of heparan sulfate proteoglycan

The EMBO Journal vol.4 no.4 pp.905-912, 1985
Immunological characterization of basement membrane
of heparan sulfate proteoglycan
Marie Dziadek, Sakuhei Fujiwaral, Mats Paulsson and
Rupert Timpl
Max-Planck-Institut fir Biochemie, D-8033 Martinsried, FRG
'Present address: 2nd Internal Medicine, Kobe University Hospital, Kobe,
Communicated by R.Timpl
Antibodies were raised against a small high-density and a
large low-density form of heparan sulfate proteoglycan from
a basement membrane-producing mouse tumor and were
characterized by radioimmunoassays, immunoprecipitation
and immunohistological methods. Antigenicity was due to the
protein cores and included epitopes unique to the low density form as well as some shared by both proteoglycans. The
antibodies did not cross-react with other basement membrane
proteins or with chondroitin sulfate proteoglycans from interstitial connective tissues. The heparan sulfate proteoglycans
occurred ubiquitously in embryonic and adult basement membranes and could be initially detected at the 2-4 cell stage
of mouse embryonic development. Low levels were also found
in serum. Biosynthetic studies demonstrated identical or
similar proteoglycans in cultures of normal and carcinoembryonic cells and in organ cultures of fetal tissues. They
could be distinguished from liver cell membrane heparan
sulfate proteoglycan, indicating that the basement membrane
types of proteoglycans represent a unique class of extraceliular
matrix proteins.
Key words: heparan sulfate proteoglycan/radioimmunoassays/biosynthesis/tissue localization/embryonic basement membranes
Heparan sulfate proteoglycans are a special class of sulfated
glycosaminoglycan-protein complexes within the large family of
proteoglycans (Heinegard and Paulsson, 1984). They are typical
constituents of basement membranes (Kanwar and Farquhar,
1979; Hassel et al., 1980) where they appear to play a central
role in controlling filtration and other biological properties (Farquhar, 1981). Structural studies of heparan sulfate proteoglycans
from the basement-membrane producing mouse EngelbrethHolm-Swarm (EHS) tumor have demonstrated a high- and lowdensity form of the proteoglycan (Fujiwara et al., 1984). The
high-density form (mol. wt. 130 000) consists of a small protein
10 000) connected to four, 30 nm long heparan
core (mol. wt.
sulfate chains (mol. wt. 29 000). The low-density proteoglycan
(mol. wt. 2400 000) contains >50% protein and thus a substantially larger protein core. Similar data were obtained in other
studies (Martin et al., 1984) and for a large proteoglycan from rat
yolk sac carcinoma (Fenger et al., 1984). A considerable structural similarity to the EHS tumor high-density form is found for
a proteoglycan from glomerular basement membranes (Kanwar
et al., 1981, 1984) which may, in addition, contain a proteinrich heparan sulfate proteoglycan (Parthasarathy and Spiro,
1984). Numerous other studies based on metabolically labelled
IRL Press Limited, Oxford, England.
proteoglycans obtained from basement membrane-producing cells
and organs indicate a considerable structural diversity among
heparan sulfate proteoglycans (Oohira et al., 1982, 1983; Oldberg
et al., 1982; Buonassisi and Colburn, 1983; Lowe-Krentz and
Keller, 1983; Hampson et al., 1984; Heathcote and Orkin, 1984;
lozzo, 1984; Tyree et al., 1984). The precise relationship between these proteoglycans and the high- and low-density forms
found in tumors remains to be elucidated. Other types of heparan
sulfate proteoglycans may also exist, such as a cell membranebound form identified initially in rat liver (Oldberg et al., 1979;
Kjellen et al., 1981) and larger variants from fibroblasts (Carlstedt
et al., 1983) which are possibly related to the transferrin receptor (Fransson et al., 1984).
The diversity of heparan sulfate proteoglycans in tissues has
become increasingly studied using immunological methods
(Hassell et al., 1980; Oohira et al., 1983; Buonassisi and Colburn, 1983; Woods et al., 1984; Fujiwara et al., 1984; Fenger
et al., 1984). Here we describe quantitative immunochemical
assays for proteoglycans from the EHS tumor and the properties of affinity-purified antibodies. Using these approaches we
could detect similar proteoglycans in a variety of biological
samples and distinguish them from other types of proteoglycans.
Production and specificity of antibodies against heparan sulfate
The low-density form of the proteoglycan purified from a mouse
tumor basement membrane produced high titer antisera in rabbits
(50% antigen binding in radioimmunoassay at dilutions 1:15 x 104). Approximately 100-fold lower titers were found in these
sera for the high-density proteoglycan and laminin, and negligible reactions with other basement membrane proteins such as
collagen IV, nidogen and fibronectin. The high-density proteoglycan was a much weaker immunogen and only moderate titers
of 1:50 - 100 were obtained after multiple antigen injections.
However, much stronger reactions were found in this antiserum
for low-density proteoglycan and laminin (titers 1: 103- 104)
presumably due to low contaminations of these antigens in the
high-density proteoglycan used for immunization.
Antibodies were purified from the sera by immunoadsorption
on columns of the low- or high-density proteoglycan after a preadsorption step on a laminin column. Both sets of purified antibodies retained a stronger binding (30- to 100-fold) to the lowdensity proteoglycan than the high-density form, but failed to
react with other basement membrane proteins (Figure 1). It was
interesting, however, that antibodies raised against the low-density proteoglycan showed a higher antigen binding capacity when
compared with those against high-density proteoglycan indicating
a 30-fold difference in affinity constants (Engel and Schalch,
1980). Purified antibodies against both high- and low-density
forms were used for immunoprecipitation and immunofluorescence analyses.
