Báo cáo khoa học: Intracellular trafficking of endogenous fibroblast growth factor-2 ppt

Intracellular trafficking of endogenous fibroblast growth
factor-2
Simona Taverna, Salvatrice Rigogliuso, Monica Salamone and Maria Letizia Vittorelli
Dipartimento di Biologia Cellulare e dello Sviluppo, Universita
`
di Palermo, Italy
Most proteins destined for secretion into the extracellu-
lar matrix are targeted by the presence of N-terminal
signal peptides, which direct their translocation into
the rough endoplasmic reticulum (rER). They are sub-
sequently transferred to the Golgi apparatus and then
secreted into the extracellular space. However, a grow-
ing number of secreted proteins have been identified
that lack N-terminal signal peptides. These proteins do
not enter the rER and their secretion is not influenced
by drugs such as brefeldin A and monensin, which
block secretion by classical mechanisms [1]. Therefore,
they are secreted by alternative, unconventional pro-
cesses; these proteins include the inflammatory cytokine
interleukin-b1, galectins, macrophage migration inhibi-
tory factor, acid and basic fibroblast growth factors
(FGF-1, FGF-2) and sphingosine kinase 1. The mecha-
nisms of their secretion are the subject of numerous
studies and different pathways have been described [2].
An unconventional secretion mechanism common
to several signalling proteins devoid of the typical
signalling sequence appears to be mediated by the
shedding of vesicles into the extracellular matrix. These
vesicles, also called exovesicles, are observed to bud
from the cell membranes and to be released into the
Keywords

FGF-2; microfilaments; microtubules;
secretion of leaderless proteins; shed
vesicles
Correspondence
M. L. Vittorelli, Dipartimento di Biologia
Cellulare e dello Sviluppo, Universita
`
di
Palermo, Viale delle Scienze ed. 16,
90128 Palermo, Italy
Fax: +39 0 9165 77430
Tel: +39 0 9165 77407
E-mail:
(Received 2 August 2007, revised 18
January 2008, accepted 31 January 2008)
doi:10.1111/j.1742-4658.2008.06316.x
We have previously reported how the release of fibroblast growth factor-2
(FGF-2) is mediated by shed vesicles. In the present study, we address the
question of how newly synthesized FGF-2 is targeted to the budding
vesicles. Considering that in vitro cultured Sk-Hep1 hepatocarcinoma cells
release FGF-2 and shed membrane vesicles only when cultured in the pres-
ence of serum, we added serum to starved cells and monitored intracellular
movements of the growth factor. FGF-2 was targeted both to the cell
periphery and to the nucleus and nucleolus. Movements toward the cell
periphery were not influenced by drugs affecting microtubules, but were
inhibited by cytocalasin B. Involvement of actin in FGF-2 trafficking
toward the cell periphery was supported by coimmunoprecipitation and
immune localization experiments. Colocalization of FGF-2 granules mov-
ing to the cell periphery and FM4-64-labelled intracellular lipids were not
observed. Ouabain and methylamine, two inhibitors of FGF-2 release, were

analyzed for their effects on FGF-2 intracellular localization and on vesicle
shedding. Ouabain inhibited FGF-2 movements toward the cell periphery.
The FGF-2 content of shed vesicles was therefore reduced. Methylamine
inhibited vesicle shedding; in its presence, FGF-2 clustered at the cell
periphery, but the rate of its release decreased. FGF-2 targeting to the
nucleus and nucleolus was not affected by cytocalasin B, whereas it was
inhibited by drugs that modify microtubule dynamics. Neither ouabain,
nor methylamine interfered with FGF-2 translocation to the nucleus and
nucleolus. FGF-2 targeting to the budding vesicles and to the nucleus and
nucleolus is therefore mediated by fundamentally different mechanisms.
Abbreviations
CM, conditioned media; FGF-2, fibroblast growth factor-2; FITC, fluorescein isothiocyanate; NLS, nuclear localization sequence;
rER, rough endoplasmic reticulum; uPA, urokinase type of plasminogen activator.
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1579
extracellular medium. The vesicle diameters are in the
range 100–1000 nm; the vesicle composition and func-
tion depend on the type of cells from which they have
been produced. Vesicles are found to be involved in
cell motility and tumour progression mechanisms as
well as in bone formation [3]. Involvement of shed ves-
icles in cell–cell and cell–matrix interactions is partially
mediated by the presence of several membrane-associ-
ated proteolytic enzymes [4–6] and signalling molecules
[7–9]. However, shed vesicles also channel the secretion
of several leaderless proteins that apparently accumu-
late in their internal lumen. For example, Cooper and
Barondes [10] reported that lectin 14, a signalling pro-
tein involved in muscle differentiation, is released as a
component of budding vesicles; MacKenzie et al. [11]
and Bianco et al. [12] reported that interleukin-b1is

secreted as a component of shed vesicles in response to
ATP acting on P2X receptors.
Pathways targeting specific molecules to the budding
vesicles have not been clarified. In a previous study,
we investigated the mechanism of FGF-2 secretion and
reported that FGF-2 is released from Sk-Hep1 cells
and from NIH 3T3 cells transfected with FGF-2
cDNA through vesicle shedding [13]. FGF-2, also
known as basic fibroblast growth factor, belongs to
the fibroblast growth factor superfamily. FGFs are
structurally related heparin-binding growth factors that
exhibit almost ubiquitous involvement in vertebrate
embryonic and fetal development [14], as well as in
many physio-pathological processes occurring in adult
organisms [15]. FGF-1 and FGF-2 are secreted from
the cell into the extracellular matrix, although they
lack the classical secretion sequence. Unlike FGF-1,
which appears to be released in response to stress con-
ditions as a component of a multiprotein complex
[16,17], FGF-2 is secreted constitutively; however, its
release by in vitro cultured cells is inhibited by serum
deprivation [18]. FGF-2 transmits pro-angiogenic
[19,20] and pro-lymphangiogenic [21] signals. It is also
involved in inducing smooth muscle and endothelial
cell growth and in regulating early development stages
of various organs [22], including the brain [23,24].
FGF-2 is also one of the most significant regulators of
human embryonic stem cell self-renewal and of cancer
tumourigenesis [25].
As a signalling protein, FGF-2 acts both directly at

the nuclear level and after secretion. Indeed, after its
synthesis, FGF-2 can be secreted [1] or transferred to
the nucleus and to the nucleolus where it behaves like
a transcriptional factor, inducing cell growth and
rRNA synthesis [26].
Five FGF-2 isoforms (18, 22, 22.5, 24 and 34 kDa)
have been identified in humans, all of which present
some nuclear localization sequences. However, although
the 34 kDa isoform presents an arginine-rich repeat
domain similar to the nuclear localization sequence
(NLS) present in HIV Rev protein, this sequence is
absent in isoforms with lower molecular weight; the 34,
24, 22.5 and 22 isoforms present several glycine-arginine
repeats identified as an NLS and a nonclassical bipartite
NLS in the C-terminus. [27]. The 18 kDa isoform only
presents the bipartite NLS. A portion of the bipartite
NLS regulates localization of the factor in nucleoli [28].
As a secreted molecule, FGF-2 interacts both with
matrix and membrane-bound proteoglycans and with
five members of a family of high-affinity tyrosine
kinase FGF receptors [29]. Receptor-mediated
signalling patterns appear to be utilized not only in
paracrine, but also in autocrine mechanisms [30].
According to Bossard et al. [31], receptor-bound
FGF-2 is endocytosed and can be transferred to the
nucleus along with associated proteoglycans; nuclear
Fig. 1. Targeting of endogenous FGF-2 to shed vesicles and nuclei. (A) Colocalization of FGF-2 and b1 integrin in Sk-Hep1 cells cultured in
3D type I collagen gels. Immunostaining of (a) FGF-2, (b) b1 integrin and (c) merging. FGF-2 was detected using Texas red-conjugated sec-
ondary antibodies; b1 integrin was detected using FITC-conjugated antibodies. Arrows indicate shed vesicles. Scale bar = 10 lm. (B) Time-
course of serum-induced FGF-2 intracellular movements observed by immunolocalization experiments. FGF-2 immunolocalization at 0, 15,

