(Received for publication, June 5, 1996, and in revised form, October 11, 1996)
From the Department of Veterinary Medical Chemistry, Large disulfide-stabilized proteoglycan complexes
were previously shown to be synthesized by the epidermis of axolotl
embryos during stages crucial to subepidermal migration of neural crest cells. We now show that the complexes contain PG-M/versican-like monomers in addition to some other component with low buoyant density.
Metabolically 35S-labeled proteoglycans were extracted from
epidermal explants and separated by size exclusion chromatography and
density equilibrium gradient centrifugation. The complexes, which elute
in the void volume on Sepharose CL-2B, were recovered at buoyant
density 1.42 g/ml in CsCl gradients, whereas the monomer proteoglycans,
which could only be liberated from the complexes by reduction, had a higher buoyant density (1.48 g/ml). The native complexes did not aggregate with hyaluronan. The purified complexes reacted with antibodies against a portion of a cloned PG-M/versican-like axolotl proteoglycan. These antibodies were found to stain the subepidermal matrix of axolotl embryos, suggesting that the proteoglycan complexes are encountered by neural crest cells during subepidermal migration. From Western blot analysis, the core protein of the PG-M/versican-like monomers was found to be of similar size ( The role of proteoglycans (PGs)1 in
the modulation of cell-matrix interaction and cell migration depends on
their structure and localization (1-3). Interstitial PGs that carry
chondroitin/dermatan sulfate (CS/DS) chains have been implicated to
have such modulatory roles by several studies (4-18). In this regard,
since both the core protein (7, 18-19) and the glycosaminoglycan
moiety (15, 18, 20) appear to be of functional importance, the
structure of the intact PG may be crucial (9, 18). In addition to the inherent potential for heterogeneity in the carbohydrate component (21), the core proteins of PGs in the PG-M/versican family, for
instance, may exist as several splice variants (22-25) with potentially diverse expression patterns and functions during various stages of the life cycle of an animal.
The importance of the extracellular matrix in the regulation of
subepidermal neural crest cell migration in the axolotl embryo was
demonstrated by transplantation experiments in which epidermal grafts
or microcarriers with matrix adsorbed in situ from
developmentally more advanced embryos were found to trigger the onset
of premature migration (26). Ultrastructural studies of the matrix
after fixation in the presence of cetylpyridinium chloride (27) or ruthenium red (28-30) revealed an abundance of granular proteoglycan precipitates along collagen fibrils. Interestingly, in embryos of the
white mutant axolotl, in which subepidermal migration of the neural
crest-derived pigment cells is defective (28-29, 31), the subepidermal
matrix was significantly less granulated (27, 29-30) suggesting a
potential role of the PGs in the matrix.
In an attempt to isolate and characterize the PGs that might be
encountered by migrating neural crest cells in the axolotl embryo, a
previous study (32) took advantage of the possibility to utilize
epidermal explants for the metabolic radiolabeling of PGs with
[35S]sulfate. Dorsal trunk epidermis was explanted at a
premigratory stage and was maintained in tissue culture together with
radiolabel for a time period corresponding to the transition from a
nonpermissive matrix to one that permits subepidermal migration of
neural crest cells. It was found that the major PGs synthesized were
large disulfide-stabilized PG complexes with monomer PGs in the size range of aggrecan or PG-M/versican but with unusually large
(Mr Embryos of the Mexican axolotl (Ambystoma
mexicanum, Urodela, Amphibia) were obtained from the Indiana
University Axolotl Colony, Bloomington, IN, and from the Department of
Zoology, Uppsala University. Before manual decapsulation, eggs were
sterilized in 70% ethanol and rinsed in modified Steinberg's solution
(MS), composed of 58 mM NaCl, 0.67 mM KCl, 0.34 mM CaCl2, 0.83 mM
MgSO4, 4.6 mM HEPES, pH 7.4 (33), supplemented
with 0.05% penicillin/streptomycin. All storage, handling, and
manipulations of embryos were done in sterile MS. Embryos were staged
according to Bordzilovskaya et al. (34).
Using sharp tungsten needles, the epidermis was
carefully removed from the dorsal trunk (the area between the levels of
the gills and anus overlying the somites and neural tube/crest) of stage 30 embryos (around the onset of neural crest cell migration). The
epidermal explants were incubated in MS containing 2.5 mCi/ml Na[35S]SO4 (carrier-free; DuPont NEN). After
incubation for 20 h at 20 °C, corresponding to the time between
stages 30 to 35 (as determined by the parallel incubation of control
embryos), the labeling medium was removed and the tissue was dissolved
in extraction buffer (about 400 µl per 50 explants) containing 4 M guanidine HCl, 50 mM sodium acetate, pH 5.8, and 0.2% Triton X-100, together with 1 mM
N-ethylmaleimide (NEM), 100 mM Extracted 35S-labeled PGs were
separated by chromatography on 1.2 × 100-cm columns of Sepharose
CL-2B (Pharmacia Biotech Inc.) eluted with extraction buffer at
4 °C. An aliquot of each fraction was subjected to liquid
scintillation analysis. The recovery of 35S radioactivity
was about 65%. Size-fractionated PGs were pooled as indicated in Fig.
