Báo cáo khoa học: Identification of versican as an isolectin B4-binding glycoprotein from mammalian spinal cord tissue pptx

Moreover, we demonstrate that the IB4-reactive glycoconjugate and the versican variant can be co-released from spinal cord membranes by hyaluronidase, and that the IB4-reactive glycoconj

glycoprotein from mammalian spinal cord tissue

Oliver Bogen1, Mathias Dreger1,*, Clemens Gillen2, Wolfgang Schro¨der2and Ferdinand Hucho1

1 Freie Universita¨t Berlin, Institut fu¨r Chemie-Biochemie, Thielallee, Berlin, Germany

2 Research and Development Gru¨nenthal GmbH, Aachen, Germany

Noxius stimuli are detected by specialized sets of

pri-mary afferent neurons, the nociceptors All nociceptors

are represented by C-fibers and Ad-fibers, neurons with

small- to medium-sized cell bodies and unmyelinated or

lightly myelinated axons, respectively [1] A subpopu-lation of these nociceptors express a cell-surface glyco-conjugate that can be labeled by the plant isolectin B4 (IB4) from Griffonia simplicifolia [2] Owing to the fact

Keywords

IB4; versican; nonpeptidergic C-fibers;

extracellular matrix; neuropathic pain

Correspondence

Freie Universita¨t Berlin, Institut fu¨r

Chemie-Biochemie, Thielallee 63, 14195

Berlin, Germany

Fax: +49 3083 853753

Tel: +49 3083 855545

E-mail: hucho@chemie.fu-berlin.de

*Present address

University Laboratory of Physiology, Parks

Road, Oxford OX1 3PT, UK

Glossary

Allodynia, sensation of pain caused by stimuli

that are normally innocuous; cross-excitation,

non-synaptic depolarization of dorsal root

ganglia neurons in response to excitation of

neighbouring neurons; hyperalgesia,

increased responsiveness of nociceptors

upon noxious stimulation; neuropathic pain,

pain initiated or caused by a primary lesion or

dysfunction in the nervous system;

nocicep-tor, primary sensory neuron that is activated

by stimuli capable of causing tissue damage.

(Received 4 November 2004, revised 11

December 2004, accepted 21 December

2004)

doi:10.1111/j.1742-4658.2005.04543.x

Nociceptors are specialized nerve fibers that transmit noxious pain stimuli

to the dorsal horn of the spinal cord A subset of nociceptors, the nonpepti-dergic C-fibers, is characterized by its reactivity for the plant isolectin B4 (IB4) from Griffonia simplicifolia The molecular nature of the IB4-reactive glycoconjugate, although used as a neuroanatomical marker for more than

a decade, has remained unknown We here present data which strongly sug-gest that a splice variant of the extracellular matrix proteoglycan versican is the IB4-reactive glycoconjugate associated with these nociceptors We isola-ted (by subcellular fractionation and IB4 affinity chromatography) a glyco-conjugate from porcine spinal cord tissue that migrated in SDS⁄ PAGE as a single distinct protein band at an apparent molecular mass of > 250 kDa

By using MALDI-TOF⁄ TOF MS, we identified this glycoconjugate unam-biguously as a V2-like variant of versican Moreover, we demonstrate that the IB4-reactive glycoconjugate and the versican variant can be co-released from spinal cord membranes by hyaluronidase, and that the IB4-reactive glycoconjugate and the versican variant can be co-precipitated by an anti-versican immunoglobulin and perfectly co-migrate in SDS⁄ PAGE Our findings shed new light on the role of the extracellular matrix, which is thought to be involved in plastic changes underlying pain-related phenom-ena such as hyperalgesia and allodynia

Abbreviations

DRG, dorsal root ganglia; ECL, enhanced chemiluminescence; GDNF, glial cell line-derived neurotrophic factor; GHAP, glial hyaluronate-binding protein; IB4, isolectin B4; NaCl ⁄ P i , phosphate-buffered saline; PSD, post source decay; NaCl ⁄ Tris, Tris-buffered saline.

that they apparently lack neuropeptide storage vesicles,

they are called nonpeptidergic C-fibers [3] The IB4

reactivity is primarily used as an anatomical marker for

these C-fibers The lectin consists of four identical

poly-peptide chains, and the homotetramer has an apparent

molecular mass of 114 kDa Purification initially was

based on its affinity to the disaccharide melibiose [4]

IB4 binds selectively to a carbohydrate epitope

contain-ing a terminal a-d-galactoside The bindcontain-ing depends on

the presence of Ca2+ions Because of the specificity for

nonpeptidergic C-fibers, the possibility exists that the

IB4-reactive glycoconjugate itself could participate in

nociception Indeed, the IB4-positive subpopulation of

nociceptors appears to play a distinct role in certain

pain transmission paradigms Spinal nerve ligation

leads to hyperalgesia and allodynia [5,6]

