Download Fibronectin and Glial Fibrillary Acidic Protein Expression in Normal

[CANCER RESEARCH 42, 168-177,
0008-5472/82/0042-OOOOS02.00
January 1982]
Fibronectin and Glial Fibrillary Acidic Protein Expression in Normal
Human Brain and Anaplastic Human Gliomas1
Trevor R. Jones, Erkki Ruoslahti, S. Clifford Schold, and Dareil D. Bigner2
Departments of Pathology [T. R. J., S. C. S., D. D. B.¡,Medicinéis. C. S.], and Surgery [D. D. B.], Duke University Medical Center, Durham, North Carolina 27710,
and La Jolla Cancer Research Foundation, La Jolla, California 92037 [E. R.]
ABSTRACT
The expression of fibronectin (FN), an extracellular matrix
glycoprotein with adhesive and opsonic properties, was ex
amined in normal and neoplastia human central nervous system
tissues. In both normal adult and fetal brain, FN expression, as
detected by ¡mmunohistochemistry, was restricted to the vasculature. Neurons and glia did not stain. FN expression in nine
of 10 gliomas examined was qualitatively similar to that in
normal fetal and adult brain in that it was confined to the blood
vessels. Nevertheless, in one case, there was interstitial
expression of FN. Four tumors exhibited varying degrees of
fibroblastic overgrowth in which cells containing FN were de
tected. Frozen sections of eight athymic mouse-borne anaplastic human gliomas were stained simultaneously by a doublelabel fluorescein-rhodamine
fluorescence technique for both
FN and glial fibrillary acidic protein (GFAP), a glial and glioma
marker. Seven of eight tumors expressed FN, seven of eight
expressed GFAP, and six of eight expressed both. FN was
detected as fine fibrils and specks between cells while GFAP
presented as a homogeneous intracellular fluorescence. In
many areas of the tumors, groups of cells expressed both FN
and GFAP. Eleven human glioma-derived cell lines were ex
amined for both FN and GFAP expression. In two cell lines,
there were cells which contained both GFAP and FN simulta
neously. Athymic mouse-borne tumors were developed by the
injection of an FN-producing glioma-derived cell line into
athymic mice. Human FN plasma levels were monitored, and
some tumors larger than 1.0 g produced FN which was de
tectable in the plasma, thereby demonstrating that tumor-pro
duced FN can be shed from human glioma-derived cells in
sufficient amounts to be detectable in the bloodstream. These
data show that FN expression in anaplastic human gliomas is
seen with the greatest frequency in athymic mouse-borne
tumors; GFAP-positive cultured cells are also capable of FN
expression. The original tumors rarely express detectable
amounts of FN. The biological basis for this heterogeneity of
FN expression rests in the nature of the individual tissue
studied, its cellular environment (i.e., in vivo or in vitro), and its
status with regard to neoplasia.
INTRODUCTION
FN3 is a glycoprotein
250,000
with a molecular weight of 220,000 to
first isolated from human plasma by Morrison ef al.
' This investigation was supported in part by National Cancer Institute Grants
CA-11898, CA-22790, and CA-28896, National Institute of Neurological and
Communicative Disorders and Stroke Grant 1-R23-NS-16974, American Cancer
Society Grant CH191, and the Roy Alexander Brain Tumor Research Fund.
2 To whom requests for reprints should be addressed.
3 The abbreviations used are: FN. fibronectin; CNS, central nervous system;
GFAP, glial fibrillary acidic protein; ZO, Richter's zinc option medium; FITC,
fluorescein isothiocyanate; DPBS, Dulbecco's phosphate-buffered saline.
Received August 13, 1981 ; accepted October 17, 1981.
168
(21) and from the cell surface of fibroblasts by Ruoslahti et al.
(32). FN has been studied extensively in vitro and shown to be
an important component of the extracellular matrix and an
integral element in cell adhesion and motility phenomena (9,
22, 47). In vivo studies have demonstrated that FN also acts as
opsonic protein. Reductions in circulating FN correlate with
depressions in reticuloendothelial system activity; FN also may
be important in the clearance of gram-positive bacteria (2, 35).
Transformation often reduces or eliminates FN production and
expression in cultured cells (11, 19, 40, 41, 45, 47), and a
loose correlation has been reported between tumorigenicity,
anchorage independence, and a lack of FN expression (5, 14).
In this study, we have examined the expression of FN in
normal and neoplastic CNS tissues. Because the expression of
FN by a given cell type can change with environment, i.e., solid
tumor versus cell culture (5), we have examined surgical biop
sies, primary and established glioma-derived cell lines, and
athymic mouse-borne tumors to determine if FN expression is
in any sense phase specific in glioma-derived systems.
The expression of a universally accepted biochemical differ
entiation marker of astrocytes, GFAP, was examined with FN
expression. GFAP is considered astrocyte specific in normal
and reactive tissues and generally glioma restricted in neoplas
tic tissues (6, 8). It is expressed in some glioma-derived cell
lines (1) and in athymic mouse-borne tumors derived from
anaplastic human gliomas (13).
We found that FN expression in normal adult and fetal human
brain and in 9 glioblastomas is restricted to blood vessels. One
glioblastoma contained FN not only in the blood vessels but
also in interstitial areas containing tumor cells. Six of 8 athymic
mouse-grown anaplastic human gliomas expressed both FN
and GFAP and 2 anaplastic human glioma-derived cell lines
expressed both simultaneously. Studies to determine if human
tumor-produced FN could be used to monitor tumor growth in
athymic mice carrying human gliomas demonstrated that mice
carrying tumors weighing greater than 1.0 g often have human
FN in their plasma.