1A, and 0.48 g/ml CsCl was added to give an initial density
of 1.42 g/ml before the samples were separated according to buoyant
density in self-generated CsCl gradient by ultracentrifugation at
100,000 × g for 72 h. The specific gravity (
For reduction of
the HMPG complexes, samples were dialyzed into a buffer of 6 M guanidine HCl, 100 mM Tris, pH 8, 0.1%
Triton X-100, and the mixture of protease inhibitors excluding NEM.
Dithiothreitol was added to a final concentration of 50 mM,
and the sample was incubated for 16 h at room temperature, after
which reduced SH groups were alkylated by addition of iodoacetamide to
a final concentration of 62.5 mM and incubated for another
1 h.
Alkaline
liberation of glycosaminoglycan side chains was achieved by adding 1 volume of 0.1 M NaOH, 0.6 M NaBH4
followed by incubation at room temperature for 24 h, after which
the samples were acidified, neutralized, and dialyzed against water.
The chain length of the glycosaminoglycans was assessed by
chromatography on a calibrated Sepharose CL-4B (Pharmacia) column as
described previously (32). Blue dextran (Pharmacia) and
N-2,4-dinitrophenyl- To obtain large size
hyaluronan, HealonTM (Pharmacia, Sweden) was
chromatographed on Sepharose CL-2B and analyzed by the carbazole method
(36), after which the void volume fraction was pooled and used for the
aggregation experiment. Aggrecan from bovine nasal cartilage was kindly
prepared by Dr. Erik Unger and quantitated by the carbazole method
(36). 35S-Labeled HMPG complexes or IMPGs were mixed with
10 µg of hyaluronan and 1 mg of aggrecan in dissociative buffer.
After dialysis into an associative buffer containing 0.5 M
sodium acetate, pH 5.8, 0.1% Triton X-100, and the protease inhibitor
mixture, samples were analyzed by CsCl gradient centrifugation or
chromatography on 0.8 × 100-cm columns of Sepharose CL-2B in
associative buffer. Samples without addition of carrier
aggrecan-hyaluronan complexes were used as negative controls. As a
positive control, 3H-labeled hyaluronan was centrifuged in
associative CsCl gradients with and without added aggrecan. The
3H radioactivity was recovered at Acid
treatment was done by titrating the sample with acetic acid to pH 3.8, after which the samples were incubated for 1.5 h at +4 °C. For
digestion with collagenase (form III; Advanced Biofactures, Lynnbrook,
NY), samples were dialyzed into a digestion buffer of 50 mM
Tris-HCl, pH 7.5, 0.36 mM CaCl2, and the
mixture of protease inhibitors excluding EDTA. To test the
effectiveness and specificity of the collagenase treatment, collagen
and fibronectin (kindly provided by Dr. Staffan Johansson, Department
of Medical Chemistry, Uppsala University) were treated in parallel
under identical conditions as positive and negative controls,
respectively, and analyzed by SDS-PAGE. Under these conditions,
collagen was completely degraded while fibronectin was left intact
(result not shown). Hyaluronidase (Seikagaku Co., Japan) treatment (5 turbidity reducing units/ml) was done in a buffer with 150 mM NaCl, 20 mM sodium acetate, pH 5.8, 0.1%
Triton X-100, with protease inhibitors at 37 °C for 3 h. As a
positive control of enzyme activity, 3H-labeled hyaluronan
was treated in parallel under identical conditions and analyzed by
chromatography on a Superose 6 HPLC column.
Before digestion with
chondroitinase ABC (protease-free; Seikagaku Co., Japan), the purified
PG samples were dialyzed into the appropriate buffer. Digestion with
chondroitinase ABC (0.1 units/ml) was done at 37 °C for 3 h in
50 mM Tris, pH 8, 30 mM NaAc, 0.1% Triton
X-100, with the mixture of protease inhibitors. As a positive control
of enzyme activity, aggrecan was digested under identical conditions
and analyzed by SDS-PAGE and immunoblotting.
For slot blotting, the chondroitinase
ABC-treated samples were diluted to <0.05% Triton X-100 in 4 M guanidine-HCl/TBS (10 mM Tris, pH 8, and 150 mM NaCl), serially diluted, and immobilized on a
nitrocellulose membrane (Schleicher & Schuell) using a Bio-Dot SF
Microfiltration Apparatus (Bio-Rad). For Western blotting, samples were
precipitated with 10% trichloroacetic acid, rinsed in ethanol:diethyl
ether (1:1), briefly dried, and dissolved in sample buffer. Reduced and
alkylated samples were subjected to SDS-PAGE in 7.5% homogeneous
PhastGels for 130 V-h followed by semi-dry electrotransfer for 30 min
onto nitrocellulose using a Pharmacia PhastSystem. Prestained SDS-PAGE
Standards (high range; Bio-Rad) and laminin (detected by Coomassie
staining of parallel gels) were used as molecular weight markers. Blots
were blocked with 5% non-fat milk in Tris-buffered saline, 0.1% Tween
(TBS-T), washed, incubated with primary antibodies (see below) in
TBS-T for 1 h at room temperature or overnight at +4 °C,
washed, incubated with horseradish peroxidase-linked anti-mouse Ig or
anti-rabbit Ig (Amersham Corp.; secondary antibodies diluted 1:5000)
for 30 min at room temperature, washed, reacted with enhanced
chemiluminescence reagent (Amersham Corp.) according to the
manufacturer's instructions, and detected by exposure to Fuji RX x-ray
film.