Concomit-antly, IB4-reactivity depletion was observed in the

affected dorsal root ganglia (DRG) and in lamina II of

the spinal cord dorsal horn [7] On the other hand, glial

cell line-derived neurotrophic factor (GDNF), a

neur-onal survival factor, has been shown to prevent the loss

of IB4-positive fibers caused by axotomy [8]

Simulta-neously it prevents the development of hyperalgesia

and allodynia [9] From these observations, a potential

therapeutic role of GDNF in states of neuropathic pain

was deduced

The mechanism of IB4-reactivity downregulation is

unclear It could be caused by neuronal cell death or

altered gene expression A third possibility would be a

change in the post-translational modification

(de-glyco-sylation) Similarly, a reversal of the changes caused

by injury could be a result of the survival and

out-growth of IB4-positive fibers, of an increased

expres-sion of the glycoconjugate, or of an increased

concentration of the IB4-binding epitope caused by

post-translational modification (glycosylation) It has

also been observed that upon nerve injury, axons of

the surviving IB4-positive neurons in the DRG may

sprout and form so-called perineuronal ring-shaped

structures around the larger diameter A-fibers [10]

This effect was interpreted as an anatomical basis for

the cross-excitation phenomenon that may underlie

allodynia [11] All of these reports suggest that the

IB4-reactive molecule may be important in pain

trans-mission However, the investigation of its functional

role was impossible because its identity remained

obscure

Here we describe experiments leading to the

identifi-cation of a molecule containing the IB4-binding

epi-tope We show that it is a protein which is enriched in

a membrane preparation obtained from spinal cord

tis-sue By means of biotinylated IB4 and Streptavidin–

agarose we extracted a macromolecule from a ‘light

membrane’ fraction which was identified by MALDI-TOF peptide mass fingerprinting, partial post source decay (PSD) sequencing and further experimental evi-dence We propose that the extracellular matrix pro-tein versican is an IB4-binding molecule in nerve tissue

Results

Enrichment of IB4-binding activity via subcellular fractionation

The central terminals of almost all nonpeptidergic C-fibers terminate in the substantia gelatinosa of the dorsal horn where they are connected to dorsal horn neurons This region is known to contain high levels of IB4-binding activity Various authors have suggested that the IB4 target molecule is a transmembrane- or a plasma membrane-associated glycoprotein [12,13] In our attempts to characterize the IB4-binding molecule,

we therefore focussed on neuronal membrane prepara-tions We fractionated pig spinal cord tissue by density-gradient centrifugation and analysed the frac-tions by PAGE and blotting, using an IB4-peroxidase conjugate to detect the IB4-binding molecule We found that one predominant IB4-binding entity was strongly enriched in the light membrane (probably containing axonal membranes) and synaptosomal frac-tions The apparent molecular mass of this component was > 250 kDa (Fig 1)

Fig 1 Isolectin B4 (IB4)-binding activity is enriched in the light membranes and synaptosome preparation Thirty micrograms of protein from different fractions of the synaptosome preparation were separated by SDS ⁄ PAGE [7.5% (w ⁄ v) gel] and electrophoreti-cally transferred to a nitrocellulose membrane The blot was devel-oped with IB4-peroxidase (IB4-PO) Lane 1, marker; lane 2, homogenate; lane 3, low-speed supernatant; lane 4, low-speed pel-let; lane 5, high-speed supernatant; lane 6, high-speed pelpel-let; lane

7, myelin; lane 8, light membranes; lane 9, synaptosomes; lane 10, mitochondria The highest IB4 reactivity (Arrow) is found in lanes 8 (light membranes) and 9 (synaptosomes).

Proof of the IB4-binding specificity

The lectin IB4 binds selectively to oligosaccharides

containing a terminal a-d-galactopyranosyl group

There are two options to prove the specificity of IB4

binding The first is destruction of the IB4 eptitope by

enzymatic treatment with a-galactosidase [14,15] and

the second is by competing with IB4 binding using

an appropriate sugar homologue Melibiose, an

a-d-galactopyranosyl glucoside, is known to be bound by

IB4 [4] We therefore used melibiose to analyse the

binding specifity As shown in Fig 2, this disaccharide

competes with IB4 binding, as detected by the

IB4-peroxidase (IB4-PO) assay, in a dose-dependent

man-ner Analysis of IB4 reactivity after a-galactosidase

treatment of light membranes and synaptosomes gave

a consistent result (data not shown) No IB4 binding

was detectable after enzymatic treatment

The IB4-binding glycoconjugate is a protein

In principle, the IB4-binding oligosaccharide can be

bound to proteins, lipids, or polymeric glycans In

order to analyse the nature of the IB4-binding

mole-cule, we incubated light membranes with proteinase K, which is known to digest the majority of proteins, leaving only oligopeptides behind As shown in Fig 3, proteinase K treatment reduced the IB4-binding capa-city dramatically (a 2-min incubation was sufficient

to degrade the IB4-binding molecule) The high-mole-cular-weight IB4-binding molecule therefore is a glyco-protein