MATERIALS
AND
METHODS
Maintenance and Growth of Cell Culture Lines. Human fetal lung
fibroblasts (WI-38) and human cell lines derived from anaplastic
gliomas (D-54 MG, D-171b MG, D-171d MG, U-251 MG, U-251 MG
sp, and U-251 MG el,, c!2, and c!3), a meningioma (D-187), an osteogenic sarcoma (2T), and a gliosarcoma (D-173 MG) were maintained
in culture in 100-mm Petri dishes (Falcon Plastics) containing ZO
(Grand Island Biological Co.) supplemented with 20% fetal calf serum,
584 mg glutamine per liter, and 50 fig gentamicin sulfate per ml. For
testing, 3- x 3-mm coverslips were placed in the Petri dishes prior to
cell plating; once cells reached confluence, the coverslips were re
moved from the dishes.
Establishment of Cell Culture Lines from Human Anaplastic
Gliomas Grown in Athymic Mice. Human anaplastic gliomas growing
CANCER
RESEARCH
VOL. 42
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
FN and GFAP Expression
m athymic BALB/c mice were excised under sterile conditions and
minced in serum-free medium (ZO) to yield 1.0-cu mm pieces which
were placed in a trypsinization flask containing a 0.05% collagenase
solution (Worthington Biochemical Corp.) at 37°. The flask contents
were stirred gently for 5 min; large pieces of the tumor were allowed to
settle out, the supernatant was removed and centrifugea at 250 x g,
and the cells were stored on ice. This procedure was repeated twice,
and all 3 supernatant lots were combined and placed in a 100-mm Petri
dish containing ZO supplemented with 20% fetal calf serum.
Absorption and Testing of Goat Anti-Human FN Serum. Aliquots
of goat anti-human FN serum prepared as described by Ruoslahti et al.
(33) were rendered nonreactive with mouse and bovine FN by absorp
tion with mouse and bovine serum proteins. Lyophilized goat anti-FN
serum was rehydrated to a concentration of 80 mg/ml, and dilutions of
the antiserum were tested in an ¡mmunofluorescence assay against
normal BALB/c mouse brain and spleen and a human glial tumor
grown in athymic mice. Mouse serum proteins, also from BALB/c mice,
were obtained by retroorbital bleedings; the blood was maintained at
37° during clotting to reduce FN binding to fibrinogen (31). Mouse
serum was dialyzed overnight in two 4-liter volumes of water, lyophilized, and rehydrated to a concentration of 160 mg/ml. A 1.5-ml aliquot
of goat anti-human FN serum (80 mg/ml) was incubated with 12 mg of
mouse serum protein at 37° for 1 hr and then overnight at 4°. After
centrifugation at 100,000 x g for 1 hr, the absorbed serum was
reacted in an indirect immunofluorescence assay to determine whether
it still contained cross-reactive activity with mouse tissue. This absorp
tion procedure was repeated 4 times after which no anti-mouse activity
was detectable by indirect immunofluorescence;
the absorption pro
cedure was then repeated a final time.
This technique for antibody absorption was repeated with another
1.5-ml aliquot of 80 mg goat anti-human FN serum per ml using bovine
serum proteins (160 mg/ml) for the absorption. Before absorption,
very little cross-reactive anti-bovine activity was detected in immuno
fluorescence assays with sections of bovine liver. After 3 absorptions
with 12 mg of bovine serum protein per absorption, no anti-bovine
activity was detectable, and the goat anti-human FN serum was ab
sorbed with bovine serum proteins a final time.
Absorption of Rabbit Anti-human FN Serum to Remove Crossreactivity with Bovine Serum. Affinity-purified rabbit anti-human FN
antibody (33) was absorbed exhaustively to remove any cross-reactivity
with bovine serum proteins. Lyophilized antibody was diluted to a
concentration of 1 ¿ig/mlin normal bovine serum, incubated for 1 hr at
25°, and centrifugea at 100,000 x g for 60 min. A saturated (27%)
ammonium sulfate fraction of normal rabbit serum (ammonium sulfate:
serum, 1:1) also absorbed with normal bovine serum served as a
negative serum control. Frozen sections of bovine liver were stained
with the unabsorbed and absorbed sera to test the effectiveness of the
absorption.
Generation and FITC Labeling of Swine Anti-Goat IgG Antibody.
Thirty ml of saturated (27%) ammonium sulfate were added drop by
drop to 30 ml of normal goat serum, and the resulting immunoglobulinenriched fraction was further enriched by DEAE A-50 Sephadex col
umn chromatography (7, 23). The resulting goat globulin was used to
hyperimmunize a 30-kg swine. After immunization, a saturated (27%)
ammonium sulfate fraction of the hyperimmune serum (ammonium
sulfate:serum, 1:1) was made, and the globulin solution was further
enriched by DEAE A-50 Sephadex column chromatography (7). FITC
was then conjugated to this globulin protein (10); unbound FITC was
removed by Sephadex G-25 column chromatography (12). The labeled
protein was placed on a DEAE A-50 Sephadex chromatography column
(7) and eluted with a discontinuous phosphate:NaCI buffer (12). The
FITC:protein ratio of the conjugate was determined spectrophotometrically (10), and those fractions containing protein with a FITC:protein
ratio of 2.0 were pooled and used at a concentration of 2.5 mg/ml.
Fluorescent Staining of Frozen Sections and Tissue Culture Cells.
Both the 8-jim frozen sections and the 3- x 3-mm coverslips bearing
JANUARY
cell monolayers
were fixed with -70° acetone for 30 sec and then
washed 2 times for 5 min each time with DPBS. The sections or cells
were then incubated at 25° for 30 min with primary antibody (rabbit
anti-human GFAP serum, 1:10; or goat anti-human FN serum, 1:100)
or normal serum and DPBS controls. After three 5-min washes, FITCconjugated secondary antibody (Cappel Laboratories; 1:16) (goat antirabbit IgG serum or rabbit anti-goat IgG serum depending on the
species of primary antibody) was added, and the sections or cells were
again incubated for 30 min at 25°. The specimens were mounted in
50% DPBS:50% glycerol and viewed under a Zeiss universal micro
scope with FITC interference (500 nm) and No. 50 barrier filters.