Mouse
monoclonal antibodies 5D4 (ascites fluid; Seikagaku) against keratan
sulfate were used diluted 1:200 in TBS-T. Mouse monoclonal antibodies
1B5, 2B6, and 3B3 (ascites fluid; ICN) against the unsulfated,
4-sulfated, and 6-sulfated, respectively, unsaturated "stubs"
generated by chondroitinase ABC digestion of chondroitin sulfate (37)
were used as a mixture at a dilution of 1:200 in TBS-T with respect to
each antibody.
A portion of the CS attachment
domain of a partially cloned axolotl PG-M/versican
homologue2 was selected for expression as a
fusion protein with glutathione S-transferase; the stretch
of 80 amino acid residues
(SAAEHAEGESDVTETKSPMTPLTAVETDEERQDVSTAYAELVSHSIRQSVTEIQDISQATYIESETAAKITPEDQTQKPS) lacks any apparent consensus signal for glycanation or
N- and O-linked glycosylation and shares no
apparent homology with other sequences obtained from the National
Center for Biotechnology Information (World Wide Web at URL
http://www3.ncbi.nlm.nih.gov). The corresponding 240-base pair
nucleotide sequence, to be cloned in the correct reading frame into the
BamHI/EcoRI site of the pGEX-3 (Pharmacia)
expression vector, was amplified from a partial cDNA
clone2 of the axolotl PG (AxPG) by polymerase chain
reaction. After ligation of the polymerase chain reaction product into
pGEX-3 and transfection of Escherichia coli JM83, a
recombinant clone with the correct insert was selected after
sequencing. The recombinant plasmid was cultured in E. coli
UT5600, and after induction with isopropyl
Rabbits were immunized intramuscularly with the purified
AxPG fusion protein using standard procedures, and the production of
antibodies was monitored by enzyme-linked immunosorbent assay. Anti-AxPG-GST antibodies (0.3 mg/ml) were obtained by affinity chromatography on AxPG-GST-coupled Sepharose 4B. The antibodies were
used in immunoblotting (30 µg/ml in TBS-T) and immunofluorescence (as
described below). As a control, similarly prepared affinity-purified antibodies against GST were used under the same conditions as the
anti-AxPG-GST antibodies.
Axolotl embryos of stage 32 or epidermal
explants of stage 30 embryos after 20 h incubation were fixed
overnight (embryos) or 15 min (epidermal explants) in 3%
paraformaldehyde, 0.1 M phosphate buffer, pH 7.4. For whole
embryos, fixative penetrance was facilitated by cutting off the head
and tail. The fixed embryos and epidermal explants were rinsed in
phosphate-buffered saline (PBS), dehydrated, embedded in Paraplast, and
sectioned transversely. Sections were dewaxed, rehydrated, washed in
PBS, and subjected to chondroitinase ABC digestion (1 unit/ml) in 50 mM Tris, pH 8, 30 mM NaAc for 30 min at
37 °C. The incubation with primary antibodies (10 µg/ml in PBS
containing 0.1% bovine serum albumin) was performed overnight at
4 °C. After washing in PBS, the sections were incubated with fluorescein isothiocyanate-conjugated secondary anti-rabbit IgG antibodies (diluted 1:200 in PBS, 0.1% bovine serum albumin;
Sigma) for 1 h at room temperature, after which
the immunofluorescence was visualized in a fluorescence microscope. As
control, anti-AxPG antibody binding was competitively blocked by the
addition of a 5-fold molar excess of AxPG fusion protein to the primary
antibody. Anti-GST antibodies were used as negative control.
To further characterize the large PG complexes previously shown to
be synthesized by the epidermis of the axolotl embryo (32), explants of
dorsal trunk epidermis from axolotl embryos were maintained in tissue
culture for a time period corresponding to stages 30 to 35 in the
presence of [35S]sulfate. After dissociative extraction
and dialysis to recover macromolecular radiolabel, the
35S-labeled PGs were separated by size exclusion
chromatography on Sepharose CL-2B under dissociative conditions (Fig.
1A). The high molecular weight PGs (HMPGs)
eluting in the void volume fractions and PGs of intermediate molecular
weight (IMPGs) eluting with Kav 0.2-0.3 were
separately pooled and subjected to density equilibrium centrifugation
in self-forming gradients of CsCl under dissociative conditions
( The HMPG fraction has previously been shown to contain PG complexes
stabilized by intermolecular disulfide bonds (32). After reduction, the
monomers were found to elute at a Kav of 0.2 when analyzed by Sepharose CL-2B chromatography (32). To investigate if
the HMPG monomers resemble the IMPGs, the HMPG fraction obtained after
chromatography on Sepharose CL-2B was subjected to CsCl gradient
centrifugation before and after reduction and alkylation (Fig.