30, 45, 60 min after serum addition (a–e). Top: sections 1 lm from the surface; thin arrows indicate FGF-2 granules. Bottom: sections 3 lm
from the surface; thick arrows indicate area of the nuclei in which large variations of FGF-2 concentration are observed. FGF-2 was detected
using Texas red-conjugated secondary antibodies. Scale bar = 10 lm. (C) Number of FGF-2 positive granules (diameters in the range 0.01–
1 lm) in immunostained sections 1 lm from the surface of cells fixed 0, 15, 30, 45 and 60 min after serum addition. Granules were counted
by
IMAGEJ software. Asterisks indicate a significant difference between each time compared to the 30-min value. (D) FGF-2 concentration in
nuclei. FGF-2 concentration (absorbance) was evaluated by
IMAGEJ software in sections 3 lm from the surface of cells fixed 0, 15, 30, 45
and 60 min after serum addition. Asterisks indicate a significant difference between each time compared to the 30-min value. Values are
mean ± SD of 15 measurements from five independent experiments in (B) to (D). (E) Induction of uPA activity in GM7373 cells by SK-Hep1
vesicles. (a) Casein ⁄ plasminogen zymographies for detection of uPA activity, performed as described in the Experimental procedures on
extracts of GM7373 cells that had been incubated in 5 mL of media for 16 h with 0, 25, 50 and 100 lg of SK-Hep1 vesicles (lanes 1–4).
Asterisks indicate a significant difference between each amount compared to the previous one. (b) Densitometric analysis of lysis bands due
to uPA activity. Arbitrary units represent densitometric values of lysis bands, with 100 considered as the basal uPA activity of GM7373
untreated cells. Values are the mean ± SD of 15 measurements from five independent experiments.
Intracellular trafficking of FGF-2 S. Taverna et al.
1580 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS
accumulation of the endocytosed molecule requires
interaction with a protein called ‘translokin’ and
depends on the tubulin cytoskeleton.
In the present study, we analyzed mechanisms allow-
ing translocation of FGF-2 from cytoplasmatic sites to
the budding vesicles and ⁄ or to other cell districts.
Foetal bovine serum was reported to induce both
vesicle shedding [6,32] and FGF-2 secretion [1,18] and,
in our previous studies, we observed that the two phe-
nomena were clearly associated. Approximately 30 min
after the serum was added to starved cells, granules
containing FGF-2 appeared in the proximity of the cell
membrane and colocalization between FGF-2 and b1

integrin, a molecule known to be clustered in vesicle
membranes [6], became evident. Shortly after that,
FGF-2 was detected in shed vesicles [13].
a
b
c
d
e
B
C
E
D
A
a
b
c
d
e
a
b
c
a
c
25 µg 50 µg 100 µg
0
100
200
300
400
500

UpA activity-arbitrary unit
C
25
50
100
**
**
**
b
0
10
20
30
40
50
60
70
Number of FGF-2 granules
0
15
30
45
60
***
**
**
**
0
20
40

60
80
100
120
140
A in the nuclei
0
15
30
45
60
***
**
*
*
S. Taverna et al. Intracellular trafficking of FGF-2
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1581
We therefore aimed to follow the FGF-2 transloca-
tion pathway to the cell membrane by analyzing events
following serum addition. To check whether cytoskele-
ton components were mediating FGF-2 trafficking, we
analyzed the effects of cytoskeleton perturbation on
FGF-2 intracellular movements. To improve our anal-
ysis of FGF-2 intracellular movements and targeting,
we also tested the effects of known FGF-2 secretion
inhibitors on both FGF-2 intracellular movements and
vesicle shedding.
Results
Localization of FGF-2 in SK-Hep1 cells and their
shed vesicles

Figure 1A shows vesicle shedding by SK-Hep1 cells
cultured in 3D collagen gels. As previously reported
[13], when SK-Hep1 cells are grown in a complete
medium, FGF-2 is observed to be localized in cell pro-
trusions in association with b1 integrin, a molecule
that is also specifically clustered in shed vesicles [6]. In
3D gels, shed vesicles remain trapped in the collagen
and can be observed using confocal microscopy. Shed
vesicles, testing positive for both FGF-2 and b1 inte-
grin antigens, are visible in Fig. 1A (arrows).
As previously reported [13], the shedding of FGF-2
positive vesicles occurs in cells showing no signs of
apoptosis or necrosis. Cell cultures used for the present
experiments were kept in serum-free media for 72 h in
most instances and then stimulated by serum addition.
However, the percentage of apoptotic ⁄ necrotic cells
was always negligible (data not shown).
Immunolocalization of FGF-2 in starved cells (i.e.
cells kept for 72 h in a serum-free medium) and in cells
that, after starvation, were cultured in the presence of
10% fetal bovine serum for varying lengths of time,
showed that serum addition resulted in FGF-2 positive
granules, which were almost entirely absent in starved
cells (Fig. 1B), appearing at the cell periphery. During
the first 45 min of cell growth in the complete media,
the number of granules localized at the cell periphery
progressively increased; however, their number
dropped (Fig. 1B,C) after 1 h, indicating that FGF-2
had been released. As previously reported [13], 1 h
after adding serum, FGF-2 rich vesicles can be

Fig. 2. Involvement of cytoskeleton elements in FGF-2 intracellular trafficking. (A) Effects of nocodazol treatment on tubulin organization and
on serum induced FGF-2 movements. (a, b) Immunolocalization of tubulin, respectively, in control cells (a) and in cells fixed 30 min after the
addition of 10 l
M nocodazol (b). (c, d) FGF-2 immunolocalization in control cells fixed 30 min after serum addition (c) and in cells fixed
30 min after serum and 10 l
M nocodazol addition (d). Sections 2 lm from the surface. Tubulin was detected using TRITC-labelled secondary
antibodies; FGF-2 was detected using Texas red-conjugated secondary antibodies. Thin arrows indicate FGF-2 granules; thick arrows indicate
nucleoli. Scale bar = 10 lm. (B) Effects of cytocalasin B treatment on serum induced FGF-2 movements. (a, b) Immunolocalization of actin
respectively in control cells (a) and in cells fixed 30 min after the addition of 1 l
M cytochalasin B. (b) Sections 2 lm from the surface. Thin
arrows indicate FGF-2 granules; thick arrows indicate nucleoli. Actin was labelled with FITC-phalloidin. FGF-2 was detected using Texas red-
conjugated secondary antibodies. Scale bar = 10 lm. (C) FGF-2 concentration in nuclei 30 min after serum addition, in control cells and in
cells treated with drugs affecting the cytoskeleton organization. FGF-2 concentration (absorbance) was evaluated by
IMAGEJ software, as
described in the Experimental procedures, in sections 3 lm from the surface, 30 min after serum (control) or serum and the following drugs
had been added to starved cells: 10 l
M nocodazol (Nocadazol), 1 lM paclitaxel (Paclitaxel) 1 lM colchicine (Colchicine) and 1 lM cytochala-
sin B (Cytochalasin). Asterisks indicate a significant difference between the treated cells and controls. (D) Numbers of FGF-2 positive gran-
ules near the cell surface 30 min after serum addition in control cells and in cells treated with drugs affecting the cytoskeleton organization.
FGF-2 positive granules (diameters in the range 0.01–1 lm) were counted in immunostained sections 1 lm from the surface by
IMAGEJ soft-
ware, as described in the Experimental procedures. Control cells (Controls); cells treated with 10 l
M nocodazol (Nocadazol); 1 lM paclitaxel
(Paclitaxel); 1 l
M colchicine (Colchicine); 1 lM Cytochalasin B (Cytochalasin). Drugs and serum were added together. (E) Amount of vesicles
recovered from conditioned media of control cells and of cells treated with drugs affecting the cytoskeleton organization. Media were condi-
tioned by 3–6 h of growth of 4 · 10
7
control cells grown in complete medium (Control) and by cells treated for the same time with complete
medium to which the following drugs had been added: 10 l