Frozen sections were used because attempts to stain conventionally
prepared formalin-fixed paraffin-embedded tissues were unsuccessful
even when sections were subjected to trypsin or pepsin digestions (4)
prior to antibody application.
The double-staining technique which allowed visualization of both
GFAP and FN in the same specimen was accomplished by the use of
secondary antibody generated in a third species. The secondary anti
body for the primary rabbit anti-human GFAP serum was a rhodamineconjugated swine anti-rabbit IgG (Dako Corp.; 1:8) while the secondary
antibody for the goat anti-human FN serum was an FITC-conjugated
swine anti-goat IgG (2.5 mg/ml). In 2 control preparations, the antiGFAP serum was replaced with DPBS and with normal rabbit serum.
No rhodamine-conjugated
anti-rabbit IgG bound and only FITC fluores
cence was observed. In 2 other control preparations, the anti-FN serum
was replaced with DPBS and normal goat serum. In this case, only
rhodamine fluorescence was observed. These controls established that
the 2 secondary antisera were not binding to one another and each
was binding only to the species of IgG against which it was directed.
These double-stained preparations were viewed with a Zeiss universal
microscope first with FITC interference (500 nm) and No. 50 barrier
filters and then with rhodamine interference (546 nm) and No. 58
barrier filters.
Peroxidase-Antiperoxidase
Staining for FN. The peroxidase-antiperoxidase technique used to stain FN (38) was modified from an
already-described
protocol (13). Briefly, 8-ftm frozen sections were
rehydrated with 10% normal goat serum to reduce nonspecific binding
of the secondary antibody, washed 3 times with DPBS, and incubated
with the primary antibody (affinity-purified rabbit anti-human FN, 1 mg/
ml; or a saturated ammonium sulfate fraction of normal rabbit serum,
1 mg/ml) for 2 hr at 37°. The slides were again washed in DPBS and
immersed in anhydrous methanol with 0.3% H2O2 to block any endog
enous peroxidase activity. The blocking step was delayed until after
primary antibody exposure to avoid FN denaturation prior to antibody
binding. After a washing, the slides were incubated with secondary
goat anti-rabbit IgG serum (1:40) for 30 min at 25°, washed, and
incubated with rabbit peroxidase-antiperoxidase
(Sternberger
and
Meyer, Inc.) for 30 min at 25°.After washing, the slides were subjected
to a 7-min incubation with diaminobenzidine solution (Sigma Chemical
Co.) (5 mg diaminobenzidine in 10 ml 0.05 M Tris-HCI, pH 7.6, and 5
H\ 30% H2O2). The reaction was stopped by immersion in deionized
H2O, and the slides were counterstained with an aqueous solution of
1% methyl green, dehydrated through a graded alcohol series to
xylene, and mounted.
Detection of Human FN in the Plasma of Anaplastic Human
Glioma-bearing Athymic Mice. Tumors induced by the injection of U118 MG into athymic mice were used in the fifth serial mouse passage
(1, 3). Thirty /il of mechanically dispersed tumor were injected into the
flank s.c. space of 35 athymic mice. On Days 10, 13, 16, and 19, 5
were bled and the tumors were removed and weighed. Three mice
were killed on Day 22, and 4 were killed on Days 25 and 28. Four
animals were lost to disease. In all cases, citrated heart blood samples
were taken and plasma samples were frozen at -70° until the end of
the study when a previously described double-antibody
competition
radioimmunoassay (33) was used to determine human FN levels in the
plasma samples.
1982
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
169
T. R. Jones et al.
RESULTS
Immunoperoxidase Staining of FN in Normal Adult Human
Brain, Fetal Human Brain, and Anaplastia Human Gliomas.
Frozen sections of 2 fetal (16 and 22 weeks) and 2 adult
human brains were stained immunohistochemically
for the
presence of FN. In all cases of both adult and fetal tissues, the
capillary walls contained substantial amounts of reaction prod
uct. Positive staining ended abruptly at the edge of the capil
laries and all neurons and glia were negative. The walls of
arteries found in the adult brain tissue were stained markedly.
FN was observed throughout these walls and was not restricted
to the luminal surfaces (Fig. 1; Table 1).
Ten human glioblastomas were stained similarly, and FN was
present in all types of blood vessels found in the tumors (Fig.
2) while areas that were clearly tumor as judged by comparison
with adjacent hematoxylin and eosin-stained sections were
negative except in one case (120) where a mesh-like network
of fibrillar reaction product coursed between and around tumor
cells. Fine punctate deposits of FN were also occasionally
associated with the fibrillar FN (Fig. 3). Four tumors showed
areas of intense perivascular fibroblastic proliferation. In 3
tumors, this fibrous tissue was FN positive while the bordering
tumor was negative (Fig. 4), and in one tumor (120), the tumor
tissue was also positive (Table 2).
Immunofluorescent Staining of FN and GFAP in Primary
and Established Cell Lines. Five primary and 5 established
human cell lines were stained for FN expression by indirect
immunofluorescence using an anti-FN serum that is not reactive
with bovine FN. All 5 primary cell lines, Wl 38 (fetal lung
fibroblasts), D-187 (meningioma derived), D-173 MG (gliosarcoma derived), D-171b MG, and D171d MG (both glioma
derived), and one established cell line, U-118 MG (glioma
derived), showed significant amounts of FN staining. The pat
tern of FN expression was fibrillar with the fibrils forming an
extracellular mesh-like network between and over the cells.
