2003; Carulli et al., 2006, 2007, 2010; Kwok paraformaldehyde and 2% glutaraldehyde in 0.1 M Tris-buffered
saline (TBS). The spinal cords were cut into vertebral segments
et al., 2010). Additionally, CSPGs are stabilised by the
and immersion-postfixed in a fixative of 4% paraformaldehyde
small glycoprotein tenascin-R (Brückner et al., 2000; solution for 72 h. After fixation, 15-mm-thick slices were
Dityatev and Schachner, 2003) forming a quaternary prepared in the transversal plane (according to Clara, 1959;
macromolecular complex in the direct microenvironment Standring, 2008). Tissue blocks containing the regions of
of the PN-/AC-ensheathed neurons or synapses. The interest (each second segment of cervical, thoracal, lumbal,
expression of aggrecan and brevican as well as the and sacral regions) were cryoprotected in 30% sucrose in
0.1 M phosphate-buffered saline, pH 7.4 (PBS). Series of 40-
formation of PNs and ACs is coincident with a certain
micrometre-thick sections were cut on a freezing microtome
level of reduced plasticity in the brain (Bradbury et al., and collected in PBS containing 0.1% sodium azide.
2002; Pizzorusso et al., 2002, 2006; Massey et al.,
2006; Galtrey et al., 2008; Garcı́a-Alı́as et al., 2009).
Identification of anatomical regions and applied
The clinical relevance of perineuronal matrix
nomenclature
assemblies and their major components, CSPGs, was
repeatedly addressed in neuropathology, regeneration,
Anatomical regions were identified on Nissl-stained and anti-
CNS repair and therapy models (for reviews see neuronal nuclear protein (NeuN) marked sections adopting the
Viapiano and Matthews, 2006; Busch and Silver, 2007; nomenclature of brain regions from Clara (1959) and Standring
Crespo et al., 2007; Galtrey and Fawcett, 2007; Garcı́a- (2008).
Alı́as and Fawcett, 2012; Bartus et al., 2011; Kwok
et al., 2011). In the spinal cord, ECM received attention Cytochemistry
mostly through its inhibitory role in regeneration after
injury (Busch and Silver, 2007; Pizzi and Crowe, 2007; Forty-micrometre sections were postfixed with 4% PFA in PBS
Fitch and Silver, 2008). Accordingly, partial removal of for 30 min. Treatment with 2% H2O2 in 60% methanol for
the CSPGs via chondroitinase ABC digestion is known 60 min to abolish endogenous peroxidase activity was followed
to enhance recovery after spinal cord injury and to by rinsing with phosphate-buffered saline/Tween (PBS-T)
(0.05% Tween). In some cases an additional pre-treatment
improve recovery of motor and sensory deficits
step by incubating the sections in sodium borohydride 1% in
(Bradbury et al., 2002; Barritt et al., 2006; Massey et al., PBS for 30 min at room temperature was performed to reduce
2006; Galtrey et al., 2008; Garcı́a-Alı́as et al., 2009; the potential GA induced auto fluorescence as well as
Alilain et al., 2011; Bradbury and Carter, 2011). potentially enhancing antigen accessibility (Romeis, 2010). A
Morphological data on the ECM properties in the spinal subsequent blocking step with 2% bovine serum albumin, 0.3%
cord were quite fragmentary until now. Although recent casein and 0.5% donkey serum in PBS-T was carried out for
advances gave fundamental details about the distribution 30 min at room temperature to prevent non-specific antibody
binding. Sections were incubated over one or two nights with
and phenotypic appearance of CSPG-immunoreactive primary antibodies, diluted in the same blocking solution.
ECM in the adult rat spinal cord (Vitellaro-Zuccarello Immunoreactivity was performed with biotinylated secondary
et al., 2007; Galtrey et al., 2008), to the best of our antibodies (donkey anti-mouse, donkey anti-rabbit, donkey anti-
knowledge there are no data available about the goat; Dianova, Hamburg, Germany) and ExtrAvidinÒ
structure and distribution of the ECM in the human spinal peroxidase complex (Sigma, Munich, Germany) visualised by
170 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Diaminobenzidine (brown) or nickel-enhanced Diaminobenzidine product (Brevican; Matthews et al., 2000; Giamanco et al.,
(black) as chromogen. For double and triple fluorescent labellings 2010), proteoglycan link-protein 1 (HAPLN-1; CRTL-1; Neame
immunoreactivity was visualised with Carbocyanine 2, 3 or 5 and Barry, 1994; Carulli et al., 2007, 2010; Giamanco et al.,
coupled secondary antibodies (Dianova). To reduce the effect 2010) and tenascin-R (Brückner et al., 2003). Additionally,
of autofluorescent lipofuscin, fluorescence-labelled slices were hyaluronan was detected with the biotinylated hyaluronan-
additionally treated with Sudan Black (Schnell et al., 1999). binding protein (B-HABP; Brückner et al., 1998b; Carulli et al.,
Generally, in control experiments primary antibodies were 2007; Morawski et al., 2010a). To test the specificity of the
omitted, yielding the unstained sections. B-HABP binding, free-floating sections were pretreated with
hyaluronidase (Hyase) from Streptomyces hyalurolyticus (50 U/
ml 0.1 M PBS, pH 5.0; Sigma, H1136) (50) for 4 h, 8 h and
Detection of ECM components 12 h at 37 °C. Binding of B-HABP was at the background level
after enzymatic treatment of sections. Visualisation of ECM
For the investigation of the ECM in the human spinal cord tissue with biotinylated lectins which are known to mark N-acetyl-
antibodies against several matrix components were used (for galactosamin in ECM constituents (Wisteria floribunda lectin
detailed description of the antibodies see Table 1). The main (WFA; Brückner et al., 1993; Härtig et al., 1992; Seeger et al.,
ECM components we investigated were aggrecan (Agg; 1994, 1996) and Vicia villosa lectin (VVA; Kobayashi et al.,
Brückner et al., 1999; Morawski et al., 2010a,c; Virgintino 1989; Murakami et al., 1999) was not possible due to matrix
et al., 2009), brevican, BEHAB/brevican 50 kD cleavage degradation because of the long PMD (Morawski et al., 2012b).

Table 1. Cytochemical markers used for detection of extracellular matrix components, neurons, characterisation of neurons and glial cells

Detected components Antibodies and binding proteins Dilution Source Lot number References

Matrix constituents
N-acetyl-galactosamine Biotinylated Wisteria floribunda 5 mg/ml Sigma #26F-8100 Härtig et al. (1992)
agglutinin
N-acetyl-galactosamine Biotinylated Vicia villosa agglutinin 5 mg/ml Vector Labs # M0327 Seeger et al. (1996)
Aggrecan, core protein Mouse anti-human aggrecan 1:10 Acris/Serotec #140510 Brückner et al. (2008)
(HAG7D4)
Brevican (50 kD fragment) Rabbit anti-human brevican 1:2000 R.T. Matthews #B50 Matthews et al. (2000)
(B50)
Link protein 1 Goat anti-human CRTL-1 1:400 R&D Systems #VBN 0107081 Carulli et al. (2007)
(HAPLN-1)
Hyaluronan Biotinylated hyaluronic acid-binding 10 mg/ml Cape Cod #D 00100295 Carulli et al. (2007)
protein (B-HABP)
Tenascin-R Mouse anti-bovine tenascin-R 1:50 R&D Systems #IZG02 Brückner et al. (2003)
(619)
Versican Mouse anti-human versican 1:25 DSHB Iowa Galtrey et al. (2008)
(12C5)
Neurocan Sheep anti-mouse neurocan 1:100 R&D Systems #CDGF0110011 Bekku and Oohashi (2010)
(AF5800)
NG2 Rabbit anti-rat NG2 CSPG 1:200 Millipore #2031329 Andrews et al. (2012)
(AB5320)

Cellular markers
Neuron-specific nuclear Mouse anti-NeuN MAB 377 1:100 Millipore #0507004415 Poirier et al. (2010)
protein clone A60
Parvalbumin Rabbit anti-rat PV28 1:1000 Swant #5.5 Celio and Heizmann
(1998)

Markers for characterisation of neuronal populations

GabaA receptor Mouse anti-GabaA receptor 1:200 Millipore #LV1424894 Ma et al. (2011)
ß2,3 chain
Choline acetyltransferase Goat anti-Chat AB144 1:500 Millipore #25030685 Morawski et al. (2010c)
Glycine receptor Mouse anti-glycine receptor 1:100 Synaptic #146011/7 Baer et al. (2009)
clone mAb4a (GlyR4a) Systems
Substance P Rabbit anti substance P 1:100 Accurate #H9887 Kozsurek et al. (2009)
Chemical
Vglut-1 Rabbit anti-rat vesicular 1:500 Synaptic #135003 Takamori et al. (2000)
glutamate transporter 1 Systems
Vglut-2 Rabbit anti-rat vesicular 1:500 Synaptic #135402 Herzog et al. (2006)
glutamate transporter 2 Systems