2). The reduced PGs were found at a higher buoyant
density (1.48 g/ml; Fig. 2B) than the unreduced HMPGs (1.42 g/ml; Fig. 2A), indicating that the complex in addition to
the HMPG monomers contain other components with lower buoyant density.
Since the significantly higher buoyant density of the IMPGs (1.55 g/ml; Fig. 1C) indicated that the HMPG monomers and the IMPGs are
not identical, it was decided to further compare the two PG
preparations.
The HMPG Monomers Are Distinctly Different from the IMPGs
After alkaline liberation from
the core protein, the glycosaminoglycan side chains of HMPG eluted
significantly earlier (Kav = 0.3) than those of
IMPG (Kav = 0.4) when chromatographed on a
Sepharose CL-4B column (Fig. 3) that had been previously
calibrated with hyaluronan standards. Relating the elution position to
hyaluronan fragments, the size of the glycosaminoglycan side chains of
HMPG correspond to Mr
A characteristic feature of
several large PGs such as aggrecan (38), PG-M/versican (39, 40), and
neurocan (41) is their ability to specifically bind to hyaluronan
(i.e. noncovalent protein-carbohydrate binding) to form
large supramolecular complexes. To investigate whether the HMPG
complexes can bind to hyaluronan, samples of 35S-labeled
HMPG without (Fig. 4A) or with (Fig.
4B) the addition of unlabeled hyaluronan and unlabeled
aggrecan were dialyzed into dissociative buffer and analyzed by
associative CsCl gradient centrifugation. As shown in Fig. 4, the
position of the 35S-labeled HMPG in the gradient was the
same in the presence and absence of the aggrecan-hyaluronan complexes
(see "Experimental Procedures"). In contrast,
35S-labeled IMPG, which was run on an associative Sepharose
CL-2B column under the same conditions, was found to elute in the void volume after addition of aggrecan and hyaluronan (Fig.
5). Apparently, the IMPG can form aggregates with
hyaluronan, whereas the HMPG complexes lack this property. Any
hyaluronan binding activity of native HMPG monomers could not be tested
since such activity is destroyed by reduction. Attempts to obtain
native HMPG monomers by treatment with acid (pH <4), collagenase, or
hyaluronidase (see "Experimental Procedures") were unsuccessful as
neither of these treatments altered the size or buoyant density of the
HMPG complexes (data not shown).
To investigate if the two PG
preparations contained keratan sulfate, equal amounts of
35S radioactivity of purified HMPG and IMPG were digested
with chondroitinase ABC and analyzed by slot blotting, using monoclonal
anti-keratan sulfate antibodies 5D4 (Fig.
6A). The antibodies reacted well with IMPG,
whereas no immunoreactivity was found for HMPG. To exclude that the
observed difference was due to lower amounts of HMPGs analyzed, a
mixture of anti-CS-stub antibodies (see below) was also used. As shown
in Fig. 6B, the HMPG staining with these antibodies was more
than 5-fold stronger than that of IMPG, arguing against this
possibility. The IMPGs thus resembled the aggrecans in their content of
keratan sulfate, whereas the HMPGs could be more related to the
PG-M/versicans.
The HMPG Monomers Are Related to PG-M/Versican
To investigate if the monomer PGs of the
HMPG complex are related to PG-M/versican, HMPG and IMPG were purified
from a large number of epidermal explants and analyzed by slot blotting
(Fig. 7) using affinity-purified polyclonal antibodies
(anti-AxPG-GST) raised against a fusion protein (AxPG-GST) of
glutathione S-transferase and an 80-amino acid segment of
the CS attachment domain of a partially cloned axolotl PG (AxPG; see
"Experimental Procedures") with homology to
PG-M/versican.2 Chondroitinase ABC-digested HMPG was
clearly immunoreactive to these antibodies, whereas IMPG as well as
bovine nasal cartilage aggrecan were unreactive. The immunopositive
signal of HMPG increased after the membrane was stripped in a reducing
buffer and reprobed, suggesting that reduction of the complex further
exposes the core protein epitopes (data not shown).
To estimate the size of the core protein of
the HMPG monomers, the PGs purified from a large number of epidermal
explants were analyzed by SDS-PAGE and immunoblotting after
chondroitinase ABC treatment and reduction. Detection of the core
protein by using the anti-AxPG-GST antibodies resulted in a weak band
with a size (
The affinity-purified anti-AxPG-GST fusion
protein antibodies stained the subepidermal extracellular matrix of
axolotl embryos (Fig. 9A), as well as the
matrix inside epidermal vesicles (Fig. 10).
Immunofluorescence was found along fibrils in the matrix. This staining
was inhibited by the addition of fusion protein (AxPG-GST), whereas
weak unspecific fluorescence of the epidermal layer remained (Fig.
9B); antibodies against GST gave no staining (data not
shown). From these results, it may be inferred that the major PGs
encountered by neural crest cells in the subepidermal extracellular
matrix of the axolotl embryo are PG-M/versican-like PGs that are
components of large disulfide-stabilized complexes.