M nocodazole (Nocadazole); 1 lM paclitaxel (Paclitaxel); 1 lM colchicine (Colchi-
cine) or 1 l
M cytochalasin B (Cytochalasin). Vesicle amount was evaluated by Bradford microassay method in at least three different experi-
ments. (F) Induction of uPA activity in GM7373 cells by vesicles shed by controls and by cells treated with drugs affecting the cytoskeleton
organization. Casein ⁄ plasminogen zymographies for detection of uPA activity were performed as described in the Experimental procedures
on GM7373 cells that had been incubated for 16 h with vesicles shed by control cells and by cells treated with drugs affecting the cytoskel-
eton organization. Vesicles had been obtained from ml of conditioned media. uPA activity of GM7373 cells incubated without vesicles (Med-
ium), with SK-Hep1 vesicles shed in media conditioned by 3–6 h of growth of 4 · 10
7
nontreated cells grown in complete medium (Control)
and by cells treated for the same length of time with complete medium to which the following drugs had been added: 10 l
M nocodazol
(Nocadazol); 1 l
M paclitaxel (Paclitaxel); 1 lM colchicine (Colchicine) and 1 lM cytochalasin B (Cytochalasin). With respect to cytochalasin B,
a comparison between the specific stimulatory activity of control vesicles and of vesicles shed by treated cells is also shown (Cytochalasin
sp. ac.) Arbitrary units represent densitometric values of lysis bands with 100 considered as the basal uPA activity of GM7373 cells. Values
are the mean ± SD of 15 measurements from five independent experiments.
Intracellular trafficking of FGF-2 S. Taverna et al.
1582 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS
recovered from cell-conditioned media (CM), whereas
FGF-2 is still undetected in vesicle-free CM. Therefore,
we suggest that, after 1 h of cell culture in the com-
plete medium, FGF-2 positive granules are released as
vesicle components.
Immunolocalization studies also showed that, when
cells were kept in the presence of serum for 30 min or
more, the average concentration of FGF-2 also
increased in the nucleus; moreover, the factor accumu-
lated in specific areas of the cell nucleus that were
FGF-2 negative (Fig. 1B,D) in starved cells. Because

FGF-2 is known to stimulate rRNA synthesis [26,27],
these nuclear areas are considered to correspond to
nucleoli.
FGF-2 can induce expression of the urokinase type
of plasminogen activator (uPA) in endothelial cells
[33,34]. We therefore performed assays of vesicle-asso-
ciated FGF-2 by testing the ability of vesicles to
induce uPA production in GM7373 endothelial cells.
We analyzed induction by incubating GM7373 cells
with vesicles for 16 h; uPA activity was then measured
by casein ⁄ plasminogen zymography of endothelial cell
extracts. Induction was dose-dependent (Fig. 1E);
moreover, we already knew that the stimulatory effect
of SK-Hep1 vesicles on uPA production by GM7373
cells was completely abolished by anti-FGF-2 serum
[13]. We therefore considered the specific stimulatory
effect of vesicles on uPA production as an indirect
method to evaluate their FGF-2 content.
Involvement of cytoskeleton in FGF-2 intracellular
trafficking
To analyze the involvement of cytoskeleton elements in
targeting endogenous FGF-2 to the cell nucleus and to
the cell periphery, Sk-Hep1 cells were treated with drugs
c
a
b
d
a
b
c

d
**
0
20
40
60
80
100
120
Proteins in shed vesicles
Control
Nocodazol
Paclitaxel
Colchicine
Cytochalasin
***
0
20
40
60
80
100
120
140
AB
C
D
E
F
A in the nuclei

Control
Nocodazol
Paclitaxel
Colchicine
Cytochalasin
0
20
40
60
80
100
120
140
160
180
UpA activity-arbitrary units
Medium
Control
Nocodazol
Paclitaxel
Colchicine
Cytochalasin
Cytochalasin sp. ac
***
**
0
20
40
60
80

100
120
140
Number of FGF-2 granules
***
Control
Nocodazol
Paclitaxel
Colchicine
Cytochalasin
S. Taverna et al. Intracellular trafficking of FGF-2
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1583
affecting tubulin or actin organization and FGF-2 im-
munolocalization was analyzed by confocal microscopy.
The molecules tested were nocodazol and colchicine,
which cause microtubule disassembly; paclitaxel, an
inhibitor of tubulin depolymerization; and cytochala-
sin B, a drug causing actin depolymerization.
The serum-induced increase of FGF-2 concentration
in the nucleus was less marked in cells treated with the
three drugs affecting microtubular organization. These
drugs also had clear inhibitory effects on FGF-2 locali-
zation in nuclear areas apparently corresponding to
nucleoli (Fig. 2A, panel d, thick arrows). On the other
hand, FGF-2 export toward the cell membrane was
not modified by treatment with drugs affecting micro-
tubular organization; in fact, granules at the cell
periphery were also clearly visible in treated cells
(Fig. 2A, panel d, thin arrows).
The effects of cytochalasin B treatment were the

reverse. As shown in Fig. 2B, FGF-2 immunolocaliza-
tion experiments demonstrated that actin depolymer-
ization did not modify nuclear localization of FGF-2
(Fig. 2B, panel d, thick arrows); however, cytochala-
sin B treatment blocked FGF-2 movements toward the
cell membrane.
To obtain a more quantitative picture detailing the
effects of drug treatments on cytoskeleton organization
and on FGF-2 movements toward the nucleus and the
cell periphery, the optical density in cell nuclei and the
number of FGF-2 granules at the cell periphery were
measured. As shown in Fig. 2C, nocodazol, colchicine
and paclitaxel had a significant inhibitory effect on
the increased concentration of FGF-2 induced by the
serum in the nucleus, whereas cytochalasin B had no
effect.
By contrast, as shown in Fig. 2D, the number of
FGF-2 positive granules observed at the cell periphery
30 min after the serum was added, was not modified
by treatment with drugs affecting microtubule organi-
zation, whereas it strongly decreased in cytochalasin B
treated cells.
In brief, and in accordance with the results pre-
viously reported by Bossard et al. [31] concerning
nuclear and nucleolar localization of exogenous
FGF-2, our data suggest that microtubule integrity is
needed for nuclear localization of endogenous FGF-2.
Actin filament integrity was not shown to be required
for targeting the factor to the cell nucleus. By contrast,
FGF-2 targeting to the cell periphery was not

influenced by treatment with drugs affecting tubulin
organization, whereas it was blocked by cytocalasin b
treatment.
We also tested the effects of drugs that influence the
cytoskeleton on cell attitude to shed membrane vesicles
and on vesicle capability to induce uPA production by
GM7373 cells. None of the three drugs affecting
microtubule organization had significant effects on the
amount of vesicles shed by treated cells (Fig. 2E), nor
on their stimulation of uPA production (Fig. 2F).
There is therefore no evidence of microtubule involve-
ment in the FGF-2 release mechanism. By contrast,
the amount of vesicles shed by cytochalasin B-treated
cells (Fig. 2E) was greatly reduced. Moreover, not only
the total, but also the specific stimulatory activity of
vesicles shed by cytochalasin B treated cells on uPA
production decreased (Fig. 2F).
Therefore, microfilament organization appears to be
required for both FGF-2 targeting to the budding vesi-
cles and for vesicle shedding.
a
b
c
2
45 kDa
actin
1
18 kDa
FGF-2
b

a
A
B
C
Fig. 3. Immunolocalization of FGF-2 in comparison with actin fila-
ments and lipidic structures. (A) Double staining for FGF-2 and
actin, 30 min after serum addition. FGF-2 was detected using
Texas red-conjugated secondary antibodies. Actin was labelled with
(a) phalloidin-labelled actin; (b) monoclonal antibody-labelled FGF-2;
and (c) merging. Arrow indicates FGF-2 granules on actin filaments.
Scale bar = 5 lm. Sections 1 lm from the surface. (B) Immunopre-
cipitation with anti-FGF-2 serum. Immunoprecipitate was immuno-
stained with either monoclonal antibodies against actin or
monoclonal antibodies against FGF-2. Lane 1, immunoprecipitation
from cells cultured in complete medium; lane 2, immunoprecipita-
tion from starved cells. (C) Double staining for FGF-2 and lipids.
FGF-2 was detected using FITC-conjugated secondary antibodies
(green fluorescence); lipids were stained with FM4-64 (red fluores-
cence). (a) Cells in serum-free media. (b) Cells 30 min after serum
addition. Sections 1 lm from the surface. Scale bar = 10 lm.
Intracellular trafficking of FGF-2 S. Taverna et al.
1584 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS
Further proof of the interaction between FGF-2
granules and the actin cytoskeleton was obtained by
double fluorescence experiments. Actin filaments were
labelled with phalloidin and FGF-2 with antibodies.
As shown in Fig. 3A, 30 min after serum addition,
FGF-2 positive granules were seen to localize on actin
filaments (Fig. 3A, arrow).
Actin involvement in FGF-2 release was confirmed

by immunoprecipitation experiments with anti-FGF-2
serum. However, as shown in Fig. 3B, actin was found
to be present only in immunoprecipitates obtained
from cells cultured in complete medium, and was
absent in immunoprecipitates obtained from starved
cells, thus indicating that FGF-2 ⁄ actin association
occurs when serum induces FGF-2 intracellular traf-
ficking toward the cell periphery.
Analysis of the intracellular localization of FGF-2
and of lipidic structures
To verify whether FGF-2 intracellular granules are
included in lipidic vesicles and whether they are targeted
to endolysosomal districts before being secreted, we
performed double fluorescence experiments labelling
lipids with FM4-64 (a red fluorescent molecule) and
FGF-2 with monoclonal antibodies recognized by
fluorescein isothiocyanate (FITC)-conjugated second-
ary antibodies. After being endocytosed (15 min of cell
culture in the presence of the dye), FM4-64 stains
intracellular lipidic structures [35]. As shown in
Fig. 3C, FGF-2 granules do not colocalize with FM4-
64-labelled intracellular lipidic vesicles in the absence
(Fig. 3C, panel a) or in the presence of serum (Fig. 3C,
panel b). Therefore, FGF-2 granules appear to move
on actin filaments without being enclosed in vesicles.
α sub
Na-K-ATPasi
actin
100 KDa
45 KDa

a
b
c
0
20
40
60
80
100
120
0
20
40
60
80
100
120
A in the nuclei
Control
Ouabain
Control
Ouabain
Control
Ouabain
Number of FGF-2
granules
**
*
0
20