Several antisera against laminin, collagen IV and nidogen
showed negligible binding for the proteoglycans in radioimmuno-
M.Dziadek et al.
assays (titers <1:20) except for a single antiserum against
laminin. This reaction was presumably due to proteoglycan contamination in the immunogen used since it could be completely
eliminated by passage of the antiserum over a proteoglycan
column. Antibodies eluted from the column reacted exclusively
with proteoglycan and no longer with laminin.
Immunochemical properties of high- and low-density heparan
sulfate proteoglycans
The antigenic structure of low-density proteoglycan and its
relationship to other extracellular matrix proteins was studied by
radioimmuno-inhibition assays (Figure 2). Release of > 80% of
the heparan sulfate from the protein core by heparitinase digestion did not change antigenic activity. In addition, lack of activity
was found for purified heparan sulfate chains, indicating that the
major epitopes are located on the protein core. The high-density
proteoglycan showed a 1000-fold lower inhibitory activity. A
similar low activity was found for laminin, collagen IV and
nidogen (Figure 2a), presumably due to a low contamination
( 0.2%) by the low-density form rather than immunological
10-1 10-2 10-3
Antibody concentration (pg/mt)
Fig. 1. Radioimmunoassay binding of purified antibodies against low-density
(closed symbols) and high-density (open symbols) heparan sulfate
proteoglycan. 125I-Labelled test antigens (2.5 ng/ml) were low-density
proteoglycan (0, 0), high-density proteoglycan (U, E) and laminin,
collagen IV, nidogen and fibronectin (A, A). The latter all showed
negligible binding (<2%).
cross-reactivity between these components and the proteoglycan.
This interpretation was confirmed by radioimmuno-inhibition
assays specific for laminin, collagen IV and nidogen, none of
which could be inhibited by a 100-fold molar excess of lowdensity proteoglycan (data not shown).
Interspecies cross-reactivity of antisera raised against the lowdensity proteoglycan from the mouse was examined by using 6 M
guanidine.HCl extracts of bovine lens capsule and ascites fluid
from a rat yolk sac tumor (Wewer, 1982) as convenient sources
of basement membrane components. Both samples produced
inhibition curves having slopes identical to those of a guanidine.HCl extract of mouse lens capsule and authentic mouse
proteoglycan in the assay (Figure 2b). Two other antisera showed only partial cross-reactivity, with maximal inhibition at a 60%
plateau. This demonstrates that sufficient epitopes are shared by
mouse, rat and bovine heparan sulfate proteoglycans to allow
a comparison with various chondroitin or dermatan sulfate proteoglycans obtained from bovine and rat cartilage, bone and aorta.
All these materials were found to be non-inhibitory in the radioimmuno-inhibition assay for the low-density form (Figure 2b).
Antigenic determinants of the high-density proteoglycan were
studied in inhibition assays using the labelled high-density form
as tracer. Four different antisera which were raised against either
the high-density or low-density form and antibodies found
accidentally in an anti-laminin antiserum could be equivalently
inhibited by both the high- and low-density proteoglycan but not
or only marginally by heparan sulfate chains, laminin, collagen
IV or nidogen (Figure 3). The protein core released from the
high-density proteoglycan by treatment with trifluoromethanesulfonic acid still possessed distinct inhibitory activity.
The data so far show that antisera raised against either the highor low-density form of the proteoglycan contained primarily antibodies to determinants unique to the low-density form, but in
addition some antibodies to determinants shared by both proteoglycans. This is presumably due to the much stronger immunogenic potential of the low-density form, such that even low
contamination in the high-density preparation produces a significant antibody response. A comparison was made of antisera
against the high-density form prior to and after purification on
a high-density adsorbent in inhibition assays with labelled low-
( ng protein /ml)
( ng/mL or jul/mI)
Fig. 2. Comparison by radioimmuno-inhibition assays of low-density heparan sulfate proteoglycan (LD-PG) with other basement membrane proteins (a),
proteoglycans from other tissues (b) and various biological samples (c). Low-density proteoglycan was used as 1251-labelled test antigen and as reference
inhibitor (0). Other antigens used for inhibition were: in (a) heparitinase-treated LD-PG (0), high-density heparan sulfate proteoglycan (V), heparan sulfate
chains from LD-PG (A), laminin (El), nidogen (O) and collagen IV (A); in (b) heparan sulfate proteoglycan (V), 6 M guanidine.HCI extracts of mouse (0)
and bovine lens capsule (A), ascites fluid from rat yolk sac carcinoma (A), and various rat (E) and bovine chondroitin sulfate proteoglycans (O); and in (c)
cell culture medium from stimulated F9 cells (0), normal mouse serum (U) and serum from EHS tumor-bearing mice (E). The antisera were against LDPG (a) or against HD-PG (b,c).
Characterization of heparan sulfate proteoglycan
Table I. Content of low-density heparan sulfate proteoglycan in various
biological mouse samples determined by radioimmuno-inhibition assay
10- 2 10-1 100 10
Inhibitor (nM)
Fig. 3. Radioimmuno-inhibition assay with antibodies against high-density
heparan sulfate proteoglycan (HD-PG) and 1251-labelled HD-PG as test
antigens. Inhibitors were HD-PG (C), low-density proteoglycan (0),
protein core of HD-PG (A), heparan sulfate chains of HD-PG (A), nidogen
(CI) and collagen IV (U). The antibodies were isolated from an antilaminin serum (see text) but identical results except for a lower sensitivity in
the assays were obtained with anti-HD-PG antisera.