Table 1
In situ expression of FN in normal adult and fetal human brain: an
immunoperoxidase study
Blood vessels
Neurons and glia
Normal adult human brain
A77137
A80281
Normal fetal human brain
F40359
N61866
+++
+—h + + , intensity of staining. -, failure to stain.
Table 2
In situ expression of FN in anaplastic human gliomas: an immunoperoxidase
study
Perivascular prolifBlood vessels
Tumor tissue
eration
NA'+
+ +"
91120165583667692702703735788+
+-t-+
++
++
NA+
++
NA+
++
NA+++
+ +
++++++
++++
NA+++
++
++++
++
NA
" H
1-+ + , intensity of staining. -, failure to stain. NA. not applicable; no
Tumor
fibroblastic
170
perivascular
proliferation
was found in sections studied.
Punctate areas of positive staining were noted around many of
the nuclei and represented either intracytoplasmic or substra
tum deposition of FN (Fig. 5). The remaining cell lines, 2T
(osteogenic sarcoma derived), D-54 MG, U-251 MG, and U251 MG sp (all glioma derived), were negative for FN. The CNS
tissue-derived cell lines were also stained for GFAP expression;
D-54 MG, D-1 71 b MG, D-1 71 d MG, D-173, D-187, and U-118
MG were negative for GFAP while in U-251 MG and U-251 MG
sp, GFAP was detected. FN and GFAP expression were not
seen together in the same cell line (Table 3).
Simultaneous Immunofluorescent
Staining of FN and
GFAP in Athymic Mouse-borne Anaplastia Human Gliomas.
Frozen sections of 8 anaplastic gliomas serially transplanted in
athymic mice were subjected to a double-label rhodaminefluorescein fluorescence technique so FN and GFAP could be
visualized in the same tissue section. The anti-FN serum used
was absorbed with mouse serum proteins and did not react
with mouse FN. Seven of 8 tumors contained FN, 7 of 8
contained GFAP, and 6 of 8 contained both. FN expression
presented as very fine fibrils and specks between cells while
GFAP appeared to be within cells usually producing a homo
geneous fluorescence in these sections rather than the char
acteristic fibrillary pattern seen commonly in attached cultured
cells. Any given area of a tumor could be negative for both,
positive for either, or positive for both (Table 4). Two tumors
(Tumors 120 and 391 ) showed areas that clearly demonstrated
expression of both FN and GFAP (Fig. 6).
Simultaneous Immunofluorescent Staining of FN and
GFAP in a Cell Line Derived from an Athymic Mouse-borne
Human Anaplastic Glioma. Athymic mouse-borne tumor 120
was placed in cell culture because both FN and GFAP were
detected in frozen sections. The cultured cells were stained for
both FN and GFAP; approximately 75% of the cells were
positive for GFAP while 20% expressed FN. Although they
constituted only about 10% of the total cells, cells positive for
both FN and GFAP were present (Fig. 7). The fibrillar pattern
of GFAP expression common to other GFAP-positive cell lines
was present in this cell line while the FN pattern of expression
was mostly of the punctate variety consistent with intracyto
plasmic or substratum expression. The extracellular striai pat
tern was, however, not uncommonly seen with the striae at
tached to GFAP-positive cells.
Immunoperoxidase Staining of an Established Anaplastic
Human Glioma-derived Cell Line and Its Clones. Coverslips
carrying cell monolayers of U-251 MG and 3 single-cell clones,
U-251 MG eli, cl2, and c!3, were stained for both FN and GFAP.
All 4 cell lines expressed marked amounts of GFAP in >95%
of their cells. The parent line (U-251 MG) and 2 of the clones
(eli and c!3) expressed no detectable amounts of FN. U-251
MG c!2, however, expressed moderate amounts of FN in the
typical striai pattern (Fig. 8).
Detection of Human FN in the Serum of Anaplastic Human
Glioma-bearing Athymic Mice. Levels of human FN in the
plasma of athymic mice which were carrying actively growing
human glioma-derived (U-118 MG) tumors were measured. By
staining with immunofluorescence,
sections of these tumors
were demonstrated to contain FN. Four of 31 mice tested had
detectable levels of human FN in their plasma. Two of 4 of the
mice had tumors weighing more than 2.0 g and had human FN
plasma levels of 2.1 and 5.8 fig/ml. The third and fourth mice
carried tumors weighing 0.19 and 1.25 g and had human FN
CANCER
RESEARCH
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
VOL. 42
FN and GFAP Expression
cells. Paetau ef a/. (26) suggest this staining results from
leakage of plasma FN into the tumor through breaches in the
blood-brain barrier. Because such breaches have been docu
mented (43) and because plasma-borne FN (cold-insoluble
level
FN20
lineWI38D-1
Cell
globulin)
and cell surface FN have been shown repeatedly to
++a4
lungGliomaGliomaGliosarcomaMeningiomaOsteosarcomaGliomaGliomaGliomaGliomaPassage
be
indistinguishable
in heterologous antiserum-based assays
MGD-171dMGD-173MGD-1
7 1b
+3
+ +
+9
++ +
(17, 18, 44), any FN stained in the tumors could be either
+2
++
hematogenous or tumor in origin. The transplantation of the
872TD-54
+46274561 +
tumors into athymic mice and the use of a human FN specific
antiserum allowed us to study only tumor FN and not plasma
GU-118MGU-251
M
FN. The detection of FN in 7 of 8 tumors and the close
+53646GFAPND'————ND——+
++
MGU-251
association of FN and GFAP expression in 6 of 8 demonstrated
+++ +
MG spOriginFetal
+ ++
that glial tumor cells are capable of FN expression. This ques
H
H+ + , intensity of staining. -, failure to stain. ND, not done.
tion of mutual expression of FN and GFAP also has been
pursued in cell culture systems.