40
60
80
100
120
UpA activity-arbitrary unit (2)
proteins in shed vesicles (1)
1 2
*
A
B C
D
E
Fig. 4. Effects of ouabain on FGF-2 intracellular trafficking. (A) Im-
munolocalization of FGF-2 in controls and in ouabain treated cells.
(a) Cells fixed in serum-free media; (b) cells fixed 30 min after
serum addition; (c) cells fixed 30 min after addition of serum and
100 l
M ouabain. Sections 1 lm from the surface. FGF-2 was
detected using Texas red-conjugated secondary antibodies. Scale
bar = 10 lm. (B) FGF-2 concentration (absorbance) in nuclei,
30 min after serum addition, in control cells and in cells treated
with ouabain. FGF-2 concentration (absorbance) was evaluated by
IMAGEJ software in sections 3 lm from the surface, 30 min after
serum (Control) or serum and 100 l
M ouabain (Ouabain) addition.
(C) Numbers of FGF-2 granules at the cell surface 30 min after
serum addition, in controls and in ouabain treated cells. FGF-2 posi-
tive granules (diameters in the range 0.01–1 lm) were counted in
immunostained sections 1 lm from the surface by

IMAGEJ software,
as described in the Experimental procedures. Controls, control cells
to which only serum was added; Ouabain, cells to which serum
and 100 l
M ouabain were added. Asterisks indicate a significant dif-
ference between ouabain treated cells and controls. (D) Amount of
vesicle recovered by control and ouabain treated cells and their
stimulatory effect on uPA expression by endothelial cells. (1)
Amount of vesicles recovered from complete medium conditioned
by 3 h of growth of 4 · 10
7
control cells (Control), and by 3 h of
growth of 4 · 10
7
cells in complete medium to which 100 lM oua-
bain had been added (Ouabain). (2) Induction of uPA activity in
GM7373 cells. Casein ⁄ plasminogen zymographies for detection of
uPA activity performed as described in the Experimental proce-
dures on GM7373 cells that had been incubated for 16 h with vesi-
cles obtained from 5 mL of media conditioned by 3 h of growth of
4 · 10
7
control cells grown in complete medium (Control) and by
3 h of growth of 4 · 10
7
cells in complete medium to which
100 l
M ouabain had been added (Ouabain). Arbitrary units repre-
sent densitometric values of uPA activity, where 100 is considered
as the basal uPA activity of GM7373 cells. Asterisks indicate a sig-

nificant difference between vesicles shed by Ouabain treated cells
and controls. (E) Detection of Na
+
⁄ K
+
-ATPase in shed vesicles.
Western blot analysis of Na
+
⁄ K
+
-ATPase in vesicles shed by
untreated (Control) and 100 l
M ouabain treated cells. Lane 1, vesi-
cles shed by untreated cells (30 lg of protein); lane 2, vesicles
shed by ouabain treated cells (30 lg of protein). Values are the
mean ± SD of 15 measurements from five independent experi-
ments.
S. Taverna et al. Intracellular trafficking of FGF-2
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1585
This result contradicts the hypothesis that shed vesicles,
such as exosomes [36], originate from multivesi-
cular bodies.
Effects of drugs inhibiting FGF-2 release
In another group of experiments, we tested two well-
known inhibitors of FGF-2 secretion, ouabain and
methylamine, for their effects on intracellular FGF-2
trafficking and on vesicle shedding.
Effects of ouabain
As shown in Fig. 4A, FGF-2 immunolocalization
experiments demonstrated that ouabain interferes with

FGF-2 intracellular trafficking toward the cell peri-
phery whereas it does not interfere with trafficking
toward the nucleus (Fig. 4A,B). FGF-2 positive gran-
ules, observed at the cell periphery 30 min after serum
addition, were virtually absent in cells treated with
both ouabain and serum (Fig. 4A,C).
Ouabain treatment did not modify the amount of
shed vesicles recovered from CM (Fig. 4D, bars
marked 1); however, when the stimulatory effect of
vesicles on uPA production by GM7373 cells was
tested, we observed that vesicles shed by ouabain-trea-
ted cells had a decreased specific activity on uPA
induction (Fig. 4D, bars marked 2). Therefore, oua-
bain interferes with FGF-2 intracellular trafficking
toward the cell periphery.
As ouabain is known to bind to a subunit of
Na
+
⁄ K
+
-ATPase [37,38], we looked for the presence
of this protein in vesicles. As shown in Fig. 4E, a
molecule of the expected apparent mass (100 kDa) was
marked by a monoclonal antibody against the a ATP-
ase subunit. The concentration of the recognized anti-
gen decreased in vesicles shed by ouabain-treated cells.
a
A

B

D
C
b
c
0
20
40
60
80
100
120
140
A at the cell periphery
0
20
40
60
80
100
120
A in the nuclei
0
1 2 3 4
20
40
60
80
100
Arbitrary unit
Control

Methylamine
Control
Methylamine
Control
Methylamine
*
*
*
Fig. 5. Effects of methylamine on FGF-2 intracellular trafficking. (A)
Immunolocalization of FGF-2 in controls and methylamine treated
cells. (a) Cells fixed in serum-free media; (b) cells fixed 30 min after
serum addition; (c) cells fixed 30 min after serum and 10 m
M
methylamine addition. Sections 1 lm from the surface. FGF-2 was
detected using Texas red-conjugated secondary antibodies. Scale
bar = 10 lm. (B) FGF-2 concentration (absorbance) in nuclei,
30 min after serum or serum and methylamine addition. FGF-2 con-
centration (absorbance) was evaluated by
IMAGEJ software in sec-
tions 1 lm from the surface, 30 min after serum (Control) or serum
and 10 m
M methylamine (Methylamine) addition. (C) FGF-2 concen-
tration (absorbance) at the cell periphery, 30 min after serum and
methylamine addition. FGF-2 concentration (absorbance) was evalu-
ated by
IMAGEJ software in sections 1 lm from the surface, 30 min
after serum (Control) or serum and methylamine addition (10
squares of length 2 lm were analyzed in each field). (D) Amount of
recovered vesicles and uPA activity induction. (1) Amount of vesi-
cles recovered from medium conditioned by 4 · 10

7
cells grown
for 3 h in complete medium and by cells treated for 3 h with com-
plete medium and 10 m
M methylamine. (2) uPA induction by 20 lg
of vesicles shed by cells grown in complete medium and by cells
treated for 3 h with complete medium and 10 m
M methylamine. (3)
uPA induction by vesicles obtained from 5 mL of media conditioned
by 4 · 10
7
cells grown in complete medium and by cells treated
for 3 h with complete medium and 10 m
M methylamine. (4) uPA
induction by 5 mL of vesicle-free media conditioned by
4 · 10
7
cells grown in complete medium and by cells treated for
3 h with complete medium and 10 m
M methylamine. Control,
untreated cells; Methylamine, cells treated for 3 h with 10 m
M
methylamine. Casein ⁄ plasminogen zymographies for detection of
uPA activity were performed as described in the Experimental pro-
cedures on GM7373 cells that had been incubated for 16 h with
vesicles obtained from media conditioned by 3 h of growth of
4 · 10
7
control cells grown in complete medium and by cells trea-
ted for 3 h with complete medium and 10 m