Isolation of the IB4-binding glycoprotein

by affinity chromatography SDS was found to be the most effective detergent for extracting the IB4-binding glycoconjugate from light membranes or synaptosomes (data not shown) More-over, we found that IB4 reactivity, as detected by Western blotting, was not affected by the presence of SDS but was strongly dependent on Ca2+ ions [2,4] For these reasons we tried to isolate the IB4-binding glycoconjugate from SDS-solubilized pig spinal cord light membranes by means of biotinylated IB4 and Streptavidin–agarose in the presence of Ca2+ ions Binding in the presence of 0.1 mm CaCl2 and elution

Fig 2 Competition with melibiose Thirty micrograms of

synapto-somal protein was electrophoresed on SDS ⁄ PAGE [7.5% (w ⁄ v) gel]

and blotted to nitrocellulose The membrane was stained with

Ponceau S [0.1% (w ⁄ v) Ponceau S in 5% (v ⁄ v) acetic acid] and the

lanes were separated from each other by cutting the blotting

mem-brane into strips using a scalpel The strips were transferred to a

strip-box and blocked overnight with 1% BSA in NaCl ⁄ Tris (TBS).

The strips were incubated for 1.5 h at room temperature with

isolectin B4-peroxidase (IB4-PO) (1 : 500) in NaCl ⁄ Tris containing

0.1 m M CaCl 2 , 0.1 m M MnCl 2 , and 0.1 m M MgCl 2 , and increasing

concentrations of melibiose (lane 2, without melibiose; lane 3,

10 l M ; lane 4, 50 l M ; lane 5, 100 l M ; lane 6, 250 l M ; lane 7,

500 l M ; lane 8, 1 m M ; lane 9, 2 m M melibiose) Lanes 1 and 10,

marker The IB4 reactivity (Arrow) decreases with increasing

melibi-ose concentration.

Fig 3 The isolectin B4 (IB4)-binding molecule is proteinaceous Forty micrograms of protein from the light membrane fraction was combined with Proteinase K (0.1 mgÆmL)1) in 100 m M Na x H x PO 4 ,

pH 8, and incubated for different time-periods at 37
C The diges-tion was stopped by adding 4· sample buffer and 10-min incuba-tion at 95
C Samples were separated on SDS ⁄ PAGE [7.5% (w ⁄ v) gel] and blotted to nitrocellulose (A) Coomassie Brilliant Blue-stained gel: lane 1, native light membranes; lane 2, light mem-branes after a 2 min incubation with proteinase K; lane 3, light membranes after a 5 min incubation with proteinase K; lane 4, light membranes after a 15 min incubation with proteinase K; lane 5, marker (B) Isolectin B4-peroxidase (IB4-PO)-developed Western blot (the arrow indicates IB4-binding activity): lane 1, native light membranes; lane 2, light membranes after a 2 min incubation with proteinase K; lane 3, light membranes after a 5 min incubation with proteinase K; lane 4, light membranes after a 15 min incubation with proteinase K; lane 5, marker.

from IB4 with NaCl⁄ Pi(PBS) containing 2 mm EDTA

yielded a protein fraction that was analysed by

SDS⁄ PAGE, Coomassie Blue staining and blotting

using IB4 peroxidase (Fig 4) Only one IB4-binding

component was detected This was strongly enriched in

the EDTA-eluate (Fig 4, lanes 6 and 8) Both the

lec-tin blot with IB4 peroxidase (Fig 4, lane 6) and the

protein stain with Coomassie Blue, respectively (Fig 4,

lane 8), showed one predominant band corresponding

to an apparent molecular mass of > 250 kDa

Identification of the > 250 kDa glycoprotein

To identify the isolated glycoprotein, it was digested

in-gel with trypsin and analysed by MALDI-MS

(Fig 5) The protein detected is the versican splice

variant, V2 (database entry AAA67565; 23 tryptic

pep-tides covering 12% of the full-length protein; see also

Scheme 1) This result was confirmed by PSD

sequen-cing of two selected peptides, which perfectly matched

with sequences of the pig versican according to

data-base entry AAF19155.1 (Fig 6)

The data bank search indicated versican as the only

significant match Laminin and the light- and

medium-sized subunits of neurofilaments, which had been

previously reported to be IB4-binding molecules [16], were not supported by our peptide mass fingerprint

Versican and IB4-binding activity are co-enriched

by subcellular fractionation of spinal cord tissue

It is known that the association of versican with the plasma membrane is mediated via binding to hyaluro-nan [17] Hyalurohyaluro-nan is a polymeric glycan which can

be specifically digested with hyaluronidase [18] In order to analyse whether versican, like the IB4-binding activity, is enriched in the same subcellular fractions,

we treated the insoluble part of each fraction of a syn-aptosome preparation with hyaluronidase We subse-quently analysed the extract by Western blotting by using a mAb, anti-(glial hyaluronate-binding protein) (anti-GHAP), which is known to detect all splice vari-ants of versican [19] As shown in Fig 7A, versican was detected in nearly all fractions, but is – like the IB4-binding glycoprotein – strongly enriched in the light membranes (see also Fig 1) Additional signals, detected with anti-GHAP, of around 66 kDa probably represent GHAP itself, the N-terminal part of versican [19–21]