Table 4
Four studies with cultured neonatal rat CNS cells (20, 27,
FN and GFAP expression as detected by immunofluorescence in human
30, 39) concluded that GFAP and expression are not seen
gliomas grown in athymic mice
together, and Raff ef al. (30) stated that FN was expressed in
Tumor91112114115120132241391Passage
FN5
level
+"5
++
the CNS only by fibroblasts and leptomeningeal cells. Kennedy
+5
++
+++
et al. (16) performed a similar study with fetal human brain and
+5
++
+++
received similar results. Kavinsky and Garber (15), however,
+3
++
++ +
studied embryonal mouse brain cultures and found extensive
+34
++
++
++
FN expression by GFAP-positive cells. Our studies of cultured
+++
+3
+
+++
cells
include primary and established cell lines derived from
+ + +GFAP—+
++
human
gliomas. Six contained FN, 6 contained GFAP, and one
H
H- + , intensity of staining. -, failure to stain.
contained neither as determined by immunohistochemistry.
Two cell lines were stained positively for both. The data dem
plasma levels of 1.29 and 2.7 ¿tg/ml,respectively. Plasma from
onstrate the heterogeneous nature of FN expression among
2 of 2 mice with tumors weighing more than 2.0 g were FN
glioma-derived cell lines. The fact that other authors (20, 23,
positive, one of 7 mice with tumors between 1.0 and 2.0 g was
27, 30, 39) have not found simultaneous expression of FN and
human FN positive, and one of 22 mice with tumors weighing
GFAP in CNS tissue-derived cell cultures is not contradictory
less than 1.0 g was FN positive (Chart 1).
to our data. The cell lines we have studied were derived from
neoplastic tissues and have been serially passed in culture
DISCUSSION
while the cells in other studies were primary expiant cultures
derived from normal tissues. The limited number of established
Studies of FN distribution within the CNS can be affected by
GFAP-positive cell cultures (1 ) has made studies of GFAP and
several factors. FN expression in other cell types has been
FN expression in transformed cells difficult, but this situation
shown to be transformation sensitive (11, 41, 45), cell cycle
may improve with the recent generation of a larger number of
sensitive (25), and dependent on cell environment (i.e., in vivo
Immunofluorescent
Table 3
staining of FN and GFAP in primary and established human
cell lines
or in vitro) (5). Changes in assay sensitivity can also cause an
apparent change in FN expression. These sources of variation
are important factors in evaluating FN expression in the CNS.
Immunoperoxidase
staining of frozen sections of normal
adult and fetal brain in this study revealed FN only in the walls
of blood vessels. Schachner ef al. (36) studied paraformaldehyde-perfused nonprimate vertebrate CNS tissue with immunocytochemical electron microscopy and found staining only
of the luminal surface of endothelial cells. The authors suggest
that this FN may be adsorbed from the plasma. In an immunofluorescent study at the light microscopic level, Paetau ef al.
(26) found FN patterns consistent with both vessel lumen and
basement membrane distribution. Our results also demonstrate
the presence of FN in the vessel wall and do not support the
claim of a luminal distribution only as made by Schachner et
al. (36).
Paetau étal.(26) also studied FN in human malignant gliomas
and found marked staining of the vasculature including the
glomeruloid vascular cell proliferations characteristic of glioblastoma. They also found some faint fluorescence in areas of
tumor. Our study also demonstrated FN in the walls of all types
of both normal and reactive blood vessels, and in one case
(120), positive staining was detected in areas containing tumor
JANUARY
6
5
Ë
-^ 4
o>
i
Z 3
o =13
a =16
» =19
•=22
•=25
» =28
0.4
40
80
12
16
2.0
TUMOR WEIGHT/g
24
2.8
32
36
Chart 1. Weights of anaplastia human glioma U-118 MG tumors grown in
athymic mice are plotted against the levels of human FN detectable in the plasma
samples of the animals. Inset legend indicates which day each tumor was
removed and weighed. Human FN plasma levels lower than 0.4 ¿ig/ml were
below the sensitivity of the assay. The lower limit of the assay is represented by
the dotted line; any tumors placed below it were taken from mice which had no
detectable human FN in the plasma.
1982
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
171
T. P. Jones et al.
GFAP-positive giioma-derived
cell lines (46).
Vaheri et al. (42) and Bigner et al. (1 ) analyzed a number of
well-characterized giioma-derived cell lines. Vaheri ef al. (42)
found that 2 of 5 cell lines contained FN when measured by
immunofluorescence, but all 5 were positive when cell extracts
were analyzed by the more sensitive radioimmunoassay. They,
however, did not concurrently study GFAP expression in these
cell lines. Bigner ef al. (1) found 14 of 15 permanent human
anaplastic giioma-derived cell lines contained detectable levels
of FN as measured by radioimmunoassay. Two cell lines (U251 MG and U-251 MG sp) contained low but measurable
amounts of FN, and immunofluorescent analysis revealed both
cell lines also contained marked amounts of GFAP. It was in a
clone of U-251 MG that we demonstrated by peroxidase immunohistochemistry both FN and GFAP. The cell culture data
in our study demonstrate that human cells of astroglial origin
as defined by the presence of GFAP are capable of FN expres
sion. While the radioimmunoassay techniques of Vaheri ef al.
(42) and Bigner ef al. (1) are capable of quantifying soluble
protein, the immunohistochemical methods used in this study
revealed both the perinuclear pattern of fluorescence consist
ent with intracytoplasmic FN expression and the striai fibrillar
pattern so characteristic of the FN extracellular matrix and also
allowed the localization of FN to GFAP-positive cells.
Several studies have followed FN secretion in vivo. Saba ef
al. (34) measured both bioassayable (opsonic) and immunoreactive plasma FN during growth of Sarcoma 180 in mice.