Proteoglycan content
6 M guanidine.HCI extracts:
EHS tumor
Reichert's membrane
Visceral yolk sac
Lens capsule
Tissue culture medium:
Reichert's membrane
Cell culture medium:
F9 cells
F9 cells + RAa
F9 cells + RA/cAMPb
PYS-2 cells
Schwann cells
Serum (average of 3-5):
Normal mice
EHS-tumor bearing mice
(1tg/mg protein)
(ptg/m1/24 h)
(ng/106 cells/24 h)
18 + 1
82 + 27
'After stimulation with l0-7 M retinoic acid.
bAfter stimulation with 10-1 M retinoic acid and
1.4 =
Fraction No.
Fig. 4. Separation of high- (HD-PG) and low-density (LD-PG) forms of
heparan sulfate proteoglycan from a 6 M guanidine.HCI extract of the
mouse EHS tumor and quantitative analysis by radioimmunoassays. The
assays were basically those described for LD-PG (0-0) in Figure 2 and
for HD-PG (0
0) in Figure 3.
density proteoglycan. Data indicated that antibodies cross-reacting
with the low-density form were retained on the high-density column. When whole antiserum was used in the assay the lowdensity proteoglycan was a 1000-fold stronger inhibitor than the
high-density form (Figure 2b), which was reduced to a 30-fold
difference when purified antibody was used (not shown). Since
a 30-fold difference was also found in direct binding studies
(Figure 1), the data indicate that although the high- and lowdensity forms of proteoglycans have common antigenic determinants, antibodies bind with a higher affinity to the shared determinants present on the low-density form than on the high-density
form (see also Timpl and Risteli, 1982). A clear distinction between high- and low-density proteoglycans is feasible after fractionation of tissue extracts by CsCl gradient centrifugation
followed by radioimmunoassays for both the low- and highdensity forms (Figure 4). A molar ratio for high- to low-density
form of 5:1 was found in a guanidine.HCl extract of the mouse
EHS tumor prepared under conditions minimizing proteolytic
l0-4 M
dibutyryl cyclic
degradation. In contrast, analyses of similar extracts of bovine
lens capsule and culture medium from retinoic acid-stimulated
F9 cells demonstrated that in these samples the low-density proteoglycan was in > 5-fold molar excess.
Immunological analysis of proteoglycans in tissues
Since radioimmuno-inhibition assays for the low- density proteoglycan showed a high sensitivity (ng/ml range) and parallel
inhibition profiles with biological samples (Figure 2) they were
useful for the quantitative analysis of extracts prepared from
various adult and embryonic basement membranes (Table I).
Small amounts of proteoglycan were detected in the serum of
normal mice, with a 3- to 6-fold increase in the serum of mice
transplanted with the EHS tumor (Table I) indicating the release
into the circulation of proteoglycan synthesized by the tumor.
Analyses of tissue sections by indirect immunofluorescence
with purified antibodies against both high- and low- density
proteoglycan (see Figure 1) demonstrated distinct staining of all
regions containing basement membranes, such as glomerular and
tubular basement membranes, Bowman's capsule and mesangium
in kidney, sinusoidal regions and vessel walls in liver, and lens
capsule and cornea in the eye. Similar strong reactions were
observed with embryonic basement membrane. Proteoglycans
showed, in most cases, close co-distribution with other basement
membrane proteins (laminin, nidogen). However, in skin slight
staining of the dermal matrix was observed in addition to a
stronger reaction of the epidermal-dermal basement membrane.
In mid-gestation extra-embryonic membranes intense positive
staining was found in Reichert's membrane (Figure 5) and a
weaker but distinct staining in amnion. In the visceral yolk sac
heparan sulfate proteoglycan was uniformly distributed through
the mesoderm layer and underlying extracellular matrix, unlike
laminin, nidogen and collagen IV which are concentrated in the
endoderm-mesoderm interface. This proteoglycan localization
pattern was similar to that for type I collagen and fibronectin.
Certain basement membrane proteins are early products during mouse embryonic development (Leivo et al., 1980; Cooper
M.Dziadek et al.
*d:'.,; '.:
Fig. 5. Different localization of basement membrane and cell membrane forms of heparan sulfate proteoglycan in Reichert's membrane (RM) and the EHS
tumor (EHS). The sections were stained in indirect immunofluorescence with antibodies against high-density proteoglycan from the EHS tumor (anti-BM) or
antiserum against proteoglycan from rat liver cell membranes (anti-CM). Phase contrast pictures of the sections stained with anti-CM are shown to the left.
Magnifications are x570 (RM) and x340 (EHS).
.-. .~~~~~.. ....
.~~~~~~~~~~~~~~~~~.....'.'.'. .'..........
" .'
'. ,,:
gS ,'
.- ....- ...
* :
....... ............
.:' :. is,>ffi: ..
.':. . '.,
'.' -
... ': '_ ...
'-;...';"'; ';; ;-;+..'.' ': .':.,,,':.:
ziL: '::
n ':
Fig. 6. Indirect immunofluorescence staining of pre-implantation mouse embryos with antibodies against basement membrane type of heparan sulfate
proteoglycan (high-density form). The examples are a 3-cell stage embryo (a), a 4-cell stage (c), an 8-cell stage (e) and an 8-cell stage treated with antibody
which was pre-absorbed with proteoglycan. Corresponding phase contrast pictures are shown in b,d,f,h. Magnification x270.
and MacQueen, 1983; Wu et al., 1983). We now extend such
studies by an immunofluorescent analysis of heparan sulfate proteoglycan. A distinct, punctate staining pattern on the cell surface could be seen on 2-4 cell stage embryos. Staining became
patchy and more dense by the 8-cell stage (Figure 6). Subsequent localization over the blastocyst cell surface, inner cell mass,
and in basement membranes separating tissue layers in the postimplantation egg cylinder was identical to that of laminin and
Characterization of heparan sulfate proteoglycan
o I
Slice No.