Glia markers
Ionised calcium-binding Rabbit anti-Iba1 1:500 Wako #CDQ5232 Kanazawa et al. (2002)
protein 1
Glial fibrillary acidic protein Rabbit anti-GFAP 1:1000 Dako #Z033401096302 Morawski et al. (2010c)
(GFAP)
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 171

Detection and characterisation of neuronal showed largely spared hyaluronan (Fig. 1C2–C4) but not
populations and glial cells tenascin-R immunoreactivity (Fig. 1D2–D4) in the dorsal
horn, especially in its thoracic, lumbar and sacral
To detect neurons and specific neuronal subpopulations the anti- segments.
NeuN antibody and the anti-parvalbumin (Parv) antibody were Aggrecan, brevican and HAPLN-1 were detected
used (PV 28; Celio and Heizmann, 1981; Härtig et al., 1995;
predominantly in matrix aggregates around neuronal
Seeger et al., 1996). For further characterisation of neuronal
populations surrounded by PNs antibodies against gamma- somata, dendrites and terminals. Since these structures
aminobutyric acid (GABA) A receptor (GABA A ß2,3 chain; Ma are concentrated in the grey matter, distinct ECM
et al., 2011), choline acetyl transferase (ChAT; Morawski et al., structures were detected in these regions leaving the
2010c), glycine receptor (GlyR4a; Baer et al., 2009), vesicular white matter practically spared from aggrecan, brevican
glutamate receptors 1 and 2 (Takamori et al. 2000; Herzog and HAPLN-1 immunoreactivity (Figs. 2 and 3).
et al., 2006) and substance P (Kozsurek et al., 2009) were Neurocan and versican are more ubiquitously distributed
used. For detailed description of markers see Table 1.
To investigate glial cells in the human spinal cord we used
throughout the grey matter similar to hyaluronan and
antibodies to ionised calcium-binding protein (Iba1) in microglia tenascin-R and do not seem to be enriched in special
(Kanazawa et al., 2002) and glial fibrillary acidic protein (GFAP) nuclei (Fig. 4); in detail versican is associated to the
in astrocytes (Morawski et al., 2010c). For detailed description nodes of Ranvier (Fig. 4A2) and neurocan occasionally
of markers see Table 1. delineates PN ensheathed neurons (Fig. 4B1–2). We
reveal that PNs show a partial overlap of their basic
Light microscopy, confocal laser scanning ECM components: aggrecan delineates neuronal
microscopy and image processing somata and dendrites with different intensities whereas
brevican and HAPLN-1 widely overlap around more
Tissue sections were examined with a Zeiss Axiovert 200 distal dendritic compartments as well (Fig. 5). Finally,
Microscope equipped with a motorised stage (Märzhäuser, multiple labelling experiments gave basic information on
Wetzlar, Germany) with MosaiX software and by means of a the relation of ECM to neuronal, transmitter and
CCD camera (Zeiss MRC) connected to an Axiovision 4.6 receptor subtypes in the human spinal cord.
image analysis system (Zeiss, Germany) for light microscopy.
Fluorescence labelling was examined with a Zeiss confocal
laser scanning microscope (LSM 510, Zeiss, Jena, Germany). Cervical segment
For secondary Cy2-labelled slices (green fluorescence), an
argon laser with 488-nm excitation was used and emission Aggrecan-based PNs were typical matrix assemblies of
from Cy2 was recorded at 510-nm applying a low-range band the cervical grey matter (Fig. 2A-A3). Somata were
pass (505–550 nm). For secondary Cy3-labelled slices (red enwrapped by thin (Fig. 2A1), robust (Fig. 2A2) or
fluorescence), a helium–neon laser with 543-nm excitation was diffuse (Fig. 2A3) types of aggrecan+ PNs without
used and emission from Cy3 at 570 nm was detected applying
a high-range band pass (560–615 nm). For secondary Cy5-
distribution preference but higher frequency in the
labelled slices (infrared fluorescence) a helium–neon laser with ventral horn and central part of the spinal cord. Proximal
633-nm excitation was used and emission from Cy5 at 650 nm dendrites were often visualised by an outstanding matrix
was detected applying a long-pass (650 nm) filter. Photoshop cluster on their surface (Fig. 2A1).
CS2 (Adobe Systems, MountainView, CA, USA) was used to Both the overall pattern and the fine
process the images with minimal alterations to the brightness, compartmentalisation asked for new descriptions on
sharpness, colour saturation and contrast.
HAPLN-1-immunostained sections. Remarkably, the
dorsal horn showed strong immunoreactivity which
RESULTS abruptly discontinued in the substantia gelatinosa and
further in the Lissauer’s zone (Fig. 2B). The medial part
General considerations
of the lateral funiculus of the white matter was distinctly
The perineuronal and perisynaptic ECM in the human segregated by HAPLN-1+ matrix which remained
spinal cord was represented by unique distribution untinged with aggrecan and brevican stainings (Fig. 2A,
patterns and a great structural and chemical diversity. C). Higher magnification occasionally revealed PNs
Description and characterisation of the CSPG-based (Fig. 2B1–2) but more often laces of small round
ECM were based on the analysis of representative structures traceable for 10–15 lm in the strongly
spinal cord segments (Fig. 1). Chemical heterogeneity immunoreactive neuropil of all three horns, likely ACs
was dissected by detecting five fundamental ECM (Fig. 2B2). In contrast, neuropil staining in the substantia
components: hyaluronan, aggrecan, brevican, stabiliser gelatinosa was weak which allowed us to identify faintly
link protein 1 (HAPLN-1) and tenascin-R. Additionally, stained sheaths or ACs in a considerable number
versican, neurocan and NG2 proteoglycan were (Fig. 2B3).
investigated exemplarily to complete the study. The typical presence of PNs was not recalled when
Hyaluronan and tenascin-R are ubiquitous using the anti-brevican antibody. All three horns were
components of the ECM (Morawski et al., 2012b); this filled with small structures that gave the grey matter a
was confirmed in both white and grey matters of the lightly stained character (Fig. 2C). These structures
human spinal cord. Grey matter was more intensely were reminiscent of ACs, previously described in
labelled with condensed matrix assemblies around different subcortical (Brückner et al., 2008), thalamic
perisomatic compartment of neurons likely forming PNs (Lendvai et al., 2012) and cortical (Morawski et al.,
(Fig. 1C1–D4). Notably, the substantia gelatinosa 2012b) areas of the human brain as well. These
172 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Fig. 1. Nissl staining, schematic drawing, hyaluronan and tenascin-R staining overviews. (A1–4) Nissl-stained sections completed with schematic
drawings (B1–B4) indicate the major nuclei or regions in the cervical, thoracic, lumbar and sacral segments of the human spinal cord. Numbers tag
following domains: 1: Substantia gelatinosa, 2: Nucleus proprius, 3: Clarke’s column, 4: Visceral grey, 5: Nucleus intermediomedialis, 6: Nucleus
intermediolateralis, 7: Nucleus dorsomedialis, 8: Nucleus dorsolateralis, 9: Nucleus ventrolateralis, 10: Nucleus ventromedialis, 11: Nucleus
retrodorsolateralis, 12: Sacral parasympatic. Scale bar = 1 mm.

2–5-lm-sized round or oval structures (Fig. 2C1) outlined Two loci were discretely outlined in the thoracic spinal
(in few cases) somata (Fig. 2C2) or formed smaller or cord via their densely packed strongly HAG7D4+ PNs
bigger clusters (Fig. 2C3). (Fig. 2D). Neurons of the Clarke’s column nucleus
(Fig. 2D1) as well as of the intermediomedial nucleus
(Fig. 2D3), were decorated by thick aggrecan-
Thoracic segment immunoreactive PNs. At the same time, faintly labelled
The phenotype and fine structure of individual aggrecan+, PNs were seen in the slightly stained neuropil of the
HAPLN-1+ and brevican+ perineuronal matrix nucleus proprius (Fig 2D2). Similarly to cervical
assemblies were largely similar in thoracic compared to segments, PNs were present in all three horns with
cervical or any other spinal cord segments. At the same lower density in the posterior part of the dorsal horn
time, thoracic spinal cord segments showed (Fig. 2D).
characteristic overall distribution patterns of PNs and Overall and detailed structure patterns were
ACs (Fig. 2D–F). recapitulated in thoracic segments when investigating
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 173