In the present study, antibodies against a portion of a partially
cloned PG-M/versican-like axolotl PG (AxPG) expressed as a fusion
protein (AxPG-GST) were demonstrated to recognize the large
disulfide-stabilized PG complexes (HMPG; Fig. 7), previously found to
be the major PGs synthesized by the epidermis of axolotl embryos during
stages crucial to the subepidermal migration of neural crest cells
(32). Another chondroitin sulfate PG (IMPG), which constitutes a minor
fraction of the PGs present in epidermal extracts, was not recognized
by these antibodies when equal amounts of 35S radioactivity
of the purified PGs were analyzed by slot blotting (Fig. 7). Since the
amount of core protein molecules present in the IMPG sample may be
lower than in the HMPG sample, as indicated by the lower
immunoreactivity to anti-CS stub antibodies (Fig. 6), it cannot be
excluded that the IMPGs may have escaped detection with the
anti-AxPG-GST antibodies (Fig. 7). It appears more likely, however,
that the IMPGs which in contrast to the HMPGs were shown to contain KS
(Fig. 6) may be related to the
aggrecans.4
Large PGs of the extracellular matrix have been suggested to interact
with a rich variety of matrix components, both proteins and
carbohydrates. Interactions may be mediated both by the
glycosaminoglycan side chains and by the core protein itself. The
classical case is the noncovalent binding of aggrecan to hyaluronan in
cartilage (38). In PG-M/versican the N-terminal domain of the core
protein exhibits hyaluronan-binding properties (39-41), whereas the
C-terminal domain has lectin-like and heparin binding activity (19) and can bind to tenascin-R (42). The HMPG complexes, in addition to the
PG-M/versican-like monomers, evidently contain some other component to
account for their lower buoyant density (1.42 g/ml; Fig. 1) compared
with that of the monomers (1.48 g/ml; Fig. 2). In the previous study
(32), it was shown that the integrity of the large HMPG complexes is
resistant to treatment with strong chaotropic agent (6 M
guanidine) but susceptible to reduction and alkylation. This suggested
that the HMPG complexes, unlike aggregates with hyaluronan, are
covalently held together by intermolecular disulfide bonds. Due to the
low pH of the extraction buffer (pH 5.8) and the presence of the
alkylating agent NEM, it appears unlikely that the PG complexes could
be artificially formed by disulfide exchange during extraction rather
than being the natural molecular organization in the embryonic
extracellular matrix. In further support of this conclusion, the HMPG
fraction has been found to have a very low relative protein
content.5 In the subepidermal matrix of the
axolotl embryo, PGs appear to be localized along collagen fibrils,
seemingly attached to them (27-30). Moreover, chicken PG-M has been
shown to be able to bind to type I collagen (39). However, neither the
size nor the buoyant density of HMPGs were altered by collagenase
treatment, arguing against the possibility that the isolated HMPG
complexes may owe their size (and buoyant density) to their association with collagen. Furthermore, since hyaluronidase treatment was also
ineffective in dissociating the complexes, the present study rules out
that any binding to hyaluronan could be involved in creating the large
size of the complexes. Hyaluronan binding activity was demonstrated for
the IMPGs (Fig. 5) but not for the HMPG complexes (Fig. 4). Although
the intact complexes seem to be unable to bind to hyaluronan under the
present experimental conditions, it cannot be precluded that the native
monomers might have the capacity to do so. The protein-protein
interactions leading to intermolecular disulfide bridges could, for
instance, obscure the recognition sites for hyaluronan. Nevertheless,
since the PG-M/versican-like PGs apparently form large supramolecular
complexes by protein-protein interaction, the aggregation with
hyaluronan may have been rendered redundant in the axolotl embryo.
Whether the complex formation is compatible with a role for
PG-M/versican in bridging cell surfaces to pericellular hyaluronan
(40), or in cell attachment-inhibitory binding to cell surface
hyaluronan (7), remains to investigate.
It is not known whether the ability to form intermolecular disulfide
bonds may be unique to the axolotl variant of PG-M/versican or if it
may be a general feature of some specific splice variant. When the
structure of bovine PG-M/versican was investigated in the transmission
electron microscope with the glycerol spraying/rotary shadowing
technique (43), a pronounced tendency for these PGs to self-aggregate
was noted. By that technique, however, it could not be judged whether
covalent intermolecular bonds were involved in stabilizing the
aggregates. Accordingly, it remains to be studied whether
PG-M/versican-like PGs exist in the form of large disulfide-stabilized PG complexes in other systems. Intriguingly, PG-1000/TAP-1, a large PG
component of disulfide-stabilized complexes associated with the
electric organ basement membranes of electric rays (44-45), might be
related to the PG-M/versicans. Very large CS/DSPGs, which like the
HMPGs of the axolotl embryo elute in the void volume in Sepharose CL-2B
chromatography under dissociative conditions, have previously been
isolated from sea urchin embryos (46-47). However, it remains unknown
whether the sea urchin PGs owe their large size to complex formation or
to the presence of exceedingly large CS/DS side chains (47). Although
the molecular identity of the large sea urchin PGs has not yet been
reported, the dependence on these molecules for the migration of
primary mesenchyme cells during sea urchin development (47-49) is
intriguing.