Immunoreactive FN paralleled tumor growth while bioassayable
FN dropped as tumor burden increased. Parsons ef al. (28, 29)
isolated a protein (MAD-2) from the serum and peritoneal and
pleural fluids of cancer patients. MAD-2 appears to be a
fragment of FN, and levels were usually increased above those
of nonneoplastic controls, but whether the tumor was the
source of the protein was unclear. We chose U-118 MG as the
cell line for our study of the feasibility of using FN to monitor
athymic mouse-borne tumor burden because it has the highest
level of FN synthesis of the well-characterized giioma-derived
cell lines proven to be tumorigenic in athymic mice (1, 3). Four
of 31 mice had detectable levels of human FN in their plasma;
3 of 4 had tumors which weighed 1.0 g or more. These results
show that some tumor FN can gain access to the blood, but
levels were too low to be useful as a tumor burden monitor with
this particular tumor. Possible explanations include a limited
ability of FN to cross into the blood stream, a change in FN
synthesis and/or shedding after heterotransplantation
to and
serial passage in athymic mice and the possibility that the
mouse may be acting as a human FN "sink." Oh ef al. (24)
have demonstrated that human FN injected into the blood
stream of mice enters the mouse tissues and assumes a distri
bution identical to endogenous mouse FN.
Having demonstrated that GFAP-positive cells derived from
anaplastic human gliomas could express FN, we then analyzed
FN expression as one element in the heterogeneous phenotypes seen with gliomas as a group. Only one in 10 surgical
biopsies expressed FN, but after serial passage in athymic
mice, 7 of 8 tumors were FN positive. This reversal may be the
result of 2 phenomena. In theory, the phenotype of athymic
mouse-borne tumors may evolve with passage; selective forces
may positively or negatively influence the survival and growth
of cellular subpopulations. On the other hand, the environment
in which a tumor exists rather than tumor progression may be
172
sufficient to alter phenotype (37). In this study, the change in
environment from human brain to mouse s.c. space may induce
the tumor to initiate or cease FN synthesis and expression. For
example, Chen ef al. (5) studied FN expression in tumorigenic
virally transformed rodent cells and found that 6 of 19 cell lines
were FN positive by immunofluorescence but 18 of 19 of the
cell lines were FN negative when grown as solid tumors. When
the tumors were returned to culture, FN expression resumed.
This demonstrated that the change was not due to an over
growth of FN-negative cells but rather a cessation of FN expres
sion and deposition by previously FN-positive cells and repre
sents a change in tumor behavior not induced by selection.
Although some authors (16, 27, 30) have claimed that GFAP
and FN expression are mutually exclusive, this study demon
strates that they are not. They can be seen together in the
same cell. This simultaneous expression in both athymic
mouse-borne anaplastic human gliomas and in anaplastic hu
man giioma-derived cell cultures supports the thesis that FN
expression is seen in transformed glia. Expression of these 2
molecules in anaplastic human gliomas should be considered
random with regard to one another until and unless a correlat
ing factor connecting them is demonstrated.
ACKNOWLEDGMENTS
The authors would like to express their appreciation to Dr. Lawrence F. Eng
for the kind gift of the anti-human GFAP serum and to Patricia Burks for her
secretarial assistance.
REFERENCES
1. Bigner, D. D.. Bigner, S. H., Pontén.J.. Westermark, B., Mahaley, M. S.,
Ruoslahti, E., Herschman, H., Eng, L. F., and Wikstrand, C. J. Heterogeneity
of genotypic and phenotypic characteristics of fifteen permanent cell lines
derived from human gliomas. J. Neuropathol. Exp. Neurol., 40: 201-229,
1981.
2. Blumenstock, F. A., Saba, T. M., Weber, P., and Laffin, R. Biochemical and
immunological characterization of human opsonic <>..SB glycoprotein: its
identity with cold-insoluble globulin. J. Biol. Chem., 253. 4287-4291,1978.
3. Bullard, D. E., Schold, S. C., Bigner. S. H., and Bigner, D. D. Growth and
chemotherapeutic response in athymic mice of tumors arising from human
giioma-derived cell lines. J. Neuropathol. Exp. Neurol., 40: 410-427, 1981.
4. Burns, J., Dixon. A. J., and Woods, J. C. Immunoperoxidase localization of
fibronectin in glomeruli of formalin fixed paraffin processed renal tissue.
Histochemistry, 67: 73-78, 1980.
5. Chen, L. B., Summerhayes, I., Hsieh, P., and Gallimore, P. H. Possible role
of fibronectin in malignancy. J. Supramol. Struct., 72: 139-150, 1979.
6. DeArmond, S. J., Eng, L. F., and Rubinstein, L. J. The application of glial
fibrillary acidic (GFA) protein immunohistochemistry
in neurooncology. Pathol. Res. Pract.. Õ68: 374-394, 1980.
7. Einstein, E. R., Richard, K.. and Cerutti, P. Quantitative determination of y
globulin in the sera by column chromatography. Anal. Biochem., 22: 1-10,
1968.
8. Eng, L. F., and Rubinstein, L. J. Contribution of immunohistochemistry
to
diagnostic problems of human cerebral tumors. J. Histochem. Cytochem.
26:513-522,
1978.
9. Grinnell. F., and Feld. M. K. Initial adhesion of human fibroblasts in serumfree medium: possible role of secreted fibronectin. Cell, ?7:117-129,1979.
10. Holborow, E. J., and Johnson, G. D. Immunofluorescence.
In: D. M. Weir
(ed.), Experimental Immunology, Ed. 1. pp. 571-596. Philadelphia: F. A.
Davis Co., 1967.
11. Hynes, R. O. Alteration of cell-surface proteins by viral transformation and
by proteolysis. Proc. Nati. Acad. Sei. U. S. A., 70: 3170-3174,
1973.