Fig. 2. Aggrecan, brevican and HAPLN-1 staining overviews in the cervical and thoracic segment. (A) In cervical segments of the spinal cord,
ventral but not dorsal horn showed strong aggrecan immunoreactivity. Thin (A1), robust (A2) or diffuse (A3) types of perineuronal nets populated the
ventral horn especially. Peridendritic sheaths were repeatedly identified either emanating from soma (A1, empty arrowhead) or separately (A1,
arrowheads). (B) Ventral and dorsal horns were equally immunoreactive for the HAPLN-1 antibody that abruptly discontinued in the substantia
gelatinosa (SG) and Lissauer’s zone (L). Fascicles (f) in the lateral funiculus were outlined by HAPLN-1+ matrix assemblies. Perineuronal nets (B1,
B2, asterisks) were formed by thick matrix clusters, partly established by tiny (1 lm) ring-shaped profiles (B2, empty arrowheads). 10–15-lm-long
pearl-lace-like structures were often identified (B2, arrows). Due to low neuropil staining, long sheaths (arrow) or single axonal coats (empty
arrowhead) were revealed in the substantia gelatinosa (B3). (C) The grey matter was loaded with B50+ axonal coats but not perineuronal nets (C1).
Axonal coats occasionally outlined somata (C2) or formed irregular clusters (C3, arrows). (D) In thoracic segments, aggrecan+ perineuronal nets
outlined the Clarke’s column nucleus (Cl in D, for high power image see D1) and the intermediomedial nucleus (Imm in D, for high power image see
D3). In the nucleus proprius, perineuronal nets were labelled only weakly and occasionally (D2). (E, F) Thoracic segments recapitulated findings in
cervical segments with CRTL-1 and B50 immunostainings. HAPLN-1+ small ring shaped profiles formed sheaths (E1, E4 arrows) occured
individually (E1, E3 empty arrowheads), or formed bouquets (E3, arrows). Perineuronal nets were found only occasionally (E2). B50+ axonal coats
filled the thoracic grey matter in a large number (F1), individually (F2) or forming strings (F3). Faint perineuronal nets were very rare (F4). Scale
bars = 1 mm (A–F low power images), 10 lm (A1–3, B1–3, C1, D1–3, E1–2, F1), 5 lm (C2–3, E3–4, F3–4).

HAPLN-1+ and B50+ immunoreactivities. As shown on (Fig. 2E1,4). Perisomatic HAPLN-1+ matrix aggregates
examples captured in the ventrolateral nucleus formed non-conventional types of PNs in few cases
(Fig. 2E1,3,4) or visceral grey Fig. 2E2), the grey matter (Fig. 2E2). Similar structures were stained with the anti-
was loaded with 3–5-lm-sized round profiles, likely brevican antibody as presented in the intermediolateral
ACs. They were dispersed individually (Fig. 2E1), (Fig. 2F1–2) or ventrolateral nuclei (Fig. 2F3–4) with no
formed bouquets (Fig. 2E3) or lined up to form 10–20- preference in distribution throughout the grey matter. In
lm-long strings possibly outlining dendritic sections contrast to cervical segments, the subfasciculation in
174 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Fig. 3. Aggrecan, brevican and HAPLN-1 staining overviews in the lumbar and sacral segment. In lumbar segments, aggrecan+ and HAPLN-1+
perineuronal nets outlined the Clarke’s column nucleus (A, B, respectively). Aggrecan+ perineuronal nets were typically frequent in the dorsolateral
(A1) and ventromedial (A3, arrows point to small round profiles) nucleus, faintly labelled phenotypes occurred in the nucleus proprius (A2). HAPLN-
1+ perineuronal nets were mostly detectable in the ventromedial (B1, asterisk labels soma, arrow point to proximal dendrites ensheathed with
matrix) and Clarke’s column nucleus (B2). Axonal coats were typically labelled with both CRTL-1 (B3, arrows) and B50 (C1, arrows) antibodies and
formed smaller (C3, arrows) and larger (C4, arrows) subsets. As in other segments, brevican+ perineuronal nets were rare items (C2). Axonal coats
were more densely set at the border of the substantia gelatinosa (C, arrows). In sacral segments, perineuronal nets were optimally visualised via
their aggrecan-immunoreactivity (D), both their conventional (D1,3, arrows point to peridendritic sheaths) or faint (D2) types. Within the strongly
HAPLN-1+ neuropil, short chains (E1, arrows) of, or individual (empty arrowheads in E3) axonal coats and perineuronal nets (asterisks in E2–3) were
repeatedly recognised. Perineuronal matrix assemblies were, although in a low number, also discovered with the brevican-immunostaining (F1–3)
with brevican + peridendritic sheaths (arrows in F1–2). (F4) Not only axonal coats (arrows) but also peridendritic sheaths (empty arrowheads) were
mostly seen individually in the densely populated grey matter (F). Scale bars = 1 mm (A–F low power images), 10 lm (A1–3, B1–2, C1–2, D1–3, E1–3,
F1–2), 5 lm (B3, C3–4, F3–4) 2 lm (inset in A3).

the lateral funiculus could not be detected in HAPLN-1 nucleus were delicate, faintly labelled by aggrecan
immunostained thoracic sections. (Fig. 3A2) recapitulating findings in thoracic segments.
Aggrecan+ conventional PNs were found mainly in the
ventral horn as shown in its dorsolateral (Fig. 3A1) or
Lumbar segment
ventromedial (Fig. 3A3) nuclei. Unlike in cervical or
Beside similarities, lumbar spinal cord segments were thoracic segments, anti-HAPLN-1 displayed PNs. They
distinguished by the large number of strongly stained were detected in larger number in the ventromedial
aggrecan+ and HAPLN-1+ PNs in their Clarke’s column (Fig. 3B1) or Clarke’s column (Fig. 3B2) nuclei. Axonal
nuclei (Fig. 3A, B). In contrast, PNs in the proper coat-like structures appeared throughout the grey matter
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 175

Sacral segment
The most caudal segments of the spinal cord are typically
distinguished by their enlarged grey matter. This was
immediately recognised with the applied CSPG-based
matrix stainings (Fig. 3D–F). Aggrecan+, HAPLN-1+ or
brevican+ structures were evenly distributed in ventral
and lateral horns. As seen in all other segments, the
PNs of robust (Fig. 3D1,3) or faint types (Fig. 3D2) were
ideally labelled with the anti-aggrecan antibody. HAPLN-
1+ PNs (Fig. 3E1–3) and short chains of ACs (Fig. 3E1)
were repeatedly identified. As in all other segments,
ACs were best revealed by their brevican
immunoreactivity (Fig. 3F1–4). Somata were contacted
by delicate brevican+ matrix assemblies (Fig. 3F1–3)
that continued to peridendritic sheaths (Fig. 3F2,4).

Representation of versican, neurocan and NG2

proteoglycan
Versican shows a widely homogenous distribution in the
grey matter with a strong representation in some parts
of the dorsal horn (Fig. 4A). The substantia gelatinosa is
mostly spared of versican staining. Higher magnification
reveals typical association of versican staining to the
nodes of Ranvier (Fig. 4A2). Neurocan staining in the
grey matter shows a largely homogenous distribution as
well (Fig. 4B). In the transition zone of grey to white
matter neurocan appears condensed (Fig. 4B1)
sporadically delineating PNs (Fig. 4B2, white asterisk).
NG2 proteoglycan staining appears weak in the grey
matter except the area of the posterior nerve roots
(Fig. 4C). Higher magnification shows NG2
immunoreactivity that seems to be associated to myelin
sheats (Fig. 4C1). The marginal layers of the white
matter potentially representing the spinocerebellar tracts
often show NG2 immunoreactive cells likely to be
oligodendrocytes (Fig. 4C2). Distribution of versican,
Fig. 4. Versican, neurocan and NG2 proteoglycan staining in the neurocan and NG2 CSPG is shown exemplarily in the
thoracic segment Versican shows a widely homogenous distribution thoracic segment.
in the grey matter with a strong representation in some parts of the
dorsal horn (A). The substantia gelatinosa is mostly spared of
versican staining (A1). Higher magnification reveals typical associa- ECM relations to special types of neurons,
tion of versican staining to the nodes of Ranvier (A2). Neurocan transmitters and receptors
staining in the grey matter shows a largely homogenous distribution
as well (B). In the transition zone of grey to white matter neurocan The relation of the ECM assembly to neuronal, transmitter
appears condensed (B1), sporadically delineating perineuronal nets and receptor subtypes were investigated in the cervical
(B2, white asterisk). NG2 proteoglycan staining appears weak in the
grey matter except the area of the posterior nerve roots (C). Higher segments of the human spinal cord.
magnification shows NG2 immunoreactivity that seems to be asso- PNs are composed of chemically different
ciated to myelin sheats (C1). The exterior layers of the white matter components (Morawski et al., 2012a). This was
potentially representing the spinocerebellar tracts often show NG2 recapitulated in the human spinal cord with distinct
immunoreactive cells likely to be oligodendrocytes (C2). Distribution
compartmentalisation around the surrounded neuron.
of versican, neurocan and NG2 CSPG was shown exemplarily in the
thoracic segment. Scale bars = 1 mm (A–C low power images), Aggrecan, HAPLN-1 and brevican were fundamental
100 lm (A1), 50 lm (B1, C1), 20 lm (A2, B2, C2). elements of PNs with a wide, overall overlap between
HAPLN-1 and brevican but less between aggrecan in
the distal neuronal compartment (Fig. 5).
that showed both HAPLN-1 (Fig. 3B3) and brevican The most conspicuous neuronal populations in the
(Fig. 3C, C1) immunoreactivity. Brevican+ ACs were spinal cord are the large cholinergic motorneurons that
different in size, representing small sized (1–2 lm, populate the ventral horn. Most of these neurons
Fig. 3C3) or bigger sized (3–5 lm, Fig. 3C4) populations (cervical segment: 71%; thoracic segment: 64%;
and formed dense clusters in the substantia gelatinosa lumbar segment: 64%, sacral segment: 81%) were
(Fig. 3C). Remarkably, brevican+ PN-like structures surrounded by aggrecan-based PNs ensheathing the
could be occasionally identified (Fig. 3C2). somatodendritic compartment of the cells with different
176 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Fig. 5. Perineuronal net in the spinal cord visualised by main ECM components. The neuron in the ventral horn is surrounded by ECM that shows a
partial overlap of its main components aggrecan, brevican and HAPLN-1 (A4). The PN is of the thin type, delineated by a fine layer of aggrecan (A1).
Brevican and HAPLN-1 largely overlap, surrounding parts of the soma and the neurites (A2–3). Scale bar = 100 lm.