To confirm that versican is the IB4-binding glyco-protein, we stripped the Western blot shown in Fig 7A and developed it with IB4 peroxidase (Fig 7B) The same band of > 250 kDa became vis-ible The 66 kDa band, probably representing GHAP, did not show up in the stripped and IB4-peroxidase developed blot, obviously because the IB4-binding epi-tope is not located within the N-terminal portion of versican

To address the posibility that the laminin b2 chain

as well as neurofilament proteins, which have been recently reported to bind IB4 [16], account for the IB4 reactivity that we observed within the fractions of por-cine spinal cord and especially within the hyaluroni-dase-released fraction, we tested the respective fraction for anti-neurofilament and anti-laminin immunoreac-tivity As positive controls for the immunoreactivity for these proteins, we used a neurofilament preparation from porcine spinal cord and commercially available laminin 1 from the Englebreth Holm-Swarm sarcoma, which is known to bind IB4 and to contain both IB4-binding b-chains [22] As shown in lane 3 of Fig 8, neither neurofilament proteins nor laminin were detec-ted within the protein fraction released from light membranes by hyaluronidase treatment However, the typical IB4-reactive signal that we demonstrated to be assignable to versican was well detected Thus, this IB4-reactive glycoprotein is neither a neurofilament protein nor laminin, in agreement with our other

Fig 4 Enrichment of the isolectin B4 (IB4)-binding glycoconjugate

with biotinylated IB4 and Streptavidin–agarose IB4-bound proteins

were specifically eluted by Ca2+ withdrawal with NaCl ⁄ P i (PBS)

containing 2 m M EDTA, 0.5% (w ⁄ v) SDS (lane 6 and lane 8)

Non-specifically bound proteins were eluted with 4· sample buffer

(lanes 7 and 9) All fractions were concentrated by using 30 kDa

cutoff microconcentrators and electrophoretically separated by

SDS ⁄ PAGE [7.5% (w ⁄ v) gel] Lanes 1–7 of the gel were blotted

onto nitrocellulose and developed with isolectin B4-peroxidase

(IB4-PO), as described above Lanes 8–10 of the gel were stained with

Coomassie Brilliant Blue R250 Lane 1, marker; lane 2, 15 lL of

supernatant of the extracted light membranes; lane 3, 15 lg of

pro-tein of the light membranes after extraction with SDS; lane 4,

15 lg of protein of the extracted (but not precipitated) proteins;

lane 5, combined washing fractions; lane 6, half of all proteins

elu-ted under Ca2+withdrawal; lane 7, half of all proteins eluted with

4· sample buffer; lane 8, half of all proteins eluted under Ca 2+

with-drawal; lane 9, half of all proteins eluted with 4· sample buffer;

lane 10, marker.

experiments which demonstrate that the glycoprotein is

versican

The IB4-binding molecule is

co-immunoprecipi-tated with versican by an antibody to GHAP

The possibility exists that two or more molecular

species co-migrate in the >250 kDa electrophoretic

band We therefore performed imunoprecipitation with

an anti-versican immunoglobulin and developed the

Western blot of the precipitate with IB4 peroxidase

As shown in Fig 9, only one IB4-binding glycoprotein

was detected This was strongly enriched in the

precipi-tate (compare lanes 8 and 9) The apparent molecular

mass of the glycoprotein was again > 250 kDa The inverse experiment – precipitation with IB4-biotin and Streptavidin–agarose and detection with the anti-versican immunoglobulin – gave the corresponding result (data not shown) This suggests that versican and the IB4-binding entity are the same molecule

Discussion

The IB4-binding epitope clearly is more than just an anatomical marker for a subpopulation of nociceptive C-fibers (see the Introduction and the references therein) The aim of our study was to elucidate the molecular identity of the glycoconjugate carrying the

Fig 5 Identification of versican by MALDI-MS, peptide mass fingerprint The affinity-purified protein (Fig 4, lane 8, indicated by the question mark) was digested in-gel with trypsin and tryptic peptides were analysed by MALDI-MS Upper figure: peptide mass fingerprint; lower figure, list of tryptic peptides that could be matched with peptides of versican V2 (database entry AAA67565.1).