12. Johnson, G. D., Holborow, E. J., and Dorling, J. Immunofluorescence
and
immunoenzyme techniques. In: D. M. Weir (ed.), Handbook of Experimental
Immunology, Ed. 3, pp. 15.1-15.30. Oxford: Blackwell Scientific Publica
tions Ltd., 1978.
13. Jones, T. R., Bigner, S. H.. Schold, S. C., Eng, L. F.. and Bigner, D. D.
Anaplastic human gliomas grown in athymic mice: morphology and glial
fibrillary acidic protein expression. Am. J. Pathol., in press, 1981.
14. Kahn, P., and Shin, S. I. Cellular tumorigenicity in nude mice. J. Cell Biol.,
82: 1-16, 1979.
CANCER
RESEARCH
VOL.
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
42
FN and GFAP Expression
15. Kavinsky, C. J . and Garber, B. B. Fibronectin associated with the glial
component of embryonic brain cell cultures. J. Supramol. Struct., 11: 269281, 1979.
16. Kennedy, P. G. E., Lisak, R. P., and Raff, M. C. Cell type-specific markers
for human glial and neuronal cells in culture. Lab. Invest. 43: 342-351,
1980.
17. Kuusela, P., Vaheri, A., Palo, J., and Ruoslahti, E. Demonstration of fibronectin in human cerebrospinal fluid. J. Lab. Clin. Med., 92: 595-601, 1978.
18. Macarak, E., Kirby. E., Kirk, T., and Kefalides, N. Synthesis of cold-insoluble
globulin by cultured calf endothelial cells. Proc. Nati. Acad. Sei. U. S. A.,
75. 2621-2625,
1978.
19. Mautner, V., and Hynes, R. O. Surface distribution of LETS protein in relation
to the cytoskeleton of normal and transformed cells. J. Cell. Biol., 75. 743768, 1977.
20. Minier, L. N., Lasher, R. S., and Erickson, P. F. Distribution of the LETS
protein (fibronectin) in rat cerebellum. Cell Tissue Res., 214: 491-500,
1981.
21. Morrison, P. R., Edsall, J. T., and Miller, S. G. Preparation and properties of
serum and plasma proteins. XVIII. The separation of purified fibrinogen from
fraction I of human plasma. J. Am. Chem. Soc., 70. 3103-3108,
1948.
22. Mosher. D. F. Fibronectin. Prog. Hemostasis Thromb.. 5: 111-151, 1980.
23. Murphy, J. B., and Kies. M. W. Note on spectrophotometric determination of
proteins in dilute solutions. Biochim. Biophys. Acta, 45: 382-384, 1960.
24. Oh, E., Pierschbacher, M., and Ruoslahti, E. Deposition of plasma fibronectin
in tissues. Proc. Nati. Acad. Sei. U. S. A., 78. 3222-3224,
1981.
25. Orly, J., and Sato. G. Fibronectin mediates cytokinesis and growth of rat
follicular cells in serum-free medium. Cell, 17. 295-305, 1979.
26. Paetau, A., Mellstrom, K.. Vaheri, A., and Haltai, M. Distribution of a major
connective tissue protein, fibronectin, in normal and neoplastic human
nervous tissue. Acta Neuropathol., 51: 47-51, 1980.
27. Paetau, A., Mellstrom, K., Westermark, B., Dahl, D., Haltia, M., and Vaheri,
A. Mutually exclusive expression of fibronectin and glial fibrillary acidic
protein in cultured brain cells. Exp. Cell Res., 129: 337-344, 1980.
28. Parsons, R. G.. Aldenderfer, P. H., and Kowal, R. Detection of a human
serum DNA-binding protein associated with malignant disease. J. Nati.
Cancer Inst., 63. 43-47, 1979.
29. Parsons, R. G., Todd, H. D., and Kowal, R. Isolation and identification of a
human serum fibronectin-like protein elevated during malignant disease.
Cancer Res., 39. 4341-4345,
1979.
30. Raff, M. C., Fields, K. L., Hakomori, S.-l., Mirsky, R., Pruss, R. M.. and
Winter, J. Cell-type-specific markers for distinguishing and studying neurons
and the major classes of glial cells in culture. Brain Res., 174 283-308,
1979.
31. Ruoslahti, E., and Vaheri, A. Interaction of soluble fibroblast surface antigen
with fibrinogen and fibrin. J. Exp. Med., 141: 497-501, 1975.
JANUARY
32. Ruoslahti, E., Vaheri, A., Kuusela, P., and Linder, E. Fibroblast surface
antigen: a new serum protein. Biochim. Biophys. Acta, 322; 352-358,1973.
33. Ruoslahti. E., Vuento, M.. and Engvall, E. Interaction of fibronectin with
antibodies and collagen in radioimmunoassay.
Biochem. Biophys. Acta,
534. 210-218. 1978.
34. Saba, T. M., Gregory, T. J., and Blumenstock, F. A. Circulating immunoreactive and bioassayable opsonic plasma fibronectin during experimental
tumour growth. Br. J. Cancer, 41: 956-965, 1980.
35. Saba, T. M., and Jaffe, E. Plasma fibronectin (opsonic glycoprotein): its
synthesis by vascular endothelial cells and role in cardiopulmonary integrity
after trauma as related to reticuloondotheli.il
function. Am. J. Med., 68
577-594, 1980.
36. Schachner. M., Schoonmaker. G., and Hynes, R. O. Cellular and subcellular
localization of LETS protein in the nervous system. Brain Res.. Õ58. 149158, 1978.
37. Scherer. H. J. The pathology of cerebral gliomas. J. Neurol. Psychiatry, 3.
147-177, 1940.
38. Sternberger, L. A. Immunocytochemistry,
Ed. 2, pp. 104-169. New York:
John Wiley & Sons, 1979.