M methylamine. Arbi-
trary units represent densitometric values of uPA activity, where
100 is considered as the basal uPA activity of GM7373 cells. Val-
ues are the mean ± SD of 15 measurements each of 10 squares of
length 2 lm from three independent experiments.
Intracellular trafficking of FGF-2 S. Taverna et al.
1586 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS
Effects of methylamine
Treatment with methylamine did not affect FGF-2
movements toward the cell nucleus (Fig. 5A,B). As a
consequence of methylamine treatment, FGF-2
positive granules were observed to accumulate at the
cell periphery and to coalesce with the plasma mem-
brane, which became intensely positive for FGF-2
immunoreactive material (Fig. 5A). In methylamine-
treated cells, FGF-2 positive granules were not easily
counted; therefore, the main immunofluorescence at
the cell periphery was evaluated. Namely, in each
microscopic field, we selected 10 squares of length
2 lm at the cell periphery and analyzed them for
absorbance. Fluorescence at the cell periphery was not
significantly modified by methylamine treatment
(Fig. 5C). However, the amount of released vesicles
diminished (Fig. 5D, bars marked 1). The specific stim-
ulatory effect of vesicles on uPA production by
GM7373 cells was not modified (Fig. 5D, bars
marked 2); however, we observed a decrease in uPA
induction caused by adding the total amount of vesi-
cles recovered from a fixed volume of CM (Fig. 5D,
bars marked 3) to GM7373 cells. A proportional

decrease was also observed in uPA induction caused
by adding the same volume of vesicle-free CM
(Fig. 5D, bars marked 4) to the cells. Therefore, we
can conclude that the inhibitory effect of methylamine
on FGF-2 secretion is due to this molecule’s ability to
inhibit vesicle shedding.
Discussion
We previously reported that the release of FGF-2, a
pro-angiogenic growth factor that lacks the signalling
sequence typical of most secreted proteins, is mediated
by vesicle-shedding. Neither FGF-2 release, nor vesicle
shedding occurred when cells were kept in serum-free
media. Thirty minutes after serum addition, FGF-2
granules were seen at the cell periphery and, after 1 h,
FGF-2 rich vesicles were recovered from CM; FGF-2
was detected in vesicle-free CM only 2 h later [13].
The present study aimed to analyze intracellular
trafficking of endogenous FGF-2, targeting the mole-
cule to the budding vesicles and ⁄ or to different cell
districts. Our strategy was to follow the changes in
FGF-2 intracellular localization shortly after adding
serum, both in controls and in cells treated with drugs
affecting cytoskeleton organization or FGF-2 release.
To make a comparison of the FGF-2 content of vesi-
cles shed by control and treated cells, we performed
analyses of uPA activity in extracts of GM7373 endo-
thelial cells treated with vesicles. Induction of uPA
activity in endothelial cells is a typical response elicited
by FGF-2 [33,34], and we previously reported that the
stimulatory effect of SK-Hep1 vesicles on uPA produc-

tion by GM7373 cells was dose-dependent and fully
neutralized by monoclonal anti-FGF-2 serum [13].
Assays of cell vitality were performed in control and
treated cells and data were collected using cell cultures
in which the number of apoptotic or necrotic cells,
evaluated by acridine orange and trypan blue staining,
was found to be negligible (i.e. less than 3%).
When using untreated cells, we observed that some
FGF-2 positive granules were detected at the cell
periphery as early as 15 min after serum addition; their
number then grew for approximately 45 min. However,
after 1 h of culturing in complete medium, the number
of FGF-2 granules abruptly dropped. The decrease in
the number of FGF-2 granules at the cell periphery
was coupled with the appearance of FGF-2 rich vesi-
cles in CM. Therefore, the results were in agreement
with the previously described [13] vesicle-mediated
mechanism of FGF-2 release.
FGF-2 is a molecule with autocrine, paracrine and
intracrine signalling mechanisms. With respect to intra-
crine mechanisms, the factor has to be localized in the
nucleus. A detectable amount of FGF-2 was also
found to be present in the nucleus of starved cells;
however, when serum was added, the nuclear concen-
tration of the molecule increased considerably. We also
noticed that, in starved cells, FGF-2 was undetectable
in specific areas of the nucleus apparently correspond-
ing to nucleoli. After serum addition, these areas of
the cell nucleus became highly positive for FGF-2
immunostaining.

To analyze the involvement of cytoskeletal elements
in FGF-2 targeting, we tested drugs that alter either
tubulin or actin dynamism. Treatment with drugs
affecting the organization of microtubules (nocodazol,
colchicine and paclitaxel) did not interfere with FGF-2
granule trafficking toward the cell membrane, nor with
the amount of vesicles shed into the extracellular med-
ium. No differences were observed in the stimulatory
effects of vesicles shed by control and treated cells on
the uPA activity of endothelial cells. Therefore, the
results of these experiments show that FGF-2 export
toward the cell periphery is mediated by mechanisms
that do not require microtubule organization.
By contrast, treatment with drugs that affect micro-
tubule dynamics neither allowed for the serum-induced
increase of FGF-2 concentration in the cell nucleus,
nor the specific localization of FGF-2 in nucleoli.
Therefore, nuclear and nucleolar localization of endog-
enous FGF-2 appears to follow a pathway requiring
maintenance of the microtubule organization. As
S. Taverna et al. Intracellular trafficking of FGF-2
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1587
reported by Bossard et al. [31], tubulin organization
is also needed for nuclear targeting of endocytosed
FGF-2.
Treatment with cytochalasin B showed that endo-
genous FGF-2 export toward the cell periphery
requires the integrity of the actin cytoskeleton. These
treatments decreased FGF-2 transportation to the cell
periphery and to the budding vesicles. Consequently,

the specific stimulatory activity of vesicles on GM7373
uPA activity decreased. Interaction between FGF-2
and actin filaments was also demonstrated by coimmu-
noprecipitation and coimmunolocalization of the two
molecules. Moreover coimmunoprecipitation experi-
ments showed that interactions between FGF-2 and
actin were induced by serum addition. Cytochalasin B
considerably decreased the rate of vesicle shedding. On
the other hand, cytochalasin B treatment did not affect
FGF-2 movements toward the nucleus and nucleolus.
Therefore, the results indicate that FGF-2 movements
toward the cell periphery or the nucleus utilize totally
different mechanisms.
We also determined whether FGF-2 positive gran-
ules, observed at the cell periphery, were enclosed in
lipidic vesicles. However, colocalization experiments
performed using the lipidic colorant FM4-64 and anti-
FGF-2 serum showed that FGF-2 granules were not
associated with lipidic structures. These results indicate
a secretion mechanism that, unlike the production of
exosomes [36], does not involve multivescicular bodies.
In a different group of experiments, we analyzed
FGF-2 intracellular localization, the amount of shed
vesicles and the stimulatory effect of vesicles on uPA
production, using SK-Hep1 cells treated with drugs
known to inhibit FGF-2 secretion. The drugs tested
were ouabain and methylamine.
Ouabain is a well-known inhibitor of Na
+
⁄ K

+
-
ATPase which binds to the a subunit of Na
+
⁄ K
+
-
ATPase. However, the molecule was also reported to
inhibit FGF-2 secretion [37] and FGF-2 release was
reported to be ouabain-insensitive in cells expressing
an ouabain resistant ATPase a subunit. A possible
explanation of the inhibitory effect of ouabain on
FGF-2 secretion may be that the electrochemical gra-
dient created by the Na
+
⁄ K
+
-ATPase is a prerequisite
for intracellular FGF-2 movements toward the cell
periphery or for its secretion. However, FGF-2 and
ATPase a subunit were shown to coimmunoprecipi-
tate, and it was therefore suggested that, to allow
FGF-2 secretion, FGF-2 and ATPase a subunit had to
interact [38]. The reasons why this interaction was
needed were not clarified. Our results show that oua-
bain treatment blocks intracellular movements of
FGF-2 positive granules. In treated cells, FGF-2 gran-
ules were not observed to accumulate at the cell
periphery. The drug did not modify the rate of vesicle
shedding; nevertheless, vesicles shed by ouabain-treated

cells had a decreased specific stimulatory effect on
GM7373 cells, indicating a decreased FGF-2 content.
The results obtained by treating cells with ouabain
are in agreement with the vesicle-mediated mechanism
of FGF-2 secretion that we described; they show that
ouabain-mediated inhibition of FGF-2 secretion is
associated with a decrease in FGF-2 targeting to
budding vesicles. However, the mechanisms by which
ouabain affects FGF-2 movements toward the cell
periphery require further analysis.
A hypothetical explanation of the drug effect is that
direct interaction of the ATPase a subunit with FGF-2,
inhibited by ouabain, is needed for the two molecules to
be transferred to the budding vesicles. This hypothesis
is validated by our observation that ouabain treatment
caused a decrease in the vesicle concentration of both
FGF-2 and of the ATPase a subunit. However, it is dif-
ficult to explain how the two molecules come across and
are allowed to run into each other. We hypothesized
that interaction could occur at the level of multivesicu-
lar bodies or in other membrane-bound intracellular
districts. However, this hypothesis was contradicted by
the results of lipidic staining, which failed to show any
colocalization between intracellular FGF-2 granules
and lipidic structures. The site of interaction between
the ATPase a subunit and FGF-2 therefore remains
unknown. To explain how such interaction could occur
in the cell cytoplasm, we are led to suspect the existence
of a splice variant of the ATPase a subunit lacking the
N-terminal signalling sequence that drives the complete