terminal a-d-galactosyl moiety, which renders a subset

of nociceptive C-fibers IB4 positive The idea behind

this was, of course, to provide insight into the special

role of these fibers in pain transmission Here we

des-cribe a first step towards elucidation of possible

func-tion We propose that the extracellular matrix protein,

versican, is the glycoconjugate targeted by IB4 The

evidence presented includes the following, namely that

the IB4-binding molecule is a protein enriched during

subcellular fractionation in synaptosomal and light

(axonal) membrane fractions Affinity chromatography

using biotinylated IB4 and streptavidin agarose beads

extracted from pig spinal cord a protein of high

relat-ive molecular mass (> 250 kDa) which was identified

by MALDI-MS (peptide mass fingerprinting and PSD

sequencing of two peptides) as versican As an

addi-tional criterion for the specificity of the binding to the

affinity matrix, we used the Ca2+ dependence of the

binding of the glycoconjugate(s) to the lectin Only one

protein, namely the V2-like variant of versican, was

recovered in this way from the affinity matrix No

other protein matched the peptide mass spectrum

sig-nificantly In particular, the light and medium subunits

of the neurofilament triad, as well as laminin b2, which

were recently proposed to be IB4-binding entities in

DRGs [16], could not be detected in the protein

frac-tion that bound in a Ca2+-dependent manner to IB4

Moreover, although neurofilaments and laminin were

detectable within the light membrane and synaptosome

fractions of porcine spinal cord, they never stained positive for IB4-reactivity in our hands and they were distributed over the subcellular fractions in a pattern that deviated from the distribution pattern of the IB4-reactive glycoprotein (data not shown)

Binding of peroxidase-linked IB4 to the > 250 kDa protein could be competitively prevented by melibiose,

an a-d-galactopyranoside-containing disaccharide, and both IB4-peroxidase and an anti-versican (anti-GHAP) immunoglobulin bound to this high-molecular-mass protein Immunoprecipitation with an anti-versican immunoglobulin yielded both versican and IB4 affinity Moreover, hyaluronidase treatment of light membranes not only released versican into the supernatant frac-tion, it also released an IB4-positive molecule that co-migrated identically with versican in SDS⁄ PAGE Neither neurofilaments stained positive for IB4, nor was the known IB4-positive protein laminin 1 present

in the fraction released from light membranes by hy-aluronidase treatment

Taken together, these data provide firm evidence that versican is the first unambiguously identified pro-tein (not necessarily the only one) accounting for the IB4 stain in the spinal cord (and probably in DRGs) described by anatomists

What are the implications of our findings with respect to C-fibre transmission of nociceptive informa-tion? What role might versican play with respect to the properties of C-fibre type nociceptors?

Scheme 1 Amino acid sequence of the

human versican splice isoform V2 (database

entry AAA67565.1): Peptides of the isolectin

B4 (IB4)-positive porcine versican that also

matched the human versican V2 are shown

in bold, peptides identified by post source

decay (PSD) are in bold and underlined.

Note that three of the matched peptides

correspond to the glycosaminoglycan (GAG)

a-domain (amino acids 348–1335) Peptides

that correspond to the GAG b-domain were

not detected.

[Abs Int * 1000]

4.50

4.25

NGFDQCDYGWLLDASVR

3.75

3.50

3.25

3.00

2.75

2.50

2.25

2.00

1.75

1.50

1.25

1.00

0.75

0.50

0.25

0.00

-0.25

[Abs Int * 1000]

7.0

6.5

6.0

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

m/z

m/z

1200

Fig 6 Post source decay (PSD) fragment ion spectra of two selected tryptic peptides: peptides of (M + H + ) + ¼ 1314.71 Da and of (M + H+)+¼ 2015.91 Da that had been observed directly within the peptide mass fingerprint (Fig 5) or after microfractionation of the tryptic digest by using desalting of the peptides by C18-Zip Tips followed by sequential elution of the peptides by increasing concentrations of organic solvent were analysed by PSD Both fragment ion spectra were consistent with the proposed peptide sequences derived from por-cine versican, according to database entry AAF19155.1.

Versican is an extracellullar matrix proteoglycan of the chondroitin sulphate proteoglycan subfamily of versican, brevican, aggrecan and neurocan Four dif-ferent splice variants of versican are currently known [21,23–25] and additional isoforms may exist [26] The

A

B

Fig 7 Co-enrichment of versican and isolectin B4 (IB4) reactivity

by subcellular fractionation (A) Subcellular fractions probed for

antibody to glial hyaluronate-binding protein (anti-GHAP) reactivity.

One milligram of protein from different fractions of a

synapto-some preparation was pelleted by ultracentrifugation at

436 000 g The pellets were resuspended in protease inhibitor

and 0.15 M NaCl containing 0.05 M NaxHxPO4, pH 5.3 and

homo-genized with a glass ⁄ glass homogenizer (0.1 mm clearence).