Fig. 7. Electrophoretic analysis of immunoprecipitates of [35S]methionine-labelled heparan sulfate proteoglycan produced by F9 cells after stimulation with
retinoic acid and dibutyryl cAMP. The precipitates were obtained with antibodies against high-density proteoglycan from EHS tumor (O 0), the
antibodies after adsorption with high-density proteoglycan (0 0) and with antiserum against proteoglycan from liver cell membranes (A--A). Frozen gels
were sliced into 3 mm sections, dissolved in 0.5 ml 30% H202 and radioactivity determined by liquid scintillation counting. The position of authentic
proteoglycan from the EHS tumor (PG) and of laminin (Ln) are indicated on top. Separation was done in agarose/polyacrylamide gels (Fujiwara et al., 1984);
the anode is to the right.
Table II. Reaction with anti-proteoglycan antibodies of [35S]sulfate labelled
high-density (HD) and low-density (LD) proteoglycans from Reichert's
Antibodies against
heparan sulfate
proteoglycan from
EHS tumor, LD, 15 jig
EHS tumor, HD, 30 jig
Liver membrane, 100 jig
Non-immunec, 38 jg
Fraction (%) of 35S-labelled
material precipitated
aAbout 500 c.p.m. per tube.
bAbout 9000 c.p.m. per tube.
cTotal IgG.
nidogen (data not shown).
Proteoglycan synthesis in cell and organ culture
Radioimmuno-inhibition assays demonstrated the presence of
heparan sulfate proteoglycan in cultures of teratocarcinoma cells
(F9, PYS-2), normal mouse cells and embryonic tissues. The
medium concentrations were in the range of 10 ng/ml to 1 Ag/ml
(Table I). Stimulation of F9 cells with retinoic acid, which causes
differentiation into endoderm-like cells, was accompanied by a
15-fold increase in the amount of secreted proteoglycan, with
little further increase after addition of dibutyryl cyclic AMP.
Metabolic labelling of F9 cells with [35S]methionine and
Reichert's membrane with [35S]sulfate allowed an analysis of
newly synthesized proteoglycans by immunoprecipitation. Agarose gel electrophoresis of the F9 cell immunoprecipitate showed
a single broad peak with the mobility of authentic heparan sulfate
proteoglycan which was clearly separated from the position of
laminin (Figure 7). A sulfate-labelled guanidine.HCl extract of
Reichert's membrane was mixed with a similar, non-labelled
extract from the EHS tumor and used for the co-purification of
both high- and low-density proteoglycans by three distinct steps
(see Materials and methods). The labelled material in the final
products could be specifically precipitated by antibodies against
proteoglycan from the EHS tumor (Table H).
Some of the proteoglycan produced by cultured F9 and PYS-2
cells was deposited in a dense, extracellular fibrillar network
which could be visualized by immunofluorescence staining. Staining was weak on F9 cells but considerably increased after retinoic
acid stimulation, which is in agreement with an increased secretion into the culture medium (Table I). Positive staining was
also observed with fibrosarcoma cells HT-1080 and choriocarcinoma cells demonstrating cross-reactions of the antibodies
with human proteoglycans.
Comparison of basement membrane and cell membrane forms
of heparan sulfate proteoglycans
An antiserum raised against heparan sulfate proteoglycan from
rat liver cell membranes (Woods et al., 1984) failed to react with
iodine-labelled proteoglycans from the EHS tumor in radioimmunoassays (dilutions 1:100-1:3200; data not shown) and
with sulfate-labelled high- and low-density proteoglycan from
Reichert's membrane (Table II). However, a distinct immunofluorescence staining pattern was observed on sections of the EHS
tumor and Reichert's membrane, which was associated with the
clusters of tumor cells and the cell surface of parietal endoderm
cells (Figure 5). In contrast, antibodies against the EHS tumor
proteoglycans produced a very intense staining of the broad basement membrane zones in both tissues and failed to react with
cells (Figure 5). A differential staining pattern was also observed
with F9 and PYS-2 cells where the antiserum against the liver
cell membrane proteoglycans reacted mainly with membranes
and intracellular, rather than extracellular, structures (not shown).
The same antiserum also precipitated labelled material from the
culture medium of F9 cells which showed a faster electrophoretic
mobility when compared with proteoglycans precipitated by anti909
M.Dziadek et al.
bodies against the major proteoglycans obtained from the EHS
(Figure 7). Since other data show substantial crossreactions between rat and mouse proteoglycans (Figure 2) these
observations strongly indicate two structurally different classes
of heparan sulfate proteoglycans in tissues.
The immunological and biosynthetic data demonstrate that the
heparan sulfate proteoglycans from the mouse EHS tumor represent a basement membrane-specific class of glycoconjugates.
The low-density variant was found to be a strong immunogen
presumably due to its large protein core with a mol. wt. of
300 000. The protein core possesses some unique determinants as well as others shared with the high-density form while
the heparan sulfate chains lack antigenicity. The lack of contribution by carbohydrate in the antibody response is in agreement
with observations for polyclonal antibodies raised against a variety
of other proteoglycans (Heinegard and Paulsson, 1984). The antisera failed to cross-react with several other basement membrane
proteins and with bone and cartilage proteoglycans, indicating
a unique structure for the protein core. Complete or partial interspecies cross-reactions were observed for rat tumor ascites fluid
(Wewer, 1982) and bovine lens capsule which both contain base-
ment membrane components.