Previously, bovine aggrecan was found to inhibit axolotl neural crest
cell migration in vitro (6-7), whereas aggrecans, when coated onto microcarriers and introduced into the migratory pathways of
axolotl embryos (50), affected neural crest cells differently depending
on what species the PGs were obtained. The choice of model PG for the
study of neural crest cell migration in the axolotl embryo may thus be
crucial. Since ample purification of PGs from axolotl embryos is not
readily feasible, it is presently difficult to directly address the
question whether the large PG-M/versican complexes affect migration in
the embryo. Such studies may have to await the identification of all
components in the complexes so that these may be assembled from
components expressed and purified from other sources or the development
of gene targeting techniques in the axolotl.
We express our gratitude to Drs. Koji Kimata
and Tamayuki Shinomura for their generous gift of chicken PG-M cDNA
and to Sandra Borland at the Indiana University Axolotl Colony,
Bloomington, IN, for patiently providing embryos. We are grateful to
Dr. Staffan Johansson for oligonucleotide synthesis, helpful
discussion, and critical reading of the manuscript. We thank Inger
Eriksson, Britt-Marie Fogelholm, and Charlotte Fällström
for technical assistance.
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
Acknowledgments
REFERENCES
500 kDa) as those of PG-M/versican variants of other species. Another chondroitin sulfate proteoglycan that was present in small amounts in the epidermal extracts was found to be distinctly different from the similarly sized
PG-M/versican-like monomers.
150,000) chondroitin/dermatan sulfate
(CS/DS) side chains. In the present work these complexes are further
studied, and their PG component is purified, characterized, and
identified. We report that the complexes contain PG-M/versican-like PG
monomers in addition to some other component of low buoyant density.
Furthermore, the PG-M/versican-like monomers of the complex are shown
to be distinctly different from another CSPG that is present in small
amounts in the epidermal extracts.
Embryos
-aminocaproic
acid, 10 mM EDTA, and 1 mM phenylmethylsulfonyl
fluoride as protease inhibitors (35). After end-over-end extraction for
24 h at 4 °C, the radiolabeled macromolecules (
5 × 106 cpm/100 epidermal explants) were recovered by extensive
dialysis against extraction buffer. For immunoblotting, 400 epidermal
explants were prepared as above but without radiolabeling. An aliquot
of the 35S-labeled material was added to the unlabeled
material to act as tracer during the purification steps.
)
of fractions collected from the bottom was determined by weighing, and
the 35S radioactivity of each fraction was determined by
liquid scintillation analysis of an aliquot. Recovery of
35S radioactivity was
70%. IMPGs, pooled as indicated
in Fig. 1C, were diluted 40 times with a buffer containing 7 M urea, 10 mM Tris, pH 8, 0.1% Triton X-100,
and a mixture of protease inhibitors (1 mM
phenylmethylsulfonyl fluoride, 2 mM NEM, 2 mM
EDTA, and 1 µg/ml pepstatin), and applied to a DEAE-Sephacel
(Pharmacia) ion exchange column, equilibrated with a buffer containing
the same ingredients together with 150 mM NaCl. After
washing with equilibration buffer and a buffer containing 150 mM NaCl, 50 mM Ac
, pH 4, 0.1%
Triton X-100, and the mixture of protease inhibitors, the PGs were
eluted with extraction buffer.
Fig. 1.
Purification of large proteoglycans (PGs) by
size exclusion chromatography and density equilibrium gradient
centrifugation. a, extracted 35S-labeled PGs
(1.3 × 106 cpm) were chromatographed on Sepharose
CL-2B and eluted with extraction buffer at a flow rate of 4 ml/h.
Fractions of 1.3 ml were collected, and 10 µl of these were assayed
for 35S radioactivity. Two size classes of large PGs, HMPGs
and IMPGs, were separately pooled as indicated by the bars.
The pools of 35S-labeled HMPGs (b) and IMPGs
(c), respectively, were subjected to CsCl gradient
centrifugation under dissociative conditions. Fractions of
0.5 ml
were collected from the bottom of the tubes, and aliquots of 5 µl
were assayed for 35S radioactivity. HMPGs with
1.42
g/ml (b), and IMPGs with
1.55 g/ml (c),
respectively, were pooled as indicated by the bars.
[View Larger Version of this Image (18K GIF file)]
-alanine was used as void volume and
total volume markers, respectively. 3H-Labeled hyaluronan
fragments of Mr = 40,000, kindly provided by Dr.
Håkan Pertoft, Department of Medical Chemistry, Uppsala University,
were used as an internal standard.
>1.65 g/ml and
=1.57 g/ml, respectively (data not shown).
-D-thiogalactopyranoside, the bacteria were further cultured and lysed, after which the axolotl PG-glutathione
S-transferase (AxPG-GST) fusion protein was purified by
affinity chromatography on a glutathione-Sepharose (Pharmacia)
column.
o
1.42 g/ml). The HMPGs were mostly found in
mid-gradient fractions with
1.42 g/ml (Fig. 1B) which
were pooled as indicated in the figure and used for further
characterization. In contrast, the majority of IMPGs were recovered in
fractions of significantly higher density (
1.55 g/ml; Fig.
1C). Evidently, the rather prominent second peak that
appears at
1.42 g/ml in Fig. 1C represents
contaminating HMPG which by its relative abundance is present in the
IMPG pool after separation by Sepharose CL-2B chromatography. The
IMPGs, pooled as indicated in the figure, were further purified from
contaminating HMPG by ion exchange chromatography3 before subsequent
characterization.