intensity (Fig. 6A1–3). Nevertheless, not all ChAT+ neuronal somata and distal parts of dendrites. The
neurons of the ventral horn were surrounded by PNs brevican-immunoreactive matrix formed small round or
and PNs were not exclusively associated to cholinergic oval-shaped structures representing ACs that contacted
neurons as shown by complementary ChAT and the neuronal surface (Fig. 6C1–3).
aggrecan immunopatterns (Fig. 6B1–3). In contrast, Firing patterns of neuronal cells and circuits are
brevican was detected with high intensity around powerfully controlled by inhibitory synapses (Freund and

Fig. 6. ECM and cholinergic motorneurons in the spinal cord. Many of the large cholinergic motorneurons in the ventral horn are surrounded by
aggrecan-based perineuronal nets detected with different intensity (A1–3). Additionally there were found PN-associated neurons of the same size
with weak or without immunoreactivity for ChAT and clearly cholinergic neurons without PN (B1–3). Some of the large cholinergic motorneurons are
contacted by structures immunoreactive for brevican, identified as axonal coats (C1–3). Brevican was detected with high intensity around distal parts
of the soma and on neurites but not delineating the whole soma like an aggrecan-based PN. Scale bars = 100 lm.
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 177

Buzsáki, 1996). In the spinal cord, inhibitory terminals contacted frequently (Fig. 8A1–3), exemplarily shown in
operate with both GABA as well as glycine as the cervical segment. In the dorsal horn, HAPLN-1-
neurotransmitters. We hypothesised that inhibitory positive dendrites were found to be partially contacted
terminals carry a matrix scaffold. Since single by Vglut-1 profiles, exemplarily shown in the sacral
perisynaptic matrix assemblies, hence, ACs were segment (Fig. 8B1–3). In the transition zone from the
selectively labelled with the anti-brevican and anti- dorsal horn into the posterior nerve roots a strong Vglut-
HAPLN-1antibodies we investigated the co-localisation 1-positive zone was detected. However, Vglut2-positive
of these matrix components with glycine and GABAa profiles were found only rarely could be detected (data
receptor markers. Glycine receptors were found widely not shown, see Technical considerations).
distributed in the grey matter, predominantly without a In the dorsal horn of the spinal cord, nociceptive
relation to HAPLN-1 containing structures (Fig. 7A1–3). primary afferents typically use substance P as
In contrast, brevican+ ACs repeatedly displayed co- transmitter (Todd et al., 2002). We focused our attention
localisation to glycine receptors (Fig. 7B1–3). Similar to this marker since the clinical relevance of substance
correlations were found concerning HAPLN-1 and P-mediated neurotransmission is exceptional in pain
GABAa receptors with a clear overlap along dendrites of syndromes and therapy. HAPLN-1 immunoreactivity was
large motorneurons (Fig. 7C1–3). low in laminae I and II of the dorsal horn but increased
Distribution of excitatory contacts was investigated by suddenly at is ventral border (Figs. 2B, E and 3B, E)
stainings for vesicular glutamate transporters 1 and 2. which allowed us to investigate the relationship of
Vglut-1 is widely distributed throughout the grey matter substance P immunoreactivity to the surrounding
of the spinal cord. In the ventral horn, somata of the perisynaptic matrix. We found that HAPLN-1 and
large cholinergic motorneurons are virtually spared of substance P showed largely non-overlapping distribution
Vglut-1 immunoreactivity. However, dendrites are (Fig. 9A1–3, B1–3).

Fig. 7. ECM and inhibitory transmitter receptors. Glycine receptors were found widely distributed in the grey matter, mostly without any connection
to HAPLN-1 containing structures, but in part virtually contacting somata of neurons (A1–3). Some cholinergic neurons (ChAT staining not displayed)
showed a strong co-localisation of glycine receptors with brevican-immunoreactive structures presumably representing synapse-associated axonal
coats (B1–3). Disperse distribution of GABAa receptors without co-localisation to ECM but sometimes with a clear overlap, found on neurites of large
motorneurons (C1–3). Scale bars = 50 lm (A1–3), 10 lm (B1–3, C1–3).
178 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Fig. 8. ECM and excitatory transmitter receptors. Vesicular glutamate transporter 1 is widely distributed throughout the grey matter of the spinal
cord. In the ventral horn, the somata of the large cholinergic motorneurons are virtually spared of Vglut-1 immunoreactivity. However, dendrites are
contacted frequently (A1–3), exemplarily shown in the cervical segment. In the dorsal horn, HAPLN-1 positive dendrites were found to be partially
contacted by Vglut-1 profiles, exemplarily shown in the sacral segment (B1–3). Scale bar = 20 lm (A1–B3).

Fig. 9. Distribution of ECM and substance P. In the dorsal horn, ECM detected by HAPLN-1 forms a clear distribution pattern that is inverse to that
from substance P. Partially the fibres of both stainings seem to be interwoven (overview: A1–3). The distribution of HAPLN-1 and substance P in
detail underlines that there is clearly no co-localisation of both markers (B1–3). Scale bars = 500 lm (A1–3), 200 lm (B1–3).

ECM and glia cell populations ECM. Microglia are less prominent and all-over
distributed surveilling defined areas of the tissue. They
In the spinal cord, astroglia is highly represented and
forms a dense network throughout the grey matter; in only rarely contact the PNs (Fig. 10A1–4). Interestingly,
the white matter they occur with less frequency. The the astrocytic population is restricted to the spinal cord
astrocytic network closely surrounds neurons and PNs; parenchyma and does not seem to cross the transition
astrocyte extensions seem to be interwoven with the zone to the peripheral nerve system, as observed in the
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 179

Fig. 10. Correlation of ECM and glia cell populations. Astroglia is widely distributed forming a dense network with close contact to neurons and
perineuronal nets. In contrast microglia occurs much less frequently, nevertheless rarely occurring in close proximity to perineuronal net ensheathed
neurons (A1–4).

area of the posterior nerve roots in the dorsal horn. In protein 1 accounting for the formation of PNs (Fawcett,
contrast, microglia also was found outside this area 2009) and ACs. This might imply that, e.g. hyaluronan is
between the nerve fibres showing slightly changed, an important component and potentially the backbone of
more amoeboid morphology, possibly due to their PNs and ACs but have other, more general functions in
monocytic character (results not shown). the spinal cord and the CNS as well.