The high-density proteoglycan of the EHS tumor has a much
smaller protein core (mol. wt. 10 000) which proved to be
a considerably weaker immunogen. Antisera raised against this
form were found to contain mainly antibodies specific for the
low-density variant, presumably due to a small contamination
(0.1 1 %) by the latter in the material used for immunization.
A comparison of epitopes shared by the high- and low-density
proteoglycans in both direct binding assays and radioimmunoinhibition assays indicated a 30-fold higher affinity of antibodies
for determinants on the low-density protein core. This could indicate that the corresponding protein segments on the two protein
identical, but differ in the amino acid sequence,
conformation or glycosylation. However, pulse-chase experiments have suggested that the high-density form of the proteoglycan is a physiological degradation product of the low-density
form (Martin et al., 1984) which would predict that both forms
share identical protein sequences.
Radioimmuno-inhibition assays for the low-density proteoglycan were sufficiently sensitive and specific to allow quantitative
analyses of tissue extracts and culture media. When compared
with authentic proteoglycan the various biological samples produced complete and parallel inhibition curves demonstrating
identical sets of epitopes and, most probably, identical protein
cores are not
core structures
(Timpl and Risteli, 1982). Their identity
supported in studies with metabolically labelled proteoglycans
from Reichert's membrane and retinoic acid-stimulated F9 cells
which resembled high- and low-density proteoglycans from the
EHS tumor in electrophoretic mobility, buoyant density in CsCl
gradients and chromatographic behaviour. A direct and specific
radioimmunoassay was not feasible for the high-density proteoglycan except after fractionation of samples by CsCl gradient centrifugation. Such analyses demonstrated that both density forms
of proteoglycan exist in tissue extracts and cell culture medium.
Since the materials were prepared under carefully controlled conditions the data show that the high-density variant exists in situ
and is not generated as an artefact during the extraction procedure.
Immunofluorescence studies of adult tissues showed a distinct
localization of the tumor heparan sulfate proteoglycans to ana910
regions containing authentic basement membranes, in
(Hassell et al., 1980;
agreement with previous observations
Fenger et al., 1984). The antibodies showed some cross-reactivity
have inditomical
with the dermal interstitial matrix, and previous studies
cated cross-reactivity with fibroblast extracellular matrices (Hedman et al., 1982; Woods et al., 1984). The structural basis for
this cross-reactivity remains to be determined. Heparan sulfate
embryonic baseproteoglycans could alsofirstbe detected onin several
ment membranes and
appeared the cell surface at early
of development. This corresponds to
pre-implantation stages
similar observations for laminin, entactin and collagen IV (Leivo
et al., 1980; Cooper and MacQueen, 1983; Wu et al., 1983)
indicating that these components are already produced at embryonic stages for which no morphological correlates of intact basement membranes exist. This emphasizes a central role of
basement membrane proteoglycans in molecular interactions with
other extracellular matrix proteins (Woodley et al., 1983; Fujiwara et al., 1984) rather than being restricted to a control of
macromolecular filtration (Farquhar, 1981).
As expected from the studies of embryonic tissues, the basement membrane proteoglycans are also produced by embryonal
carcinoma cells such as F9 cells. Stimulation of differentiation
with retinoic acid caused a 15-fold increase in secretion of proform.
teoglycan by these cells, of which 80% was the low-density
A similar increase after stimulation has been reported for synthesis of laminin and collagen IV (Strickland et al., 1980; Howe
and Solter, 1980; Prehm et al., 1982). Proteoglycan synthesis
was, however, insensitive to addition of dibutyryl cyclic AMP
which is thought to initiate further differentiation events. Distinct
production of basement membrane types of proteoglycans cells
also observed for Schwann cells, PYS-2 teratocarcinoma
and endodermal cells of Reichert's membrane in agreement with
metabolic labelling data (Hogan et al., 1982; Oohira et al., 1982;
Tyree et al., 1984). The immunological analysis of these
systems could be valuable for more detailed studies of basement
membrane proteoglycan biosynthesis.
Small amounts of proteoglycan could be detected in normal
serum by radioimmunoassay, presumably originating from cells
of the vessel walls. An increased concentration was observed in
serum of mice with transplants of the EHS tumor, indicating that
pathological production of basement membrane proteoglycan
be monitored by serum analysis. Circulating forms of laminin
and collagen IV have been demonstrated (Risteli et al., 1981)
which increase in concentration during experimental diabetes
(Risteli et al., 1982). Since proteoglycan synthesis appears to
be decreased in diabetic conditions (Rohrbach et al., 1983; Kanwar et al., 1983) specific radioimmunoassays may become
valuable tools for studying molecular events in these and other
basement membrane-associated diseases.
Cell culture studies have indicated a considerable structural
diversity of heparan sulfate proteoglycans (Lowe-Krentz
Keller, 1983;
Morris, 1984; Hampson et al., 1984). This may in part be due
to variations in the structure of heparan sulfate side chains (Radhadiverkrishnamurthy et al., 1984), while little is known about thelack
sity protein
proteoimmunological cross-reactivity between heparan sulfate
cell membranes.
glycans from basement membranes and liver
The latter proteoglycan has a smaller size than the tumor highform and contains a hydrophobic peptide segment which
allows its insertion into plasma membranes (Oldberg et al., 1979;
et al., 1981). Proteolytic cleavage of the cell membrane
forms can cause release from the cell surface and subsequent
Characterization of heparan sulfate proteoglycan
deposition in the extracellular environment. Immunofluorescent
analysis demonstrated a different distribution of basement membrane and cell membrane proteoglycans in tissues. The latter form
could also be detected in association with EHS tumor cells in
tissue sections. We assume that this component is usually lost
during the preparation of the more abundant basement membrane
proteoglycans from the tumor. Taken together the data indicate
that at least two or three distinct types of heparan sulfate proteoglycan exist in tissues which differ in their protein cores. A similar
diversity has recently been established by immunological and
structural studies of chondroitin sulfate/dermatan sulfate proteoglycans (Heinegard et al. in preparation; Brennan et al., 1983)
and may be a characteristic feature of these two major classes
of proteoglycans.