protein to the rER. For the Na
+
⁄ K
+
-ATPase a sub-
unit, such a variant is not known; however, it was
described for an ouabain-sensitive H
+
⁄ K
+
-ATPase a2
subunit. It was suggested that the alternative splicing of
this molecule dictates isoform-specific differences in
membrane targeting or cytoskeletal association [39].
Different explanations of the effects of ouabain are
also possible. For example, FGF-2 trafficking to the
budding vesicles might be inhibited by ouabain in
response to one of the several signal transduction path-
ways modulated by the alkaloid interaction with the
Na
+
⁄ K
+
-ATPase a subunits [40].
The results obtained after methylamine treatment
clearly support the prominent role of shed vesicles in
FGF-2 secretion. Methylamine has been described to
inhibit both FGF-2 release [1] and vesicle shedding
[41]. The results contained herein indicate that the
methylamine inhibitory effect on FGF-2 release was

due to its action on vesicle shedding. FGF-2 granules
coalesced with the cell membrane but were not
Intracellular trafficking of FGF-2 S. Taverna et al.
1588 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS
released. The specific stimulatory activity of vesicles
shed by methylamine treated cells on uPA production
by endothelial cells was not modified compared to
vesicles shed by control cells; however, the amount
of recovered vesicles decreased. As a result, the
stimulatory effect obtained by adding vesicles recovered
from a fixed volume of CM to GM7373 cells decreased.
A parallel decrease in uPA induction was observed
when fixed volumes of vesicle-free CM were tested for
their uPA stimulatory effects on GM7373 cells.
In our previous study, we suggested that FGF-2 is
released into the medium from vesicle disruption. Vesi-
cles were shown to have a short half-life and, whereas
FGF-2 rich vesicles were already recovered from condi-
tioned medium after 1 h of cell culture in complete med-
ium, the presence of FGF-2 in vesicle-free CM was
detected only 2 h later [13]. The stimulatory effect of
vesicle-free media conditioned by methylamine treated
cells decreased in comparison to vesicle-free media con-
ditioned by control cells, and the percentage of this
decrease was parallel to the decrease in the amount of
vesicle-associated FGF-2. Therefore, the reduced amount
of FGF-2 in the medium can be explained by the inhi-
bitory effect of methylamine on vesicles shedding.
Although treatment of GM7373 cells with CM con-
taining methylamine for 16 h was tolerated, treatment

with CM containing drugs that affect the cytoskeleton
organization or with ouabain affected their vitality. For
this reason, the uPA stimulatory activity of vesicle-free
CM was only analyzed after methylamine treatment.
In conclusion, our results show that FGF-2 release
involves transporting the molecule in granules, which
are not associated to lipidic structures, along actin mi-
crofilaments. Ouabain inhibits intracellular movements
of FGF-2 via unknown mechanisms, whereas methyl-
amine restrains FGF-2 secretion by inhibiting vesicle
shedding. Targeting endogenous FGF-2 to nuclei and
nucleoli is achieved by a different mechanism, which is
mediated by microtubules and is not inhibited by
either ouabain or by methylamine.
Experimental procedures
Cells and culture media
Human SK-Hep1 hepatocarcinoma cells were grown in
DMEM (Celbio, Milan, Italy) supplemented with 10% fetal
bovine serum (Euroclone, Celbio). Bovine GM7373 fetal
aortic endothelial cells [42] were grown in Eagle’s MEM
(Celbio) supplemented with 10% fetal bovine serum, essen-
tial and non-essential amino acids. All the cells were nega-
tive for mycoplasma contamination through the Hoechst
33258 (Sigma-Aldrich, St Louis, MO, USA) staining assay.
SK-Hep1 cells were grown overnight in complete medium
to study the effect of serum on the intracellular distribution
of FGF-2, and then they were kept in serum-free medium
for 3 days with three medium changes. Medium supple-
mented with 10% fetal bovine serum was added and after
various culture durations, cells were fixed and the intracel-

lular distribution of FGF-2 and of other molecules were
analyzed using immunofluorescence assays.
Immunofluorescence assay
Cells, seeded at low density (2000 cells per well) onto
microscope coverslips in 12-well culture plates (Corning,
obtained from Celbio), were fixed in 3.7% formaldehyde in
NaCl ⁄ P
i
for 10 min followed by permeabilization with
0.1% Triton X-100 for 5 min. Tubulin was detected with
mouse monoclonal anti-tubulin serum (clone DM1A,
1 : 100; Sigma-Aldrich) and monoclonal anti-mouse
TRITC-conjugated serum (1 : 50; Sigma-Aldrich). Actin
was detected using FITC-labelled phalloidin (1 : 300;
Sigma-Aldrich). FGF-2 was detected with mouse monoclo-
nal anti-FGF-2 serum (0.5 mgÆmL
)1
; type II, Upstate
Biotechnology, Lake Placid, NY, USA) and Texas red-
conjugated anti-mouse serum (1 : 500; Amersham Bio-
sciences, Little Chalfont, UK). FGF-2 granules (diameters
in the range 0.01–1.0 lm) were counted by imagej software
(NIH Image, Bethesda, MD, USA) (approximately 30 cells
were present in each analyzed microscopic image). To eval-
uate immunofluorescence in nuclei, in each microscopic
image, 10 nuclei were selected and analyzed for absorbance.
To evaluate immunofluorescence in the cell periphery, 10
squares of length 2 lm were selected in each field and ana-
lyzed. Data are reported as arbitrary units taking 100 as
the absorbance control. For each experiment, a minimum

of three different fields were considered. The results shown
are mediated from data obtained in at least three indepen-
dent experiments.
Double staining for FGF-2 and lipids
FGF-2 was detected using FITC-conjugated secondary
antibodies; lipids were stained with FM4-64. FM4-64 was
dissolved in ice-cold NaCl ⁄ P
i
without calcium and magne-
sium and added at a final concentration of 5 lgÆmL
)1
.
Cells plated on coverslips were removed from the culture
medium and quickly immersed in the staining solution, on
ice, for 15 min. The coverslips were removed from the
staining solution, fixed with ice-cold 4% formaldehyde in
NaCl ⁄ P
i
without calcium and magnesium and placed on ice
for 10 min. Coverslips were rinsed three times with NaCl ⁄ P
i
without calcium and magnesium, mounted on a microscope
slide and analyzed. Immunostained cells were analyzed by
confocal microscopy (Olympus 1X70 with Melles Griot
laser system; Olympus, Tokyo, Japan).
S. Taverna et al. Intracellular trafficking of FGF-2
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1589
Cell treatments
SK-Hep1 cells were treated with cytochalasin B (1 lm),
nocodazol (10 lm), paclitaxel (1 lm), colchicine (1 lm),

ouabain (100 lm) and methylamine (10 mm) (all obtained
from Sigma-Aldrich). Treatments were performed by add-
ing serum and drugs at the same time, respectively, 30 min
before cells were fixed for immunolocalization experiments,
and 3 or 6 h before conditioned media were collected for
analyses on vesicles. For methylamine treatment, cells were
pre-incubated for 16 h with methylamine-containing com-
plete media, then transferred to fresh methylamine-contain-
ing complete media and kept for an additional 3 h.
Cells cultured in 3D type 1 collagen gels
Type 1 collagen (Rat tail BD Biosciences, Milan, Italy; sold
as a 3.9 mgÆmL
)1
solution in 0.02 m acetic acid) 500 lL,
was mixed v ⁄ v with cells (3 · 10
5
mL
)1
) suspended in 10%
fetal bovine serum-DMEM containing 50 mm NaHCO
3
,
and plated 100 lL per well, in 96-multiwell plates (Costar
3596, obtained from Celbio). As soon as the gel had poly-
merized, 200 lL of complete medium was added to each
well. After 4 days of incubation, cells were fixed by adding
3.7% formaldehyde (Sigma-Aldrich) in NaCl ⁄ P
i
for 10 min.
FGF-2 was detected with mouse monoclonal anti-FGF-2