DISCUSSION Matrix components show heterogeneity in their

cellular and compartmental recruitment
This study describes the distribution and phenotypic
appearance of major ECM components hyaluronan, The different compartmentalisations of aggrecan,
aggrecan, brevican, link protein 1 (HAPLN-1) and brevican and link protein argue for different functions at
tenascin-R in the whole dimension of the human spinal the cell surface and contacting terminals (Fig. 5). In this
cord. Whilst these molecules are most typical and regard, Frischknecht et al. (2009) described that
eminent components of the ECM of the CNS, other removal of ECM can lead to both diffusion and
CSPGs investigated like versican, neurocan and NG2 exchange of synaptic receptors. As recently shown, the
are present but do not show eye-catching distribution ECM can form compartments at the cell surface and
patterns in the non-diseased human spinal cord. We separate adjacent synapses (Blosa et al., 2013). This
demonstrate the presence, nucleus-specific distribution might facilitate locus-specific modifications of surface
and chemical heterogeneity of PNs and ACs. The molecules, and thereby regulate the localisation and
results indicate that these molecules, known to functions of synaptic receptors and ion channels.
contribute essentially to the ECM scaffold in the brain, Aggrecan in the human spinal cord extends also far
show distinct and specific distribution patterns and co- from soma around the cell surface and is virtually
localise with specific neuronal populations. Strikingly, restricted to PNs which might foster the stabilisation of
the overall distribution patterns vary between the the synapse as suggested by Hockfield and colleagues
different ECM components as well as with regard to the (1990) and/or could provide a diffusion barrier for
various laminae and different spinal cord segments positively charged molecules (Morawski et al., 2012b;
investigated. In general accordance with previously Lendvai et al., 2012, 2013; Blosa et al., 2013). Brevican
published data on ECM distribution and composition in immunoreactivity in the spinal cord is localised close to
rodent’s spinal cord (Kalb and Hockfield, 1988; the cell surface, sometimes delimiting the cell soma,
Bertolotto et al., 1996; Takahashi-Iwanaga et al., 1998; rarely revealing PNs and in most instances potentially
Deepa et al., 2006; Galtrey et al., 2008), major ECM covering synaptic endings showing massive ACs in the
components that are associated with PNs and ACs can grey matter in all segments. As hypothesised by our
be separated into two general groups. The first group, group and colleagues (Blosa et al., 2013) this could
encompassing aggrecan, brevican and link protein 1, is potentially restrict transmitters to the synaptic cleft and
virtually restricted to the grey matter solely of stabilise the receptor distribution on the neuronal
perineuronal and perisynaptic origin. This distribution membrane. HAPLN-1 immunoreactivity in the spinal
pattern implies that aggrecan, brevican and link protein cord is found to be most widely distributed in the grey
1 have an important role in the formation of PNs and matter, likely being a ‘‘link’’ between aggrecan and
ACs. Members of the second group, encompassing brevican patterns as HAPLN-1 is partially overlapping
hyaluronan and tenascin-R, are distributed much more with both proteins in both PNs and ACs. In accordance
ubiquitously; they appear in PNs and ACs of the grey with previous findings of our group (Brückner et al.,
matter but are equally present amorphously in the white 2008; Lendvai et al., 2012, 2013; Morawski et al.,
matter. Based on the mRNA expression of tenascin-R 2012b; Blosa et al., 2013) we show that HAPLN-1 and/
and HASs in new born rats Fawcett and colleagues or brevican positive ACs enwrap non-perisomatic
suggested a much earlier formation of these ubiquitous boutons in the human spinal cord (Fig. 7). Since these
representatives than of aggrecan, brevican and link assemblies are independent from PNs we supposed
180 C. Jäger et al. / Neuroscience 238 (2013) 168–184

that matrix of the ACs is contributed by the presynaptic vulnerable and to degenerate in Alzheimer’s disease
neuron. (Morawski et al., 2010a,c). We find a frequent although
not invariable association of PNs and cholinergic
Distribution patterns of ECM components and motorneurons in the human spinal cord which might
intersegmental differences imply functional differences between these cells.
Further, it suggests that the cholinergic nature of a
The most striking difference in matrix component neuron alone does not propose the presence or lack of
distribution included (i) the lack of aggrecan PNs in the nervous system.
immunoreactivity in the dorsal horn, (ii) the presence of Cholinergic motorneurons are the last order neurons
HAPLN-1+ and brevican+ structures in the substantia in the somatomotor system (Clara, 1959). Afferents on
gelatinosa but not in the Lissauer’s zone, (iii) the these cells represent therefore the ‘‘last chance’’ for the
demarcation of Clarke–Stilling and intermediomedial modulation of ‘‘final command exit’’. Inhibitory tuning is a
nuclei via their neurons surrounded by thick PNs and powerful way to fine-tune motorneuron activity (Geyer
(iv) the large number of ventral horn motorneurons et al., 1987). We hypothesised that inhibitory synapses
associated with PNs. associate with ECM assemblies. Indeed, glycine and
PNs are typically present in the efferent domain of the GABAa receptors were related to both peridendritic
spinal cord, thus, ventral and lateral/intermediate columns sheaths and single perisynaptic matrix aggregates,
which is reflected in the presence of aggrecan- hence, ACs.
immunoreactivity of the grey matter. Nuclei with distinct The dorsal horn receives fibres of diverse modalities
PNs include the Clarke–Stilling nucleus, from the periphery. Due to its clinical relevance, slow
intermediomedial nucleus and the group of large system pain fibres took central stage in neuroanatomical
motorneurons. Whilst these units are associated to research that use substance P as neurotransmitter. We
different systems they share the similar characteristic show that despite the large number of ACs in this
that they demarcate long-range projection neurons of region, substance P+ fibres and terminals are largely
the afferent cerebellar, efferent sympathetic or devoid of perisynaptic matrix assemblies. The
somatomotor pathways, respectively. This is in contrast complementary distribution of ACs and substance P+
to most brain regions where PNs are associated to local fibres as well as the absence of ACs in the Lissauer’s
circuit but not principal neurons (Brückner et al., 1994). zone suggest that pain fibres and synapses lack
Nevertheless, the ‘‘coupling’’ between projection individual matrix sheath in contrast to fibres of other
neurons and PNs is far from being exclusive since modalities.
neurons major efferent nuclei, e.g. intermediolateral
nucleus are not invariably surrounded by PNs. The impact of ECM assembly in the human spinal
The dorsal horn receives a large number of afferent cord
fibres which access the grey matter via the Lissauer’s
zone. Whilst nociceptive fibres terminate here in a large The recovery of function in the damaged spinal cord in
number, axons transmitting other modalities synapse human individuals is a longed-for goal but is limited due
only in the neighbouring substantia gelatinosa (Todd to the virtual absence of axonal regeneration and
et al., 2002). We argue that these latter synapses are relatively limited level of plasticity. In the adult CNS
surrounded by distinct, individual matrix assemblies plasticity is restricted especially in areas enriched in
which appear as HAPLN-1+ or brevican+ ACs on our PNs and ACs around neuronal cell bodies, dendrites
specimens. These unique matrix aggregates might and axon terminals, e.g. in primary sensory brain areas
support the synaptic integrity of afferent synapses and (Brückner et al., 1999). These PNs and ACs which form
can be promising targets of clinical approaches and at the end of the critical period comprise different
therapies. inhibitory CSPGs (mainly aggrecan and brevican in the
adult CNS). The level of plasticity in the adult CNS and
especially in the spinal cord was shown to be
ECM relations to special types of neurons,
reactivatable by treatment with chondroitinase ABC, an
transmitters and receptors
enzyme which selectively removes chondroitin
Brain regions and their neuronal subsets are sulphated glycosaminoglycan (CS-GAG) chains from
characterised by distinct ECM patterns. Thus, PNs proteoglycans (predominantly present on aggrecan) on
typically surround interneurons (Brückner et al., 1994) of PNs and ACs. This chondroitinase treatment is able to
the fast-spiking type (Härtig et al., 1992) that preferably promote sprouting of new connections in the damaged
populate areas associated with low plasticity (Brückner spinal cord (Fitch and Silver, 2008; Fawcett and Curt,
et al., 1999). We show that human spinal neurons are 2009; Bartus et al., 2011; for reviews see Bradbury and
differently associated to the same matrix components. Carter, 2011; Kwok et al., 2011). As reported by
PNs are dominating around large, hence, projection Hunanyan et al. (2010) and Andrews et al. (2012), in
neurons that may belong, however, to different systems. case of spinal cord injuries (SCI), neurocan (Andrews
Along the same lines, the association of PNs to neurons et al., 2012) and NG2 (Hunanyan et al., 2010; Andrews
with similar neurochemical make-up is strikingly different et al., 2012) expression is increased. In addition to the
in the brain and spinal cord. Cholinergic neurons in the fact that ECM proteoglycans reduce axonal growth and
basal forebrain are spared from PNs (Brauer et al., sprouting which can partially be restored by
1993; Adams et al., 2001) which are known to be chondroitinase digestion, they might take action on
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 181