Materials and methods
Purification and fragments of basement membrane proteoglvcans
High-density heparan sulfate proteoglycan was purified from 0.5 M NaCl extracts of the mouse EHS tumor as previously described (Fujiwara et al., 1984).
The residual tissue (Timpl et al., 1983) was then extracted twice overnight at
4°C with 6 M guanidine.HCI, 0.05 M Tris-HCI pH 7.4 containing 10 mM EDTA,
2 mM N-ethylmalemide (NEM) and 2 mM phenylmethanesulfonyl fluoride
(PMSF) as protease inhibitors. Low-density proteoglycans in these extracts were
separated from the bulk of other proteins on DEAE-cellulose (7 M urea, 0.05 M
Tris-HCI pH 7.4, 2.5 mM EDTA, 0.5 mM NEM, 0.5 mM PMSF) and eluted
in the second half of a 0- 0.6 M NaCl gradient. Further purification steps were
on Sephacryl S-400 in 6 M guanidine.HCI with protease inhibitors followed by
CsCl gradient centrifugation in 6 M guanidine.HCI buffer with a starting density of 1.34 g/ml (45 000 r.p.m., 48 h, 18°C). The final material collected from
the central portion of the gradient (6= 1.33- 1.39 g/ml) was essentially free from
high-density proteoglycan, nucleic acids and glycoproteins and appeared
homogeneous by ultracentrifugation in 6 M guanidine.HCI (J.Engel, personal
A 6 M guanidine.HCI extract of [35S]sulfate-labelled Reichert's membrane was
mixed with an excess of a similar but non-labelled extract of the EHS tumor as
carrier. Passage over DEAE-cellullose (see above) separated the low-density proteoglycan (eluted at 0.3 M NaCl) from the high-density form (0.4 M NaCl) which
contained - 14 % and 68 %, respectively, of the total incorporated radioactivity.
Further purifications were done as described above, except for a higher starting
density (1.48 g/ml) for the CsCl gradient centrifugation.
The protein core was released from the high-density proteoglycan by treatment with trifluoromethanesulfonic acid (Edge et al., 1981) and purified on
Sephadex-G50. Treatment with heparitinase using 0.3 mIU/Ag proteoglycan
(Seikagaku Kogyo Co., Tokyo; see Fujiwara et al., 1984) followed by dialysis
was used to prepare the protein core from the low-density proteoglycan. Alkaline
treatment was used to prepare the heparan sulfate chains from both proteoglycans
(Fujiwara et al., 1984).
Other antigens
Collagen IV and laminin from the mouse EHS tumor were those used in previous
studies (Timpl et al., 1978, 1979). The 150 000 mol. wt. form of nidogen (Timpl
et al., 1983) was isolated from the 6 M guanidine.HCI extract (see above) and
purified to electrophoretic homogeneity (unpublished). Purified chondroitin/dermatan sulfate proteoglycans of large molecular mass from bovine nasal cartilage
and aorta and from rat chondrosarcoma (Heinegard and Paulsson, 1984; S.Gardell
of rat nasal
and D.Heinegard, unpublished) and small chondroitin
cartilage and rat or bovine diaphyseal bone (Heinegard and Paulsson, 1984;
Franzen and Heinegard, 1984) were prepared by established methods and kindly supplied by Dr D.Heinegard, Lund.
Tissue extracts were made immediately after dissection in 6 M guanidine.HCI
pH 7.4, 10 mM EDTA, 0.3 mM PMSF and 0.2 mM p-hydroxymercuribenzoate
in order to minimize proteolytic degradation. Reichert's membranes (10 per ml),
lens capsules (20 per ml), visceral yolk sacs (4 per ml) and non-necrotic EHS
tumor (0.5 g/ml) were frozen on dry ice, then extracted overnight at 4°C and
supernatants were stored at -70°C. Aliquots of these extracts were separated
by CsCl gradient centrifugation using the same conditions as described above
for Reichert's membrane proteoglycans.
Cell and organ cultures
Mouse PYS-2 teratocarcinoma cells (Lehman et al., 1974), mouse F9 embryonal
carcinoma cells prior to and after stimulation for 6 days with retinoic acid and
dibutyryl cyclic AMP (Prehm et al., 1982), human HT-1080 fibrosarcoma cells
(American Type Culture Collection) and human choriocarcinoma cells (ENAMI,
gift of G.Wick, Innsbruck) were cultured in Dulbecco's modified Eagle's medium
(DMEM, Flow Laboratories) containing 5 % fetal calf serum (FCS). 24 h supernatants of confluent cultures were taken for radioimmunoassays. Other cells were
labelled with [35S]methionine [100 jLCi/ml, 1058.6 Ci/mmol, New England
Nuclear (NEN)] for 24 h in DMEM containing 2% FCS and 10% of the usual
methionine concentration. Conditioned medium from mouse Schwann cells was
provided by Dr D.Edgar, Martinsried, and fluid from ascites culture of rat yolk
sac carcinoma L2 (Wewer, 1982) were kindly given by Dr U.Wewer, Copenhagen.