serum (0.5 mgÆ mL
)1
; type II, Upstate Biotechnology) and
Texas red-conjugated anti-mouse serum (1 : 500; Amersham
Biosciences). b1 Integrin was detected with rat monoclonal
C27 anti-b1 integrin serum [43] (0.5 mgÆmL
)1
) and FITC-
conjugated anti-rat serum (1 : 200). Immunostained cells
were analyzed by confocal microscopy [Olympus 1X70 with
a Melles Griot laser system (Carlsbad, CA, USA)].
Vesicle purification from conditioned medium
Vesicles were purified from conditioned medium as
described [32]. Briefly, medium conditioned by subconfluent
healthy cells for 3, 6 or 24 h were centrifuged at 2000 g and
at 4000 g for 15 min. The supernatant was ultracentrifuged
at 105 000 g for 90 min. Pelleted vesicles were resuspended
in NaCl ⁄ P
i
.
The amount of isolated vesicles was determined by mea-
suring protein concentration through the Bradford micro-
assay method (Bio-Rad, Segrate, Milan, Italy) using bovine
serum albumin (Sigma-Aldrich) as a standard.
Western blotting
After electrophoresis by SDS ⁄ PAGE, the proteins were
blotted onto a nitrocellulose membrane (Hybond; Amer-
sham Biosciences), which was saturated with 5% nonfat
dry milk, 0.1% Tween 20 in NaCl ⁄ P
i

for 4 h and then
incubated with monoclonal anti-FGF-2 (1 lgÆmL
)1
; type II,
Upstate Biotechnology) or monoclonal anti-Na
+
⁄ K
+
-
ATPase (Upstate Biotechnology) primary sera followed by
horseradish peroxidase-conjugated antimouse IgG serum
(1 : 5000; Amersham Biosciences) for 1 h at room tempera-
ture. Immunocomplexes were visualized with the ECL
Western blotting kit (Amersham Biosciences) using Hyper-
film (Amersham Biosciences).
Immonoprecipitation assay
3 · 10
5
Sk-Hep1 cells were grown for 48 h in 20 mL of com-
plete medium or in serum-free medium. After harvesting the
cell culture medium, cells were washed twice with NaCl ⁄ P
i
and subsequently removed from the plate using 1 mL lysis
buffer (1% Triton X-100 and Protease Inhibitor Cocktail
Set III Calbiochem (Darmstadt, Germany), obtained from
Inalco S.P.A., Milan, Italy 1 : 1000, dissolved in NaCl ⁄ P
i
)
and incubated for 10 min at room temperature. Following
centrifugation (8000 g for 10 min), cell lysates were processed

for immunoprecipitation: CNBr activated-Sepharose 6MB
was suspended in mm HCl and washed for 15 min. Mono-
clonal anti-FGF-2 (1 lgÆlL
)1
; Upstate Biotechnology) was
dissolved in coupling buffer (0.5 m NaCl in 0.1 m NaHCO
3
,
pH 8.3) and mixed with the gel in rotation overnight at 4 °C.
Excess ligand was removed by washing Sepharose 6MB
with coupling buffer. All remaining active groups were
blocked by 2 h of rotational agitation at room temperature
with glycine (Sigma-Aldrich) (0.2 m), which had been dis-
solved in coupling buffer. Immune complexes were washed
with three cycles of alternating pH. Each cycle consisted of
a wash with 0.5 m NaCl in acetate buffer (0.1 m, pH 4),
followed by a wash with 0.5 m NaCl in Tris buffer (0.1 m,
pH 8).
Antibody-conjugated Sepharose 6MB was mixed with
250 lg of protein extract from cells grown either in com-
plete or in serum-free medium and maintained in rotation
overnight at 4 °C.
Immune complexes were then eluted into Laemmli SDS
gel sample buffer. After 7.5% SDS-polyacrylamide gel elec-
trophoresis under reducing conditions, proteins were trans-
ferred to a nitrocellulose filter, and analyzed as described
for western blotting experiments with one of the following
probes: (a) 1 : 500 dilution of the monoclonal anti-actin
serum (Sigma-Aldrich) or (b) 1 : 500 dilution of a monoclo-
nal anti-FGF-2 serum (Upstate Biotechnology).

Assay for uPA activity
GM7373 cells (approximately 70 000 cellsÆcm
)2
) were grown
for 16 h in 5 mL of medium supplemented with varying
concentrations of vesicles shed by Sk-Hep1 control cells or
by Sk-Hep1 cells treated with various drugs. GM7373 cell
cultures were washed twice with NaCl ⁄ P
i
and incubated for
18 h in serum-free medium. After harvesting the culture
Intracellular trafficking of FGF-2 S. Taverna et al.
1590 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS
supernatant, the cells were washed twice with NaCl ⁄ P
i
and
detached with a scraper. Following centrifugation at 800 g
for 10 min at 4 °C, protein concentration was measured
using the Bradford method. Cell extracts (30 lg of proteins)
were loaded onto 7.5% SDS ⁄ polyacrylamide gels. Zymo-
graphy was performed as described [44] with overlay gels
containing 3% nonfat dry milk and 40 lgÆmL
)1
of bovine
plasminogen (Sigma-Aldrich). Densitometric analysis of
the lysis bands was performed using Eastman Kodak Co.
science 1d image analyzer software (Eastman Kodak,
Rochester, NY, USA). White bands of caseynolysis of
images were converted to dark bands to better display the
activity. Histograms were constructed using the results of

at least three totally independent experiments.
Statistical analysis
Results are expressed as the mean ± SD or number and
percentage. Data were analyzed using Student’s t-test to
assess the difference between control and treated samples.
P < 0.001 (***); < 0.005 (**) or < 0.05 (*) was consid-
ered statistically significant.
Acknowledgements
This work was supported by the Italian Ministry of
University and Scientific and Technological Research
and by the Italian Association for Cancer Research
(AIRC) (M. Letizia Vittorelli). Simona Taverna was
supported by a fellowship from AIRC. We would like
to thank Salvatore Agnello for his assistance with the
confocal microscopy and Italia Di Liegro for critically
reading the manuscript.
References
1 Mignatti P, Morimoto T & Rifkin D (1992) Basic fibro-
blast growth factor, a protein devoid of secretory signal
sequence, is released by cells via a pathway independent
of the endoplasmic reticulum-Golgi complex. J Cell
Physiol 151, 81–93.
2 Nickel W (2003) The mystery of nonclassical protein
secretion. Eur J Biochem 270, 2109–2119.
3 Vittorelli ML (2003) Shed membrane vesicles and
clustering of membrane-bound proteolytic enzyme. Curr
Top Dev Biol 54, 411–432.
4 Zucker S, Wieman JM, Lysik RM, Wilkie DP,
Ramamurthy N & Lane B (1987) Metastatic mouse
melanoma cells release collagen-gelatin degrading

metalloproteinases as components of shed membrane
vesicles. Biochim Biophys Acta 924, 225–237.
5 Ginestra A, Monea S, Seghezzi G, Dolo V, Nagase H,
Mignatti P & Vittorelli ML (1997) Urokinase
plasminogen activator and gelatinases are associated
with membrane vesicles shed by human HT1080
fibrosarcoma cells. J Biol Chem 272, 17216–17222.
6 Dolo V, Ginestra A, Cassara
`
D, Violini S, Lucania G,
Torrisi MR, Nagase H, Canevari S, Pavan A &
Vittorelli ML (1998) Selective localization of MMP-9,
b1 integrins and HLA-I molecules on membrane
vesicles shed by 8701-BC breast carcinoma cells. Cancer
Res 58, 4468–4474.
7 Abrahams VM, Straszewski SL, Kamsteeg M,
Hanczaruk B, Schwartz PE, Rutherford TJ & Mor G
(2003) Epithelial ovarian cancer cells secrete functional
Fas ligand. Cancer Res 63, 5573–5581.
8 Sidhu SS, Mengistab AT, Tauscher AN, LaVail J &
Basbaum C (2004) The microvesicle as a vehicle for
EMMPRIN in tumor-stromal interactions. Oncogene
23, 956–963.
9 D’Agostino S, Salamone M, Di Liegro I & Vitttorelli ML
(2006) Membrane vesicles shed by oligodendroglioma
cells induce neuronal apoptosis. Int J Oncol 29, 1075–1085.
10 Cooper DNW & Barondes SH (1990) Evidence for
export of a muscle lectin from cytosol to extracellular
matrix and a novel secretory mechanism. J Cell Biol
110, 1681–1691.