39. Stieg. P. E., Kimelberg. H. K., Mazurkiewicz, J. E., and Banker, G. A.
Distribution of glial fibrillary acidic protein and fibronectin in primary astroglial cultures from rat brain. Brain Res., )99. 493-500. 1980.
40. Vaheri, A., and Ruoslahti. E. Disappearance of a major cell-type specific
surface glycoprotein antigen (SF) after transformation of fibroblasts by Rous
sarcoma virus. Int. J. Cancer. 13: 579-586, 1974.
41. Vaheri, A., and Ruoslahti, E. Fibroblast surface antigen produced but not
retained by virus-transformed human cells. J. Exp. Med.. 142: 530-535,
1975.
42. Vaheri, A., Ruoslahti, E.. Westermark, B., and Pontén,J. A common celltype specific surface antigen in cultured human glial cells and fibroblasts:
loss in malignant cells. J. Exp. Med., J43. 64-72, 1976.
43. Vick, N. A., Khandekar, J. D., and Bigner, D. D. Chemotherapy of brain
tumors. Arch. Neurol., 34: 523-526, 1977.
44. Vuento, M., Wrann, M., and Ruoslahti, E. Similarity of fibronectins isolated
from human plasma and spent fibroblast culture medium. FEBS Lett., 82.
227-231. 1977.
45. Wartiovaara. J.. Linder. E.. Ruoslahti, E., and Vaheri, A. Distribution of
fibroblast surface antigen. J. Exp. Med., »40:1522-1533,
1974.
46. Westermark, B.. Larrson. E., Brunk, U.. Lubitz, W.. and Mark. J. Establish
ment of attached and non—attached cell lines from an uncommon human
glioma. Int. J. Cancer, 28. 341-351, 1981.
47. Yamada, K. M., Yamada, S. S.. and Pastan, I. Cell surface protein partially
restores morphology, adhesiveness, and contact inhibition of movement to
transformed fibroblasts. Proc. Nati. Acad. Sei. U. S. A., 73: 1217-1221,
1976.
1982
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
173
T. P. Jones et al.
1A
l B
3B
Fig. 1. Frozen sections of human brain stained by the peroxidase-antiperoxidase
technique for FN. In A, the walls of these capillaries in normal fetal brain are
positive (arrow) while neurons and glia do not stain, x 425. In B, FN distribution in normal adult brain is also restricted to the blood vessels (arrows), x 425. In C, this
artery in normal adult brain has FN throughout its wall, x 425.
Fig. 2. Peroxidase-antiperoxidase
stain of FN in a frozen section of glioblastoma 702. The walls of all types of blood vessels (arrows) in this tumor contain FN.
x 100.
Fig. 3. Serial frozen sections of glioblastoma 120 stained by H & E and by peroxidase-antiperoxidase
for FN. In A, an area containing tumor cells is shown. H &
E, x 425. In B, a serial section stained for FN shows a mesh-like network of fibrillar reaction product (large arrow) coursing around cells. Punctate deposits of (small
arrow) are also present, x 425.
174
CANCER
RESEARCH
VOL.
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
42
FN and GFAP Expression
/-
-
£>
'•
Fig. A. Serial frozen sections demonstrating perivascular fibroblastic proliferation and FN expression in Tumor 692. In A, perivascular fibroblastic proliferation
coursing around nesjs of tumor cells (arrows) are shown. H & E, x 100. In 8, these fibroblastic proliferations are stained markedly for FN (arrows) while surrounding
tumor tissue is negative. Light methyl green counterstain, x 100. In C, an area of marked fibroblastic proliferation (arrows) rests between nests of tumor cells. H &
E, x 100. In D. fibroblastic tissue is positive for FN in this peroxidase-antiperoxidase
stain (arrows); tumor cells are negative. Light methyl green counterstain, x 100.
Fig. 5. In this anaplastic human glioma-derived cell line (171 d MG) stained by indirect immunofluorescence for FN, both the fibrillar extracellular network (double
arrows) and the punctate perinuclear (single arrow) patterns of expression are demonstrated, x 425.
JANUARY
1982
175
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
T. R. Jones et al.
Fig. 6. Frozen sections of athymic mouse-borne anaplastia human gliomas stained simultaneously in the same section by rhodamine-fluorescein
double indirect
immunofluorescence for FN and GFAP. In A, most cells in tumor 120 are positive for GFAP as demonstrated by intense rhodamine fluorescence, x 425. In B, the
same section is also rich in FN as revealed by the presence of fluorescein fluorescence, x 425. In C, tumor 391 has one area rich in GFAP while other cells are
negative, x 425. In D, FN is seen in the same area as the GFAP. x 425.
Fig. 7. Rhodamine-fluorescein
double indirect immunofluorescent staining of the same cells of tumor 120 in culture. In A, this cell (arrow) is rich in GFAP. X 425.
In B. the same cell is also expressing FN (arrow), x 425. In C. 2 cells contain GFAP (arrows), x 425. In D. the 2 cells in C, among others, also stain for FN (arrows),
x 425.
176
CANCER
RESEARCH
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
VOL.
42
FN and GFAP Expression
.
8B
Fig. 8. Anaplastia human glioma-derived cell line (U-251 MG cl..) stained by the peroxidase-antiperoxidase
technique. In A. greater than 95% of the cells contain
GFAP. Methyl green counterstain, x 425. In B, the fibrillar FU forms a network which runs between and over the cells (arrows). Methyl green counterstain, x 425.
JANUARY
1982
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.
177
Fibronectin and Glial Fibrillary Acidic Protein Expression in
Normal Human Brain and Anaplastic Human Gliomas
Trevor R. Jones, Erkki Ruoslahti, S. Clifford Schold, et al.
Cancer Res 1982;42:168-177.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/42/1/168
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1982 American Association for Cancer Research.