Fig. 2.
Buoyant densities of HMPG complexes and HMPG
monomers. 35S-Labeled HMPGs were subjected to CsCl
gradient centrifugation under dissociative conditions before
(a) and after (b) reduction and alkylation.
Fractions of 0.5 ml were collected, and aliquots of 200 µl were
assayed for 35S radioactivity.
[View Larger Version of this Image (20K GIF file)]
150,000, as previously
reported (32), whereas the glycosaminoglycan side chains of IMPG are
shorter (Mr
75,000).
Fig. 3.
Glycosaminoglycan chain length of HMPGs and
IMPGs. 35S-Labeled glycosaminoglycan side chains (3,000 cpm),
liberated by treatment with alkaline borohydride, of purified HMPG
(filled squares) and IMPG (open squares) were
chromatographed on a Sepharose CL-4B column previously calibrated with
hyaluronan fragments (32). The arrow shows the elution
position of 3H-labeled hyaluronan fragments of
Mr = 40,000, which were used as internal
standard.
[View Larger Version of this Image (25K GIF file)]
Fig. 4.
Hyaluronan binding activity of HMPG.
CsCl gradient centrifugation of 35S-labeled HMPG under
associative conditions in the absence (a) and presence
(b) of exogenously added unlabeled hyaluronan and aggrecan.
Fractions of 0.5 ml were collected, and aliquots of 200 µl were
weighed and assayed for 35S radioactivity. Under the same
conditions, 3H-labeled hyaluronan in the presence and
absence of added aggrecan (see "Experimental Procedures") was
recovered at
>1.65 g/ml and
= 1.57 g/ml, respectively.
[View Larger Version of this Image (19K GIF file)]
Fig. 5.
Hyaluronan binding activity of IMPG.
Sepharose CL-2B chromatography of 35S-labeled IMPG under
associative conditions in the absence (a) and presence
(b) of exogenously added unlabeled hyaluronan and aggrecan.
Flow rate was 1 ml/h, and fractions of 0.3 ml were collected. The
entire fractions were assayed for 35S radioactivity.
[View Larger Version of this Image (15K GIF file)]
Fig. 6.
Keratan sulfate content of HMPGs and
IMPGs. Samples of HMPG and IMPG (equal amount of
35S-tracer; see "Experimental Procedures"),
corresponding to purified PGs from about 50 epidermal explants, were
digested with chondroitinase ABC, stepwise diluted 1:4, and analyzed by
slot blotting using monoclonal antibody 5D4 directed against keratan
sulfate (a), and a mixture of monoclonal antibodies 3B3,
2B6, and 1B5 (36) directed against the chondroitin sulfate (CS)
"stubs" remaining attached to the core protein after digestion with
chondroitinase ABC (b). Indicated amounts of bovine aggrecan
was used for reference.
[View Larger Version of this Image (18K GIF file)]
Fig. 7.
Immunoreactivity of HMPG and IMPG to
antibodies against axolotl-PG-glutathione S-transferase
(AxPG-GST) fusion protein. Chondroitinase ABC-digested samples of
HMPG and IMPG (same amount as in Fig. 6) were serially diluted 1:4 and
analyzed by slot blotting using affinity-purified anti-AxPG-GST
antibodies. Indicated amounts of AxPG-GST and bovine aggrecan were used
for reference.
[View Larger Version of this Image (46K GIF file)]
500 kDa) in the same range as the core proteins of
PG-M/versicans (data not shown). A stronger signal was obtained when
using a mixture of the monoclonal antibodies 3B3, 2B6, and 1B5 (37), directed against the unsaturated stubs generated by digestion with
chondroitinase ABC, which recognized the same band (Fig. 8). Since only CS/DS is present in the HMPG preparation
(32), and only one band (
500 kDa) was detected with anti-CS-stub
antibodies, the detected core protein must be derived from the major PG
in the preparation. It thus appears unlikely that any significant amount of contaminating PGs would be present in the HMPG preparation. The core protein of IMPG could not be detected by immunoblotting using
these conditions.
Fig. 8.
Size of the HMPG core protein.
Chondroitinase ABC (CSase ABC)-digested (+)/undigested (),
reduced and alkylated HMPGs and IMPGs (equal amount of
35S-tracer; see "Experimental Procedures") were
detected with the same mixture of anti-CS-stub antibodies as in Fig. 6,
after electrophoresis in a homogeneous 7.5% PhastTM
SDS-PAGE gel run for 130 V-h and electroblotting to nitrocellulose membrane. The molecular weight of high molecular weight standards and
reduced and alkylated laminin are indicated. Similarly treated bovine
aggrecan (Aggr) was run for reference.
[View Larger Version of this Image (70K GIF file)]
Fig. 9.
Immunofluorescence of embryonic subepidermal
extracellular matrix with antibodies against axolotl-PG-glutathione
S-transferase (AxPG-GST) fusion protein. Sections of
the dorsal trunk of axolotl embryos at stage 32 were incubated with
anti-AxPG-GST antibodies in the absence (a) and presence
(b) of exogenously added AxPG-GST fusion protein. The bound
primary antibodies were detected by immunofluorescence using
fluorescein isothiocyanate-conjugated anti-rabbit Ig as secondary
antibodies. Ep, epidermis; NC, neural crest;
NT, neural tube; S, somites. Bar = 20 µm.