Two-hundred and fifty micrograms of each fraction was treated

with 50 units of hyaluronidase for 2 h at 37
C Fifteen

micro-grams of the concentrated (3 kDa cut-off microconcentrators)

pro-tein extract from each fraction was separated by SDS ⁄ PAGE

[7.5% (w⁄ v) gel], electrophoretically blotted to nitrocellulose and

developed by using monoclonal anti-GHAP [12C5; 1 : 250 dilution

in NaCl⁄ Tris (TBS) containing 5% (w ⁄ v) dry milk and 0.1% (w ⁄ v)

Tween 20] Lane 1, homogenate; lane 2, low-speed supernatant;

lane 3, low-speed pellet; lane 4, high-speed supernatant; lane 5,

high-speed pellet; lane 6, myelin; lane 7, light membranes; lane

8, synaptosomes; lane 9, mitochondria; lane 10, marker The top

arrow indicates the IB4-binding versican, the arrow at 66 kDa

indicates GHAP, the N-terminal portion of versican (B) Blotted

proteins corresponding to Fig 7A, probed for IB4-reactivity The

Western blot from Fig 7A was stripped by a 30 min incubation

at 40
C with 2% (w ⁄ v) SDS, 10 m M b-mercaptoethanol in

62.5 m M Tris ⁄ HCl, pH 6.7 and washed extensively with NaCl ⁄

Tris The membrane was blocked by overnight incubation in

NaCl⁄ Tris containing 1% BSA and developed with IB4-PO, as

described in the Materials and methods Lane 1, homogenate;

lane 2, low-speed supernatant; lane 3, low-speed pellet; lane 4,

high-speed supernatant; lane 5, high-speed pellet; lane 6, myelin;

lane 7, light membranes; lane 8, synaptosomes; lane 9,

mito-chondria; lane 10, marker The arrow indicates the IB4-reactive

signals.

Fig 8 Neither laminin nor neurofilaments account for the isolectin B4 (IB4) reactivity in the protein fraction released from light mem-branes by hyaluronidase Left, proteins of a neurofilament prepar-ation (lane 2), hyaluronidase-released light membrane proteins (lane3), and commercially available laminin 1 (lane 4) were separ-ated by SDS ⁄ PAGE and visualized by Coomassie staining Lane 1, molecular mass marker Right: proteins according to the gel shown

on the left were transferred to a nitrocellulose membrane and probed for various immunoreactivities by Western blot or lectin blot analysis Light membrane hyaluronidase extract (lane 3) was probed for IB4 reactivity [by using isolectin B4-peroxidase (IB4-PO)], anti-laminin immunoreactivity (anti-L1), and anti-neurofilament NF-M reactivity (anti-NF-M) Although IB4 reactivity is present (arrow), no reactivity for laminin or NF-M was observed NF-M was detected in

a neurofilament preparation (lane 2), but no IB4 reactivity was observed in this preparation Commercially available laminin 1 was readily detected by using laminin-specific antibody, and the b1 and b2 chains stained positive also for IB4 reactivity (lane 4).

Fig 9 Co-immunoprecipitation of versican and isolectin B4 (IB4) reactivity Western blot using isolectin B4-peroxidase (IB4-PO) for detection of IB4 reactivity (arrow) Lane 1, marker; lane 2, 30 lg of light membranes; lane 3, 10 lg of extracted light membranes; lane

4, 10 lg of hyaluronidase extract; lane 5, 5 lg of nonprecipitated proteins; lane 6, 5 lg of protein from washing step 1; lane 7, 5 lg

of protein from washing step 2; lane 8, half volume of the eluted proteins; lane 9, 30 lg of light membranes; lane 10, Marker.

versican variant V2 is the dominant splice variant of

versican in neuronal tissue and the one that gave the

best match with our mass spectrometric data

Versican isoforms are expressed in a variety of

tis-sues [27], including in the extracellular matrix of the

brain [28] Known functional effects of versican in the

nervous system were reported to be the impairment of

axonal or neurite growth by versican V2 [29,30], and

the promotion of neurite outgrowth and neuronal

dif-ferentiation in vitro by a different isoform, versican V1

[31] Intriguingly, versican shares the structural

fea-tures of the other chondroitin sulphate proteoglycans,

providing a binding module for the interaction with

hyaluronan and a C-type lectin domain though which

the interaction with other extracellular matrix

mole-cules may occur This protein family was thus also

called ‘hyalectans’ or ‘lecticans’ and have been

sugges-ted to form ternary complexes with hyaluronan and

tenascin R [32] These complexes contribute to the

molecular makeup of perineuronal nets, extracellular

matrix-based structures that surround neurons and

that create, e.g barriers that shield the neurons from

the outside and also prevent axonal sprouting [33,34]

Notably, there are reports of neuronal synthesis of at

least one type of lectican, namely aggrecan [35,36]