axonal conduction, too. Especially NG2 is reported to Alzheimer Forschungsinitiative e.V. (AFI #11861) and the
have a strengthening potential on weak projections by German Research Foundation MO 2249/2-1 within the SPP
influencing axonal conduction (Hunanyan et al., 2010). 1608 to M. Morawski. This work was supported by the Scottish
The chronic changes in ECM composition after SCI Universities Life Science Alliance to A. Alpar.
implicate the ECM as a viable target for the
development of combined therapies aiming to restore
spinal cord function after SCI (Hunanyan et al., 2010; REFERENCES
Andrews et al., 2012).
Treatments enhancing plasticity e.g. by removing the Adams I, Brauer K, Arélin C, Härtig W, Fine A, Mäder M, Arendt T,
CS-GAG chains may re-open a window for successful Brückner G (2001) Perineuronal nets in the rhesus monkey and
rehabilitation of spinal cord damage; nevertheless, they human basal forebrain including basal ganglia. Neuroscience
108:285–298.
may also cause side-effects due to pleiotropic and
Alilain WJ, Horn KP, Hu H, Dick TE, Silver J (2011) Functional
unknown functions of CSPGs. Thus, identifying the regeneration of respiratory pathways after spinal cord injury.
therapeutic utility of chondroitinase ABC or any other Nature 475(7355):196–200.
ECM degrading enzymatic treatment in the CNS, Andrews EM, Richards RJ, Yin FQ, Viapiano MS, Jakeman LB (2012)
especially human spinal cord, needs an in-depth Alterations in chondroitin sulfate proteoglycan expression occur
analysis of CSPG distribution and composition. This both at and far from the site of spinal contusion injury. Exp Neurol
235:174–187.
knowledge was in humans largely lacking until now.
Baer K, Waldvogel HJ, Faull RL, Rees MI (2009) Localization of
glycine receptors in the human forebrain, brainstem, and cervical
Technical considerations spinal cord: an immunohistochemical review. Front Mol Neurosci
2:25.
Visualisation of ECM with biotinylated lectins which are Bandtlow CE, Zimmermann DR (2000) Proteoglycans in the
known to mark N-acetyl-galactosamin in ECM developing brain: new conceptual insights for old proteins.
constituents (Wisteria floribunda lectin (WFA; Brückner Physiol Rev 80:1267–1290.
et al., 1993; Härtig et al., 1992; Seeger et al., 1994, Barritt AW, Davies M, Marchand F, Hartley R, Grist J, Yip P,
McMahon SB, Bradbury EJ (2006) Chondroitinase ABC promotes
1996) and Vicia villosa lectin (VVA; Kobayashi et al.,
sprouting of intact and injured spinal systems after spinal cord
1989; Murakami et al., 1999)) was not successful, injury. J Neurosci 26(42):10856–10867.
potentially due to matrix degradation because of the Bartus K, James ND, Bosch KD, Bradbury EJ (2011) Chondroitin
long PMD (Morawski et al., 2012b). For that reason it sulphate proteoglycans: key modulators of spinal cord and brain
was not possible to study the amount of CS-GAG that is plasticity. Exp Neurol 235(1):5–17.
known to be different in distinct PN populations Bekku Y, Su WD, Hirakawa S, Fässler R, Ohtsuka A, Kang JS,
Sanders J, Murakami T, Ninomiya Y, Oohashi T (2003) Molecular
(Morawski et al., 2012b). We were virtually unable to
cloning of Bral2, a novel brain-specific link protein, and
detect trustworthy signals for vesicular glutamate immunohistochemical colocalization with brevican in
transporter 2 which might reflect previous perineuronal nets. Mol Cell Neurosci 24:148–159.
immunohistochemical limitations on human nervous Bekku Y, Oohashi T (2010) Neurocan contributes to the molecular
tissue (Lendvai et al., 2012) or the very rare appearance heterogeneity of the perinodal ECM. Arch Histol Cytol 73:95–102.
of vesicular glutamate transporter 2 in the investigated Bertolotto A, Manzardo E, Guglielmone R (1996)
Immunohistochemical mapping of perineuronal nets containing
human spinal cord areas.
chondroitin unsulfated proteoglycan in the rat central nervous
Due to identical overall matrix distribution principles in system. Cell Tissue Res 283(2):283–295.
the total cranio-caudal dimension of the spinal cord, we Berardi N, Pizzorusso T, Ratto GM, Maffei L (2003) Molecular basis
limited transmitter- and receptor subtype-related of plasticity in the visual cortex. Trends Neurosci 26:369–378.
investigations mainly to the cervical segment of the Blosa M, Sonntag M, Brückner G, Jäger C, Seeger G, Matthews RT,
spinal cord. Rübsamen R, Arendt T, Morawski M (2013) Unique features of
extracellular matrix in the mouse medial nucleus of trapezoid
body – implications for physiological functions. Neuroscience
CONCLUDING REMARKS 228:215–234.
Bradbury EJ, Moon LD, Popat RJ, King VR, Bennett GS, Patel PN,
The human spinal cord is characterised by typical ECM Fawcett JW, McMahon SB (2002) Chondroitinase ABC promotes
blueprints in its various regions and around its distinct functional recovery after spinal cord injury. Nature
neuronal subsets. We believe and show examples that 416(6881):636–640.
the versatile recruitment of matrix components in the Brauer K, Härtig W, Bigl V, Brückner G (1993) Distribution of
parvalbumin-containing neurons and lectin-binding perineuronal
different spinal cord modules reflects functional
nets in the rat basal forebrain. Brain Res. 631:167–170.
differences. Bradbury EJ, Carter LM (2011) Manipulating the glial scar:
Currently, there are a lot of regenerative approaches chondroitinase ABC as a therapy for spinal cord injury. Brain
using ECM degradation by chondroitinase ABC to treat Res Bull 84(4–5):306–316.
SCI. A detailed knowledge of the ECM properties in the Brückner G, Brauer K, Härtig W, Wolff JR, Rickmann MJ, Derouiche
human spinal cord can help to improve these A, Delpech B, Girard N, Oertel WH, Reichenbach A (1993)
approaches and might bring new therapy concepts. Perineuronal nets provide a polyanionic, glia-associated form of
microenvironment around certain neurons in many parts of the rat
brain. Glia 8:183–200.
Acknowledgements—This work was supported by the German Brückner G, Seeger G, Brauer K, Härtig W, Kacza J, Bigl V (1994)
Research Foundation GRK 1097 ‘‘INTERNEURO,’’ the EU-Pro- Cortical areas are revealed by distribution patterns of
ject ‘‘Neuropro’’ (Grant Agreement No. 223077), COST Action proteoglycan components and parvalbumin in the Mongolian
BM1001 ‘‘Brain Extracellular Matrix in Health and Disease’’, the gerbil and rat. Brain Res 658:67–86.
182 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Brückner G, Härtig W, Kacza J, Seeger J, Welt K, Brauer K (1996) and functional microdomains in the brain. Trends Neurosci
Extracellular matrix organization in various regions of rat brain 33:503–512.
grey matter. J Neurocytol 25:333–346. Donaggio A (1898) Nuove osservazioni sulla struttura della cellula
Brückner G, Härtig W, Seeger J, Rübsamen R, Reimer K, Brauer K nervosa. Riv Sper Freniatr 24:772–778.
(1998a) Cortical perineuronal nets in the gray short-tailed Fawcett JW (2009) Recovery from spinal cord injury: regeneration,
opossum (Monodelphis domestica): a distribution pattern plasticity and rehabilitation. Brain 132(6):1417–1418.
contrasting with that shown in placental mammals. Anat Fawcett JW, Curt A (2009) Damage control in the nervous
Embryol 197:249–262. system: rehabilitation in a plastic environment. Nat Med
Brückner G, Bringmann A, Härtig W, Köppe G, Delpech B, Brauer K 15(7):735–736.
(1998b) Acute and long-lasting changes in extracellular-matrix Fitch MT, Silver J (2008) CNS injury, glial scars, and inflammation:
chondroitin-sulfate proteoglycans induced by injection of inhibitory extracellular matrices and regeneration failure. Exp
chondroitinase ABC in the adult rat brain. Exp Brain Res Neurol 209(2):294–301.
121:300–310. Freund TF, Buzsáki G (1996) Interneurons of the hippocampus.
Brückner G, Hausen D, Härtig W, Drlicek M, Arendt T, Brauer K Hippocampus 6:347–470.
(1999) Cortical areas abundant in extracellular matrix chondroitin Frischknecht R, Heine M, Perrais D, Seidenbecher C, Choquet D,
sulphate proteoglycans are less affected by cytoskeletal changes Gundelfinger ED (2009) Brain extracellular matrix affects AMPA
in Alzheimer’s disease. Neuroscience 92(3):791–805. receptor lateral mobility and short-term synaptic plasticity. Nat
Brückner G, Grosche J, Schmidt S, Härtig W, Margolis RU, Delpech Neurosci 12:897–904.
B, Seidenbecher CI, Czaniera R, Schachner M (2000) Postnatal Galtrey CM, Fawcett JW (2007) The role of chondroitin sulfate
development of perineuronal nets in wild-type mice and in a proteoglycans in regeneration and plasticity in the central nervous
mutant deficient in tenascin-R. J Comp Neurol 428:616–629. system. Brain Res Rev 54:1–18.
Brückner G, Grosche J, Hartlage-Rübsamen M, Schmidt S, Galtrey CM, Kwok JC, Carulli D, Rhodes KE, Fawcett JW (2008)
Schachner M (2003) Region and lamina-specific distribution of Distribution and synthesis of extracellular matrix proteoglycans,
extracellular matrix proteoglycans, hyaluronan and tenascin-R in hyaluronan, link proteins and tenascin-R in the rat spinal cord. Eur
the mouse hippocampal formation. J Chem Neuroanat 26:37–50. J Neurosci 27(6):1373–1390.
Brückner G, Szeöke S, Pavlica S, Grosche J, Kacza J (2006) Axon Garcı́a-Alı́as G, Barkhuysen S, Buckle M, Fawcett JW (2009)
initial segment ensheathed by extracellular matrix in perineuronal Chondroitinase ABC treatment opens a window of opportunity
nets. Neuroscience 138:365–375. for task-specific rehabilitation. Nat Neurosci 12(9):1145–1151.
Brückner G, Morawski M, Arendt T (2008) Aggrecan-based Garcı́a-Alı́as G, Fawcett JW (2012) Training and anti-CSPG
extracellular matrix is an integral part of the human basal combination therapy for spinal cord injury. Exp Neurol 235:26–32.
ganglia circuit. Neuroscience 151:489–504. Geyer SW, Gudden W, Betz H, Gnahn H, Weindl A (1987) Co-
Busch SA, Silver J (2007) The role of extracellular matrix in CNS localization of choline acetyltransferase and postsynaptic glycine
regeneration. Curr Opin Neurobiol 17(1):120–127. receptors in motoneurons of rat spinal cord demonstrated by
Carulli D, Rhodes KE, Brown DJ, Bonnert TP, Pollack SJ, Oliver K, immunocytochemistry. Neurosci Lett 82:11–15.
Strata P, Fawcett JW (2006) Composition of perineuronal nets in Giamanco KA, Morawski M, Matthews RT (2010) Perineuronal net
the adult rat cerebellum and the cellular origin of their formation and structure in aggrecan knockout mice. Neuroscience
components. J Comp Neurol 494:559–577. 170(4):1314–1327.
Carulli D, Rhodes KE, Fawcett JW (2007) Upregulation of aggrecan, Golgi C (1882) Nervoso (Sistema). In: Enciclopedia medica Vallardi,
link protein 1, and hyaluronan synthases during formation of Vol. 3, No. 1, pp. 93–122.
perineuronal nets in the rat cerebellum. J Comp Neurol Golgi C (1893) Intorno all’origine del quarto nervo cerebrale e una
501(1):83–94. questione isto-fisiologica che a questo argomento si collega.
Carulli D, Pizzorusso T, Kwok JC, Putignano E, Poli A, Forostyak S, Rendiconti della Reale Accademia dei Lincei 2:379–389.
Andrews MR, Deepa SS, Glant TT, Fawcett JW (2010) Animals Golgi C (1898) Intorno alla struttura delle celle nervose. Boll Soc Med
lacking link protein have attenuated perineuronal nets and Chir Pavia 1:1–14.
persistent plasticity. Brain 133(8):2331–2347. Härtig W, Brauer K, Brückner G (1992) Wisteria floribunda agglutinin-
Celio MR, Blümcke I (1994) Perineuronal nets – a specialized form of labelled nets surround parvalbumin-containing neurons.
extracellular matrix in the adult nervous system. Brain Res Rev Neuroreport 3(10):869–872.
19(1):128–145. Härtig W, Brückner G, Brauer K, Schmidt C, Bigl V (1995) Allocation
Celio MR, Heizmann CW (1981) Calcium-binding protein parvalbumin of perineuronal nets and parvalbumin-, calbindin-D28- and
as a neuronal marker. Nature 293:300–302. glutamic acid decarboxylase-immunoreactivity in the amygdala
Celio MR, Spreafico R, De Biasi S, Vitellaro-Zuccarello L (1998) of the rhesus monkey. Brain Res 698:265–269.
Perineuronal nets: past and present. Trends Neurosci Härtig W, Derouiche A, Welt K, Brauer K, Grosche J, Mader M,
21(12):510–515. Reichenbach A, Brückner G (1999) Cortical neurons
Clara M (1959) Das Nervensystem des Menschen. J.A Barth. p. 772. immunoreactive for the potassium channel Kv3.1b subunit are
Crespo D, Asher RA, Lin R, Rhodes KE, Fawcett JW (2007) How predominantly surrounded by perineuronal nets presumed as a
does chondroitinase promote functional recovery in the damaged buffering system for cations. Brain Res 842:15–29.
CNS? Exp Neurol 206(2):159–171. Herzog E, Takamori S, Jahn R, Brose N, Wojcik SM (2006) Synaptic
Deepa SS, Carulli D, Galtrey C, Rhodes K, Fukuda J, Mikami T, and vesicular co-localization of the glutamate transporters
Sugahara K, Fawcett JW (2006) Composition of perineuronal net VGLUT1 and VGLUT2 in the mouse hippocampus. J
extracellular matrix in rat brain: a different disaccharide Neurochem 99:1011–1018.
composition for the net-associated proteoglycans. J Biol Chem Hockfield S, McKay RD (1983) A surface antigen expressed by a
281(26):17789–17800. subset of neurons in the vertebrate central nervous system. Proc
Dino MR, Harroch S, Hockfield S, Matthews RT (2006) Monoclonal Natl Acad Sci U S A 80:5758–5761.
antibody Cat-315 detects a glycoform of receptor protein tyrosine Hockfield S, Kalb RG, Zaremba S, Fryer H (1990) Expression of
phosphatase beta/phosphacan early in CNS development that neural proteoglycans correlates with the acquisition of mature
localizes to extrasynaptic sites prior to synapse formation. neuronal properties in the mammalian brain.-. Cold Spring Harb
Neuroscience 142:1055–1069. Symp Quant Biol 55:505–514.
Dityatev A, Schachner M (2003) Extracellular matrix molecules and Hunanyan AS, Garcı́a-Alı́as G, Alessi V, Levine JM, Fawcett JW,
synaptic plasticity. Nat Rev Neurosci 4:456–468. Mendell LM, Arvanian VL (2010) Role of chondroitin sulfate
Dityatev A, Seidenbecher CI, Schachner M (2010) proteoglycans in axonal conduction in mammalian spinal cord. J
Compartmentalization from the outside: the extracellular matrix Neurosci 9(30):7761–7769.
 C. Jäger et al. / Neuroscience 238 (2013) 168–184 183