Fetal extra-embryonic membranes were dissected from 14th day gestation C3H
BALB/c mice and cultured for 24 h in DMEM with 5% FCS. Labelling of
Reichert's membrane with [35S]sulfuric acid (0.5 mCi/ml, 43 Ci/mg, NEN) was
done for 3 h in DMEM containing 1% FCS.
Antisera and purification of antibodies
Rabbit antisera against low-density proteoglycan were prepared by two injections
of 0.5 mg in complete Freund's adjuvant following a standard protocol (Timpl,
1982). Immunization with high-density proteoglycan (antigen doses 0.2- 1 mg)
started with the same protocol and was followed by 5-8 booster injections at
irregular intervals. High-density (4.6 mg) and low-density (1 mg) proteoglycans
were coupled to 1 g CNBr-activated Sepharose for affinity-purification of antibodies (Timpl, 1982). Antisera were passed over a similarly prepared lamininimmunoadsorbent prior to their passage over the appropriate proteoglycanimmunoadsorbent. For immunofluorescence and immunoprecipitation controls,
antibodies against high-density proteoglycan (35 JLg) were absorbed with highdensity proteoglycan (200 Ag) for 24 h at 4°C. An antiserum against heparan
sulfate proteoglycan from rat liver (Woods et al., 1984) was a kind gift from
Dr M.Hook, Birmingham, Alabama. Antisera and purified antibodies against collagen IV (Timpl et al., 1978), laminin (Timpl et al., 1979) and nidogen (Timpl
et al., 1983) were those used in previous studies.
Immunological methods
Antigens were labelled with 1251 by the chloramine T procedure and used in binding and inhibition (sequential saturation) radioimmunoassays (Timpl and Risteli,
1982). Immunoprecipitations of 0.1 ml aliquots of labelled proteoglycan were
carried out in the presence of phosphate buffered saline pH 7.2, 0.04% Tween
(PBS/Tween) by incubation with purified antibodies (7-40 pg), absorbed antibodies, antiserum to liver cell membrane proteoglycan (10 A1) and normal rabbit
IgG (38 jg) for 24 h at 4°C. Immune complexes were bound to goat anti-rabbit
IgG linked to Staphylococcus aureus (0.1 -0.5 ml Tachisorb, Hoechst AG) for
18-24 h at 4°C, the material was washed with PBS-Tween and then dissolved
by boiling either in sample buffer for 1.2% polyacrylamide/l % agarose gel electrophoresis (Fujiwara et al., 1984) or in 1 M NaCl, 2% SDS for radioactive counting.
For indirect immunofluorescence, sections were incubated with purified antibodies (20-40 Ag/ml in PBS), antisera (dilutions 1:5 1:20) and as controls absorbed antibody or normal rabbit IgG for 40-60 min followed by
FITC-conjugated goat antiserum against rabbit IgG (dilution 1:20, 30 min incubation, from Miles Laboratories). Cell monolayers were washed with PBS,
fixed with 95% methanol (1 min) and air-dried. Cryostat tissue sections
I m) were air-dried and fixed in 70% ethanol (0.5 min). Pre-implantation
stage embryos were flushed from the oviduct or uterus into Tyrode's solution
and zonae pellucidae were removed with acid Tyrode's pH 2.5 (Nicolson et al.,
1975). Embryos were incubated for at least 1 h prior to antibody staining. Sections and cell cultures were mounted in glycerol and embryos in a minimum volume
of Tyrode's solution under a layer of paraffin oil (Boots, UK) on a glass coverslip
and viewed with a Zeiss ICM 405 epifluorescent inverted microscope.
We thank Drs D.Edgar, D.Heinegard, M.Hook, U.Wewer and G.Wick for the
gift of samples and Ms Hildegard Reiter for expert technical assistance. M.D.
and M.P. were supported by fellowships from EMBO and the Alexander von
Humboldt Foundation. The study was also supported by a grant from the Deutsche
Forschungsgemeinschaft (project Ti 95/6-1).
Brennan,M.J., Oldberg, A., Ruoslahti,E., Brown,K. and Schwartz,N. (1983)
Dev. Biol., 98, 139-147.
Buonassisi,V. and Colburn,P. (1983) Biochim. Biophys. Acta, 760, 1-12.
Carlstedt,I., Coster,L., Malmstrom,A. and Fransson,L. A. (1983) J. Biol. Chem.,
258, 11629-11635.
96, 467471.
Cooper,A.R. and MacQueen,H.A. (1983) Dev. Biol.,
Edge,A.S.B., Faltynek,C.R., Hof,L., Reichert,L.E. and Weber,P. (1981) Anal.
Biochem., 118, 131-137.
Engel,J. and Schalch,W. (1980) Mol. Immunol., 17, 675-680.
Farquhar,M.G. (1981) in Hay,E.D. (ed.), Cell Biology of Extracellular Matrix,
Plenum Press, NY, pp. 335-378.
and Albrechtsen,R.
Fenger,M., Wewer,U.
(1984) FEBS Lett., 173,
M.Dziadek et al.
Fransson,L. A., Carlstedt,I., Coster,L. and Malmstrom,A. (1984) Proc. Natl.
Acad. Sci. USA, 81, 5657-5661.
Franzen,A. and Heinegard,D. (1984) Biochem. J., 224, 47-58.
Fujiwara,S., Wiedemann,H., Timpl,R., Lustig,A. and Engel,J. (1984) Eur. J.
Biochem., 143, 145-157.
Hampson,I.N., Kumar,S. and Gallagher,J.T. (1984) Biochim. Biophys. Acta,
801, 306-313.