Interestingly, in the latter study the authors even

dem-onstrated the specific expression of differentially

gly-cosylated variants of the same core protein by

different subsets of neurons, even in some cases

restric-ted to a particular lamina within the gray matter of

the spinal cord [36] We propose that this may also

apply to the IB4-positive versican variant

There are indeed reports that versican can be a

com-ponent of perineuronal nets [37] In the case of versican

V2, however, there has been no previous report of a

neuronal expression, but V2 expression has been

suggested to be assignable to oligodendrocytes and

Schwann cells [29,30] This contrasts with

immunohisto-chemical data on the expression of IB4 reactivity in

the dorsal root ganglion, which appears, owing to the

unambiguous stain of neuronal cell bodies including the

Golgi apparatus, clearly neuronal [10,12] Therefore,

an immunohistochemical stain for versican within the

spinal cord and within the DRG should resolve this

problem In summary, we suggest that there is a

versi-can V2-like or V2-related versiversi-can variant that is

modified by IB4-reactive carbohydrates and that is

syn-thesized by neurons

It is an exciting feature of potential high medical

rele-vance that the IB4-reactive moiety underlies dynamic

changes in experimental paradigms of neuropathic pain,

namely a loss of IB4 reactivity within the dorsal horn of

the spinal cord and within the DRG, that can be

allevi-ated or reversed by GDNF [7,8] Moreover, nerve injury can lead to the formation of IB4-reactive basket-like structures that emanate from IB4-positive C-fibres and that surround the cell bodies of large-diameter A-neu-rons within the DRG [10] As nerve injury renders DRG neurons hyperexcitable, afferent impulses invading the somata of A-neurons may initiate ectopic discharges in the surrounding C-fibers of these basket-like structures This cross-excitation phenomenon between A- and C-fibers in the DRG is discussed as a candidate for the development of allodynia in neuropathic pain [11] Our identification of versican is obviously remarkable in the light of the data of Li & Zhou [10] who described the above-mentioned basket-like structures that emanate from C-fibres to surround A-cell bodies, because it is obvious that these structures may be at least partially made up of extracellular matrix proteins Our findings thus open the door for further investigations of these phenomena

There is another emerging set of evidence for the important role of extracellular matrix molecules in pain transmission There was a recent report that pep-tide fragments from two other important extracellular matrix proteins, laminin and fibronectin, inhibited hyperalgesia caused by prostaglandin E2 and epineph-rine, respectively Both extracellular matrix proteins are involved in signalling through integrins, and mAbs directed against the b1-integrin subunit, as well as a knockdown of b1-integrin expression, inhibited inflam-matory hyperalgesia [38] The C-terminal domain of versican has been shown in pull-down and co-immuno-precipitation assays to bind to b1-integrin and to regu-late glioma cell adhesion and free radical-induced apoptosis [39] Moreover, Wu et al recently reported that PC12 cell differentiation and neurite outgrowth was dependent on integrin signalling and was blocked

by application of an anti-b1 integrin immunoglobulin [31] Versican V2, however, exerted no such effects on PC12 cell differentiation as compared to the V1 splice variant

Taken together, our finding that versican with most similarity to versican V2 among the known versican variants is the principal IB4-binding protein in the spi-nal cord (and probably also in the DRG) fuels a signi-ficant new aspect into the investigation of perineuronal nets and provides long sought-after information in the ongoing struggle to elucidate the molecular basis of neuropathic pain

Experimental procedures

All substances and biochemicals were of the highest purity commercially available The mAb anti-GHAP, developed

by R A Asher [18], was obtained from the Developmental

Studies Hybridoma Bank founded under the auspices of the

National Institute of Child Health and Human

Develop-ment (NICHD) and maintained by the University of Iowa

(Department of Biological Sciences, Iowa City, IA, USA)

Subcellular fractionation

Pig spinal cords were obtained from a local slaughterhouse,

separated from the meninges and taken to the laboratory in

liquid nitrogen All procedures, including all centrifugation

steps, were carried out at 4
C

Preparation of synaptosomes was based on the method

established by Gray & Whittaker [40], with some minor

modifications: Briefly, frozen pieces of pig spinal cord were

homogenized in homogenization buffer (10 mm Hepes,

pH 7.4, 1 mm EDTA, 320 mm sucrose) containing a

prote-ase inhibitor cocktail (Roche Diagnostics) Homogenization

was performed with a motor-driven glass-Teflon

homo-genizer (0.2 mm clearance) by 12 up-and-down strokes at

800 r.p.m The homogenate was centrifuged at 1000 g for

10 min The supernatant (S1) was removed and placed on

ice The pellet (P1) was resuspended in homogenization

buf-fer and homogenized again as described above The

homo-genate was centrifuged at 1000 g for 10 min The resulting

pellet (P1¢, cell debris and nuclei) was discarded The

super-natant (S1¢) was combined with supersuper-natant S1 and

centri-fuged at 12 000 g for 15 min The supernatant (S2) was

discarded, the pellet (P2, crude membrane fraction) was

resuspended in homogenization buffer and homogenized

again with six up-and-down strokes at 800 r.p.m using the

motor-driven glass-Teflon homogenizer The homogenate

was centrifuged at 12 000 g for 20 min The supernatant

(S2¢) was discarded, the pellet (P2¢) was resuspended with

0.32 mm sucrose in 5 mm Tris⁄ HCl, pH 8.1, layered onto a

discontinuous sucrose gradient (1.2 ⁄ 1.0 ⁄ 0.85 m sucrose)