Kalb RG, Hockfield S (1988) Molecular evidence for early activity- hedgehog tenrec (Echinops telfairi Martin, 1838). Neuroscience
dependent development of hamster motor neurons. J Neurosci 165:831–849.
8(7):2350–2360. Morawski M, Brückner G, Jäger C, Seeger G, Arendt T (2010c)
Kanazawa H, Ohsawa K, Sasaki Y, Kohsaka S, Imai Y (2002) Neurons associated with aggrecan-based perineuronal nets are
Macrophage/microglia-specific protein Iba1 enhances membrane protected against tau pathology in subcortical regions in
ruffling and Rac activation via phospholipase C-gamma- Alzheimer’s disease. Neuroscience 169:1347–1363.
dependent pathway. J Biol Chem 277(22):20026–20032. Morawski M, Brückner G, Arendt T, Matthews RT (2012a) Aggrecan:
Kobayashi K, Emson PC, Mountjoy CQ (1989) Vicia villosa lectin- beyond cartilage and into the brain. Int J Biochem Cell Biol
positive neurones in human cerebral cortex. Loss in Alzheimer- 44(5):690–693.
type dementia. Brain Res 498:170–174. Morawski M, Brückner G, Jäger C, Seeger G, Matthews RT, Arendt T
Köppe G, Brückner G, Brauer K, Härtig W, Bigl V (1997) (2012b) Involvement of perineuronal and perisynaptic
Developmental patterns of proteoglycan-containing extracellular extracellular matrix in Alzheimer’s disease neuropathology.
matrix in perineuronal nets and neuropil of the postnatal rat brain. Brain Pathol 22(4):547–561.
Cell Tissue Res 288:33–41. Murakami T, Ohtsuka A, Su WD, Taguchi T, Oohashi T, Murakami T,
Kozsurek M, Lukácsi E, Fekete C, Puskár Z (2009) Nonselective Abe K, Ninomiya Y (1999) The extracellular matrix in the mouse
innervation of lamina I projection neurons by cocaine- and brain: its reactions to endo-alpha-N-acetylgalactosaminidase and
amphetamine-regulated transcript peptide (CART)- certain other enzymes. Arch Histol Cytol 62(3):273–281.
immunoreactive fibres in the rat spinal dorsal horn. Eur J Neame PJ, Barry FP (1993) The link proteins. Experientia
Neurosci 29:2375–2387. 49:393–402.
Kwok JC, Carulli D, Fawcett JW (2010) In vitro modeling of Neame PJ, Barry FP (1994) The link proteins. Experientia 70:53–72.
perineuronal nets: hyaluronan synthase and link protein are Ohyama J, Ojima H (1997) Labeling of pyramidal and nonpyramidal
necessary for their formation and integrity. J Neurochem neurons with lectin Vicia villosa during postnatal development of
114(5):1447–1459. the guinea pig. J Comp Neurol 389:453–468.
Kwok JC, Dick G, Wang D, Fawcett JW (2011) Extracellular matrix Pizzi MA, Crowe MJ (2007) Matrix metalloproteinases and
and perineuronal nets in CNS repair. Dev Neurobiol proteoglycans in axonal regeneration. Exp Neurol
71(11):1073–1089. 204(2):496–511.
Lendvai D, Morawski M, Brückner G, Négyessy L, Baksa G, Glasz T, Pizzorusso T, Medini P, Berardi N, Chierzi S, Fawcett JW, Maffei L
Patonay L, Matthews RT, Arendt T, Alpár A (2012) Perisynaptic (2002) Reactivation of ocular dominance plasticity in the adult
aggrecan-based extracellular matrix coats in the human lateral visual cortex. Science 298:1248–1251.
geniculate body devoid of perineuronal nets. J Neurosci Res Pizzorusso T, Medini P, Landi S, Baldini S, Berardi N, Maffei L (2006)
90(2):376–387. Structural and functional recovery from early monocular
Lendvai D, Morawski M, Negyessy L, Gati G, Jäger C, Baksa G, deprivation in adult rats. Proc Natl Acad Sci U S A
Glasz T, Attems J, Tanila H, Arendt T, Harkany T, Alpar A (2013) 103(22):8517–8522.
Neurochemical mapping of the human hippocampus reveals Ramón y Cajal SF (1897) Las células de cilindro-eje corto de la capa
perisynaptic matrix around functional synapses in Alzheimer’s molecular del cerebro. Rev Trim Microgr 2:105–127.
disease. Acta Neuropathol 125:215–229. Rauch U (2007) Brain matrix: structure, turnover and necessity.
Lugaro E (1895) Sulla struttura del nucleo dentato del cervelletto Biochem Soc Trans 35:656–660.
nell’uomo. Monit Zool Ital 6:5–12. Reinert T, Morawski M, Arendt T, Butz T (2003) Quantitative
Ma XM, Huang JP, Kim EJ, Zhu Q, Kuchel GA, Mains RE, Eipper BA microanalysis of perineuronal nets in brain tissue. Nucl Instrum
(2011) Kalirin-7, an important component of excitatory synapses, Meth B 210:395–400.
is regulated by estradiol in hippocampal neurons. Hippocampus Rhodes KE, Fawcett JW (2004) Chondroitin sulphate proteoglycans:
21(6):661–677. preventing plasticity or protecting the CNS? J Anat 204:33–48.
Massey JM, Hubscher CH, Wagoner MR, Decker JA, Amps J, Silver Romeis – Mikroskopische Technik (2010) Mulisch, Maria; Welsch,
J, Onifer SM (2006) Chondroitinase ABC digestion of the Ulrich (Hrsg.) 18. Aufl., XII, 556.
perineuronal net promotes functional collateral sprouting in the Schnell SA, Staines WA, Wessendorf MW (1999) Reduction of
cuneate nucleus after cervical spinal cord injury. J Neurosci lipofuscin-like autofluorescence in fluorescently labeled tissue. J
26:4406–4414. Histochem Cytochem 47(6):719–730.
Matthews RT, Gary SC, Zerillo C, Pratta M, Solomon K, Arner EC, Seeger G, Brauer K, Härtig W, Brückner G (1994) Mapping of
Hockfield S (2000) Brain-enriched hyaluronan binding (BEHAB)/ perineuronal nets in the rat brain stained by colloidal iron
brevican cleavage in a glioma cell line is mediated by a disintegrin hydroxide histochemistry and lectin cytochemistry.
and metalloproteinase with thrombospondin motifs (ADAMTS) Neuroscience 58(2):371–388.
family member. J Biol Chem 275(30):22695–22703. Seeger G, Lüth HJ, Winkelmann E, Brauer K (1996) Distribution
Matthews RT, Kelly GM, Zerillo CA, Gray G, Tiemeyer M, patterns of Wisteria floribunda agglutinin binding sites and
Hockfield S (2002) Aggrecan glycoforms contribute to the parvalbumin-immunoreactive neurons in the human visual
molecular heterogeneity of perineuronal nets. J Neurosci cortex: a double-labelling study. J Hirnforsch 37(3):351–366.
22:7536–7547. Spicer AP, Joo A, Bowling Jr RA (2003) A hyaluronan binding link
Morawski M, Brückner MK, Riederer P, Brückner G, Arendt T (2004) protein gene family whose members are physically linked
Perineuronal nets potentially protect against oxidative stress. Exp adjacent to chondroitin sulfate proteoglycan core protein genes:
Neurol 188:309–315. the missing links. J Biol Chem 278(23):21083–21091.
Morawski M, Alpár A, Brückner G, Fiedler A, Jäger C, Gati G, Stieler Standring S, editor. Gray’s Anatomy: The Anatomical Basis of Clinical
JT, Arendt T (2009) Chondroitin sulfate proteoglycan-based Practice. Elsevier.
extracellular matrix in chicken (Gallus domesticus) brain. Brain Szigeti ZM, Matesz C, Szekely G, Felszeghy S, Bácskai T, Halasi G,
Res 1275:10–23. Mészár Z, Módis L (2006) Distribution of hyaluronan in the central
Morawski M, Pavlica S, Seeger G, Grosche J, Kouznetsova E, nervous system of the frog. J Comp Neurol 496(6):819–831.
Schliebs R, Brückner G, Arendt T (2010a) Perineuronal nets are Takahashi-Iwanaga H, Murakami T, Abe K (1998) Three-dimensional
largely unaffected in Alzheimer model Tg2576 mice. Neurobiol microanatomy of perineuronal proteoglycan nets enveloping
Aging 31:1254–1256. motor neurons in the rat spinal cord. J Neurocytol
Morawski M, Brückner G, Jäger C, Seeger G, Künzle H, Arendt T 27(11):817–827.
(2010b) Aggrecan-based extracellular matrix shows unique Takamori S, Rhee JS, Rosenmund C, Jahn R (2000) Identification of
cortical features and conserved subcortical principles of a vesicular glutamate transporter that defines a glutamatergic
mammalian brain organization in the Madagascan lesser phenotype in neurons. Nature 407:189–194.
184 C. Jäger et al. / Neuroscience 238 (2013) 168–184