Hassell,J.R., Gehron Robey,P., Barrach,H.J., Wilczek,J., Rennard,S.I. and Martin,G.R. (1980) Proc. Natl. Acad. Sci. USA, 77, 44944498.
Heathcote,J.G. and Orkin,R.W. (1984) J. Cell Biol., 99, 852-860.
Hedman,K., Johansson,S., Vartio,T., Kjellen,L., Vaheri,A. and Hook,M. (1982)
Cell, 28, 663-671.
Heinegard,D. and Paulsson,M. (1984) in Piez,K.A. and Reddi,A.H. (eds.), Extracellular Matrix Biochemistry, Elsevier, NY, pp. 277-328.
Hogan,B.L.M., Taylor,A. and Cooper,A.R. (1982) Dev. Biol., 90, 210-214.
Howe,C.C. and Solter,D. (1980) Dev. Biol., 77, 480-487.
Iozzo,R.V. (1984) J. Cell Biol., 99, 403-417.
Kanwar,Y.S. and Farquhar,M.G. (1979) Proc. Natl. Acad. Sci. USA, 76, 13031307.
Kanwar,Y.S., Hascall,V.C. and Farquhar,M.G. (1981) J. Cell Biol., 90, 527-532.
Kanwar,Y.S., Veis,A., Kimura,J.H. and Jakubowski,M.L. (1984) Proc. Natl.
Acad. Sci. USA, 81, 762-766.
Kanwar,Y.S., Veis,A., Kimura,J.H. and Jakubowsi,M.L. (1984) Proc. Natl.
Acad. Sci. USA, 81, 762-766.
Kjellen,L., Pettersson,I. and Hook,M. (1981) Proc. Natl. Acad. Sci. USA, 78,
Lehman,J.M., Speers,W.C., Swartzendruber,D.E. and Pierce,G.B. (1974) J.
Cell. Physiol., 84, 13-27.
Leivo,I., Vaheri,A., Timpl,R. and Wartiovaara,J. (1980) Dev. Biol., 76, 100-114.
Lowe-Krentz,L.J. and Keller,J.M. (1983) Biochemistry (Wash.), 22, 4412-4419.
Martin,G.R., Kleinman,H.K., Terranova,V.P., Ledbetter,S. and Hassell,J.R.
(1984) in Porter,R. and Whelan,J. (eds.), Basement Membranes and Cell Movement, CIBA Foundation Symposium 108, Pitman, London, pp. 197-209.
Morris,J.E. (1984) Arch. Biochem. Biophys., 235, 127-140.
Nicolson,G.L., Yanagimachi,R. and Yanagimachi,H. (1975) J. Cell Biol., 66,
Oldberg,A., Kjellen,L. and Hook,M. (1979) J. Biol. Chem., 254, 8505-85 10.
Oldberg,A., Schwartz,C. and Ruoslahti,E. (1982) Arch. Biochem. Biophys., 216,
Oohira,A., Wight,T.N., McPherson,J. and Bornstein,P. (1982) J. Cell Biol.,
92, 357-367.
Oohira,A., Wight,T.N. and Bornstein,P. (1983) J. Biol. Chem., 258, 2014-2021.
Parthasarathy,N. and Spiro,R.G. (1984) J. Biol. Chem., 259, 12749-12755.
Prehm,P., Dessau,W. and Timpl,R. (1982) Connective Tissue Res., 10, 275-285.
Radakrishnamurthy,B., Jeansome,N.E. and Berenson,G.S. (1984) Biochim.
Biophys. Acta, 802, 314-320.
Risteli,J., Rohde,H. and Timpl,R. (1981) Anal. Biochem., 113, 372-378.
Risteli,J., Draeger,K.E., Regitz,G. and Neubauer,H.P. (1982) Diabetologia, 23,
Rohrbach,D.H., Wagner,C.W., Star,V.L., Martin,G.R., Brown,K.S. and Yoon,
J.W. (1983) J. Biol. Chem., 258, 11672-11677.
Strickland,S., Smith,K.K. and Marotti,K.R. (1980) Cell, 21, 347-355.
Timpl,R. (1982) Methods Enzymol., 82A. 472-498.
Timpl,R. and Risteli,L. (1982) in Furthmayr,H. (ed.), Immunochemistry of the
Extracellular Matrix, Vol. I, CRC Press, Boca Raton, pp. 199-235.
Timpl,R., Martin,G.R., Bruckner,P., Wick,G. and Wiedemann,H. (1978) Eur.
J. Biochem., 84, 43-52.
Timpl,R., Rohde,H., Gehron Robey,P., Rennard,S.I., Foidart,J.M. and Martin,G.R. (1979) J. Biol. Chem., 254, 9933-9937.
Timpl,R., Dziadek,M., Fujiwara,S., Nowack,H. and Wick,G. (1983) Eur. J.
Biochem., 137, 455-465.
Tyree,B., Horigan,E.A., Klippenstein,D.L. and Hassell,J.R. (1984) Arch.
Biochem. Biophys., 231, 328-335.
Wewer,U. (1982) Dev. Biol., 93, 416-421.
Woodley,D.T., Rao,C.N., Hassell,J.R., Liotta,L.A., Martin,G.R. and Kleinman,H.K. (1983) Biochim. Biophys. Acta, 761, 278-283.
Woods,A., Hook,M., Kjellen,L., Smith,C.G. and Rees,D.A. (1984) J. Cell Biol.,
99, 1743-1753.
Wu,T.C., Wan,Y.J., Chung,A.E. and Damjanov,I. (1983) Dev. Biol., 100, 496505.
Received on 21 January 1985