and centrifuged at 85 000 g for 2 h The resulting

subcellu-lar fractions were harvested using a widened Pasteur

pip-ette Myelin accumulated at the 0.32 ⁄ 0.85 m sucrose

interface, light membranes at the 0.85 ⁄ 1.0 m sucrose

inter-face, synaptosomes at the 1.0 ⁄ 1.2 m sucrose interface and

mitochondria at the bottom of the centrifugation tube All

fractions were diluted to a final sucrose concentration of

less than 0.3 m with protease inhibitor-containing NaCl⁄ Pi

(PBS) (137 mm NaCl, 2.7 mm KCl, 8 mm Na2HPO4,

1.5 mm KH2PO4, pH 7.4), centrifuged at 12 000 g for

10 min, and recovered from the bottom of the tube with

protease inhibitor containing NaCl⁄ Pi The protein

concen-tration was determined using the Bradford assay [41] with

BSA (type V; Pierce) as standard

Western blot analysis of IB4-binding activity

Samples (30–40 lg of protein) were combined with sample

buffer [final concentration: 62.5 mm Tris⁄ HCl, pH 6.8, 3%

(w⁄ v) SDS, 10% (v ⁄ v) glycerol, 5% (v ⁄ v) b-mercaptoetha-nol, 0.025% (w⁄ v) Bromophenol blue], heated for 10 min

at 60
C and electrophoresed on 7.5% (w ⁄ v) polyacryl-amide gels in 25 mm Tris containing 192 mm glycine and 0.1% (w⁄ v) SDS [42] Proteins were electrophoretically transferred to nitrocellulose by using the semidry method [transfer time was 2 h at 1.5 mAÆcm)2, with 47.9 mm Tris, 38.9 mm glycine, 0.038% (w⁄ v) SDS and 20% (v ⁄ v) meth-anol] Blots were blocked overnight with 1% (w⁄ v) BSA in NaCl⁄ Tris (Tris-buffered saline; 20 mm Tris, 150 mm NaCl), incubated for 1.5 h at room temperature with

IB4-PO (Sigma, 1 : 500) in NaCl⁄ Tris containing 0.1 mm CaCl2, 0.1 mm MnCl2, and 0.1 mm MgCl2, and washed three times with NaCl⁄ Tris-T [NaCl ⁄ Tris containing 0.1% (v ⁄ v) Tween 20] containing 0.1 mm CaCl2, 0.1 mm MnCl2, and 0.1 mm MgCl2 Lectin-reactivity was visualized by using the enhanced chemiluminescence (ECL) detection system (Amersham Biosciences)

Affinity chromatography

A 1.25 mg sample of freshly prepared light membranes was extracted for 1 h at room temperature in 1% (w⁄ v) SDS containing 0.1 mm CaCl2, 0.1 mm MnCl2, 0.1 mm MgCl2, and protease inhibitor-containing NaCl⁄ Tris Extracted proteins were separated by centrifugation at 10 000 g for

10 min, combined with 25 lL (50 lg) of IB4-biotin (Sigma) and incubated for 16 h at 4
C under continuous rotation The SDS concentration was reduced to 0.5% by adding an equal volume of 0.1 mm CaCl2, 0.1 mm MnCl2, 0.1 mm MgCl2, and protease inhibitor containing NaCl⁄ Tris IB4-labelled proteins were affinity-bound by adding 750 lL of Streptavidin–agarose (Sigma) in 0.01 m NaxHxPO4, pH 7.2, containing 0.15 m NaCl and 0.02% (w⁄ v) Na3N The sam-ple was incubated under vigorous shaking for 3 h at 4
C Beads were centrifuged and washed twice under vigorous shaking with 0.1 mm CaCl2, 0.1 mm MnCl2, 0.1 mm MgCl2 and protease inhibitor containing NaCl⁄ Tris IB4-captured proteins were eluted from the beads by using 0.5% (w⁄ v) SDS and 2 mm EDTA containing NaCl⁄ Pi Nonspecifically bound proteins were eluted with 4· sample buffer according

to Laemmli [42] All fractions were concentrated by using micro concentrators with a molecular weight cut-off of

30 kDa (Amicon), electrophoresed by SDS⁄ PAGE [7.5% (w⁄ v) gel] and visualized by staining with Coomassie Brilli-ant Blue or blotted onto nitrocellulose and analysed with IB4-PO, as described above

MALDI-MS The protein of interest was digested in-gel with trypsin according to standard protocols [43] The MALDI-MS measurements were performed using dihydroxy benzoic acid (Sigma) or a-cyano 4-hydroxy cinnamic acid (Bruker Dal-tonics, Leipzig, Germany) as matrix substances A Bruker