Todd AJ, Puskar Z, Spike RC, Hughes C, Watt C, Forrest L (2002) (2009) Differential distribution of aggrecan isoforms in
Projection neurons in lamina I of rat spinal cord with the perineuronal nets of the human cerebral cortex. J Cell Mol Med
neurokinin 1 receptor are selectively innervated by substance 13:3151–3173.
p-containing afferents and respond to noxious stimulation. J Vitellaro-Zuccarello L, Bosisio P, Mazzetti S, Monti C, De Biasi S
Neurosci 22:4103–4113. (2007) Differential expression of several molecules of the
Vargová L, Syková E (2008) Extracellular space diffusion and extracellular matrix in functionally and developmentally distinct
extrasynaptic transmission. Physiol Res 57(Suppl. 3):S89–S99. regions of rat spinal cord. Cell Tissue Res 327(3):433–447.
Viapiano MS, Matthews RT (2006) From barriers to bridges: Wu Y, Wu J, Lee DY, Yee A, Cao L, Zhang Y, Kiani C, Yang BB
chondroitin sulfate proteoglycans in neuropathology. Trends Mol (2005) Versican protects cells from oxidative stress-induced
Med 12(10):488–496. apoptosis. Matrix Biol 24:3–13.
Virgintino D, Perissinotto D, Girolamo F, Mucignat MT, Montanini L, Yamaguchi Y (2000) Lecticans: organizers of the brain extracellular
Errede M, Kaneiwa T, Yamada S, Sugahara K, Roncali L, Perris R matrix. Cell Mol Life Sci 57:276–289.

(Accepted 8 February 2013)

(Available online 18 February 2013)