[View Larger Version of this Image (85K GIF file)]
Fig. 10.
Immunofluorescence of extracellular matrix
of epidermal explants with antibodies against axolotl-PG-glutathione
S-transferase (AxPG-GST) fusion protein. Sections of
epidermal vesicles were incubated with anti-AxPG-GST antibodies and
detected by immunofluorescence (a) as described in the
legend to Fig. 9. To assess the extracellular matrix localization of
the immunoreactivity, the same section as in a was
photographed with combined immunofluorescence and phase-contrast optics
(b). Bar = 10 µm.
[View Larger Version of this Image (61K GIF file)]
*
This work was supported by Grant 6525 from the Swedish
Medical Research Council, Konung Gustaf V:s 80-Årsfond, and the
Swedish Natural Science Research Council Grant B-AA/BU 03810-311. The costs of publication of this
article were defrayed in part by the
payment of page charges. The article
must therefore be hereby marked
"advertisement" in
accordance with 18 U.S.C. Section
1734 solely to indicate this fact.
To whom correspondence should be addressed: Dept. of Veterinary
Medical Chemistry, SLU, BMC, Box 575, S-751 23 Uppsala, Sweden. Tel.:
+46-18-174276; Fax: +46-18-550762; E-mail:
michael.stigson{at}vmk.slu.se.
1
The abbreviations used are: PG, proteoglycan;
HMPG, high molecular weight PG; IMPG, intermediate molecular weight PG;
CS, chondroitin sulfate; DS, dermatan sulfate; AxPG, axolotl
proteoglycan; GST, glutathione S-transferase; PAGE,
polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline;
NEM, N-ethylmaleimide.
2
Currently, an axolotl PG (AxPG) with homology to
PG-M/versican is being cloned (M. Stigson and L. Kjellén, work in
progress). Briefly, a cDNA probe corresponding to a portion of the
conserved C-terminal domain of chicken PG-M (22) prepared from a
partial cDNA clone generously provided by Drs. K. Kimata and T. Shinomura was used for the screening of an oligo(dT)-primed embryonic
axolotl cDNA library. A 4.67-kilobase pair AxPG partial cDNA
clone (the full transcript was estimated by Northern analysis to be
around 12-13 kilobase pairs) was sequenced and found to consist of
3422 of coding and 1245 base pairs of 3-untranslated sequence. The deduced amino acid sequence comprises the C-terminal 1140 amino acid
residues of the core protein including 830 amino acid residues of the
CS attachment domain and the complete conserved C-terminal end with two
epidermal growth factor-like domains, a lectin-like domain, and a
complement regulatory protein-like domain. The C-terminal domain of
AxPG shows an extensive amino acid sequence identity to that of the
published chicken (22), mouse (24), and human (51) PG-M/versican,
i.e. 83% identity to the avian and 78% to the mammalian
species, whereas the similarity to other large PGs is significantly
lower, e.g. 58% identity to chicken aggrecan (52) and 56%
to rat neurocan (53). However, the CS attachment region apparently
shows limited sequence conservation between the species and was chosen
for expression as a fusion protein and generation of AxPG-specific
antibodies (cf. 54).
3
Introducing ion exchange chromatography on
DEAE-Sephacel as a first step before Sepharose CL-2B chromatography or
CsCl gradient centrifugation was found to result in a significantly
reduced recovery of HMPGs compared with IMPGs (M. Stigson, unpublished data). For unknown reasons, HMPGs appear to bind irreversibly to the
ion exchange resin and are only partly eluted by increasing the salt
concentration. This property of the HMPG was taken advantage of for
purification of IMPGs, which were enriched after ion exchange chromatography of HMPG-contaminated IMPG preparations.
4
Since the buoyant density of the HMPG monomers
is lower, 1.48 g/ml (Fig. 2) versus 1.55 g/ml (Fig. 1) for
IMPG, while their glycosaminoglycan side chains are longer,
Mr 150,000 versus Mr
75,000 for IMPG (Fig. 3), HMPG is likely to have fewer side chains
per core protein molecule than IMPG. Nevertheless, the immunoreactivity
of IMPGs to anti-CS-stub antibodies was found to be lower than for the
HMPGs when equal amounts of 35S radioactivity of purified
PGs were analyzed by slot blotting (Fig. 6). A restricted access of
primary antibodies due to a relative epitope clustering could account
for the lower immunoreactivity of IMPG.
5
When epidermal explants were simultaneously
labeled with [3H]leucine and [35S]sulfate
(M. Stigson, unpublished data) and analyzed by chromatography on
Sepharose CL-2B, essentially all 3H-labeled macromolecules
of the epidermal explants eluted with Kav
0.8, whereas <1% of the 3H radioactivity was recovered
in the void volume. The low protein content of the HMPG fraction did
not allow for further characterization of its protein component.
©1997 by The American Society for Biochemistry and Molecular Biology, Inc.