11 MacKenzie A, Wilson HL, Kiss-Toth E, Dower SK,
North RA & Surprenant A (2001) Rapid secretion of
interleukin-1B by microvesicle shedding. Immunity 8,
825–835.
12 Bianco F, Pravettoni E, Colombo A, Schenk U, Moller
T, Matteoli M & Verderio C (2005) Astrocyte-derived
ATP induces vesicle shedding and IL-1 beta release
from microglia. J Immunol 174, 7268–7277.
13 Taverna S, Ghersi G, Ginsetra A, Rigogliuso S,
Pecorella S, Alaimo G, Saladino F, Dolo V, Dell’Era P,
Pavan A et al. (2003) Shedding of membrane vesicles
mediates fibroblast growth factor-2 release from cells.
J Biol Chem 278, 81911–81919.
14 Bottcher RT & Niehrs C (2005) Fibroblast growth
factor signaling during early vertebrate development.
Endocr Rev 26, 63–77.
15 Grose R & Dickson C (2005) Fibroblast growth factor
signaling in tumorigenesis. Cytokine Growth Factor Rev
16, 179–186.
16 Landriscina M, Bagala C, Mandinova A, Soldi R,
Micucci I, Bellum S, Prudovsky I & Maciag T (2001)
Copper induces the assembly of a multiprotein aggre-
gate implicated in the release of fibroblast growth factor
1 in response to stress. J Biol Chem 276, 25549–25557.
17 Prudovsky I, Bagala C, Tarantini F, Mandinova A,
Soldi R, Bellum S & Maciag T (2002) The intracellular
translocation of the components of the fibroblast
growth factor 1 release complex precedes their assembly
prior to export. J Cell Biol 158, 201–208.
18 Seghezzi G, Patel S, Ren CJ, Gualandris A, Pintucci G,

Robbins ES, Shapiro RL, Galloway AC, Rifkin DB &
S. Taverna et al. Intracellular trafficking of FGF-2
FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS 1591
Mignatti P (1998) Fibroblast growth factor-2 (FGF-2)
induces vascular endothelial growth factor (VEGF)
expression in the endothelial cells of forming capillaries:
an autocrine mechanism contributing to angiogenesis.
J Cell Biol 141, 1659–1673.
19 Folkman J (1995) Angiogenesis in cancer, vascular,
rheumatoid and other disease. J Nat Med 1, 27–31.
20 Straume O & Akslen LA (2002) Importance of vascular
phenotype by basic fibroblast growth factor, and
influence of the angiogenic factors basic fibroblast
growth factor ⁄ fibroblast growth factor receptor-1 and
ephrin-A1 ⁄ EphA2 on melanoma progression. Am J
Pathol 160, 1009–1019.
21 Chang LK, Garcia-Carden G, Farnebo F, Fannon M,
Chen EJ, Butterfield C, Moses MA, Mulligan RC,
Folkman J & Kaipainen A (2004) Dose-dependent
response of FGF-2 for lymphangiogenesis. PNAS 101,
11658–11663.
22 Bikfalvi A, Klein S, Pintucci G & Rifkin DB (1997)
Biological roles of fibroblast growth factor-2. Endocr
Rev 18, 26–45.
23 Weickert CS, Kittell DA, Saunders RC, Herman MM,
Horlick RA, Kleinman JE & Hyde TM (2005) Basic
fibroblast growth factor and fibroblast growth factor
receptor-1 in the human hippocampal formation.
Neuroscience 131, 219–233.
24 Martin C, Bueno D, Alonso MI, Moro JA, Callejo S,

Parada C, Martin P, Carnicero E & Gato A (2006)
FGF2 plays a key role in embryonic cerebrospinal fluid
trophic properties over chick embryo neuroepithelial
stem cells. Dev Biol 297, 402–416.
25 Dvorak P, Dvorakova D & Hampl A (2006) Fibroblast
growth factor signaling in embryonic cancer steam cells.
FEBS Lett 580, 2869–2874.
26 Sheng Z, Lian Y, Lin CY, Comai L & Chirico WJ
(2006) Direct regulation of rRNA transcription by
fibroblast growth factor 2. Mol Cell Biol 25, 9419–9426.
27 Sorensen V, Nilsen T & Wiedlocha A (2006) Functional
diversity of FGF-2 isoforms by intracellular sorting.
Bioessays 28, 504–514.
28 Sheng Z, Lewis JA & Chirico WJ (2004) Nuclear and
nucleolar localization of 18kDa FGF-2 is controlled by
C-terminal signals. J Biol Chem 279, 40153–40160.
29 Sleeman M, Fraser J, McDonald M, Yuan S, White D,
Kumble K, Grandison JG, Watson JD & Murison JG
(2001) Identification of a new fibroblast growth factor
receptor, FGFR5. Gene 271, 171–182.
30 Mignatti P & Rifkin DB (1991) Release of basic fibro-
blast growth factor, an angiogenic factor devoid of secre-
tory signal sequence: a trivial phenomenon or a novel
secretion mechanism. J Cell Biochem 47, 201–207.
31 Bossard C, Laurell H, Van de Berghe L, Zanibellato C
& Prats H (2003) Translokin is an intracellular media-
tor of FGF-2 trafficking. Nat Cell Biol 5, 433–439.
32 Dolo V, Ginestra A, Ghersi G, Nagase H & Vittorelli
ML (1994) Human breast carcinoma cells cultured in
the presence of serum shed membrane vesicles rich in

gelatinolytic activities. J Submicrosc Cytol Pathol 26,
173–180.
33 Presta M, Moscatelli D, Joseph-Silverstein J & Rifkin
DB (1986) Purification from a human hepatoma cell
line of a basic fibroblast growth factor-like molecule
that stimulates capillary endothelial cell plasminogen
activator production, DNA synthesis, and migration.
Mol Cell Biol 6, 4060–4066.
34 Moscatelli D, Presta M, Joseph-Silverstein J & Rifkin
DB (1986) Both normal and tumor cells produce basic
fibroblast growth factor. J Cell Physiol 129, 273–276.
35 Wiederkehr A, Avaro S, Prescianotto-Baschong C,
Haguenauer-Tsapis R & Riezman H (2000) The F-box
protein Rcy1p is involved in endocytic membrane traffic
and recycling out of an early endosome in Saccharomy-
ces cerevisiae. J Cell Biol 149, 397–410.
36 Fevrier B & Raposo G (2004) Exosomes: endosomal-
derived vesicles shipping extracellular messages. Curr
Opin Cell Biol 16, 415–421.
37 Florkiewicz RZ, Anchin J & Baird A (1998) The inhibi-
tion of FGF-2 export by cardenolides implies a novel
function for the catalytic subunit of Na
+
-K
+
ATPase.
J Biol Chem 273, 544–551.
38 Dahl JP, Binda A, Canfield VA & Levenson R (2000)
Participation of Na,K-ATPase in FGF-2 secretion:
rescue of ouabain-inhibitable FGF-2 secretion by

ouabain-resistant Na,K-ATPase alpha subunits.
Biochemistry 39, 14877–14883.
39 Kone BC & Higham SC (1998) A novel N-terminal
splice variant of the rat H+-K+-ATPase alpha2
subunit. Cloning, functional expression, and renal
adaptive response to chronic hypokalemia. J Biol Chem
273, 2543–2552.
40 Kaplan JH (2005) A moving new role for the sodium
pump in epithelial cells and carcinomas. Science STKE
289, 31.
41 Cassara
`
D, Ginestra A, Dolo V, Miele M, Caruso G,
Lucania G & Vittorelli ML (1998) Modulation of
vesicle shedding in 8701-BC human breast cancer cells.
J Submiscrosc Cytol Pathol 30, 45–53.
42 Grinspan JB, Mueller SN & Levine EM (1983) Bovine
endothelial cells transformed in vitro by benzo(a)pyrene.
J Cell Physiol 114, 328–338.
43 Ghersi G, Dong H, Goldstein LA, Yeh Y, Hakkinen L,
Larjava HS & Chen WT (2002) Regulation of fibroblast
migration on collagenous matrix by a cell surface
peptidase complex. J Biol Chem 277, 29231–29241.
44 Vassalli JD, Dayer JM, Wohlwend A & Belin D (1984)
Concomitant secretion of prourokinase and of a plas-
minogen activator-specific inhibitor by cultured human
monocytes-macrophages. J Exp Med 159, 1653–1668.
Intracellular trafficking of FGF-2 S. Taverna et al.
1592 FEBS Journal 275 (2008) 1579–1592 ª 2008 The Authors Journal compilation ª 2008 FEBS