Am J Physiol Endocrinol Metab
281: E1221–E1229, 2001.
MDCK cells secrete neutral proteases cleaving
insulin-like growth factor-binding protein-2 to -6
LILIANA SHALAMANOVA,1 BERND KÜBLER,1
JENS-GERD SCHARF,2 AND THOMAS BRAULKE1
University of Hamburg, Children’s Hospital, Department of Biochemistry,
D-20246 Hamburg; and Division of Gastroenterology and Endocrinology,
Department of Medicine, University of Göttingen, D-37075 Göttingen, Germany
Received 15 November 2000; accepted in final form 16 July 2001
INSULIN-LIKE GROWTH FACTORS (IGFs) I and II are singlechain polypeptides expressed in many tissues and participate in the regulation of growth and differentiation
of various cell types as autocrine and/or paracrine
factors. Most of the metabolic and mitogenic effects of
IGFs are mediated by IGF-I or insulin receptors exhibiting tyrosine kinase activity upon ligand binding (18).
The capacity of IGFs to affect cell growth and metabolism via interaction with cell surface receptors is
controlled by a family of IGF-binding proteins
(IGFBPs). Six distinct high-affinity IGFBPs, designated IGFBP-1 to IGFBP-6, have been characterized,
differing in molecular mass, posttranslational modifi-
cations, and tissue and developmental regulated expression (10, 44). Additional components of the IGF
system are IGFBP-specific proteases. Limited proteolysis of IGFBPs is believed to be the major mechanism for the release of IGFs from IGFBP 䡠 IGF complexes, generating fragments with reduced affinity
for IGFs (26). Proteolytic activity at neutral pH has
been detected for IGFBP-2 (15), IGFBP-3 (16, 42),
IGFBP-4 (14, 22, 30), and IGFBP-5 (9, 39), which are
characterized by distinct fragmentation patterns,
substrate specificity, and inhibitor profile. Furthermore, several studies suggest that acidic proteases
may be involved in inactivation and the regulation of
the extracellular IGFBP level, presumably in the
lysosomal degradation pathway (2, 3, 8, 12, 27, 36).
Two IGFBP proteases, the IGF-dependent IGFBP-4
protease and the IGFBP-5 serine protease secreted
by cultured fibroblasts, have recently been reported
to be identical with the pregnancy-associated plasma
protein A and the complement component C1s, respectively (6, 23).
IGFs, IGFBPs, and IGF receptors have been reported to contribute to processes of cell proliferation
and differentiation of epithelial cells (11, 32, 46).
Prerequisite for the constitution of an epithelial permeability barrier are highly polarized cells characterized by morphologically, functionally, and biochemically distinct apical and basolateral plasma
membranes. The basolateral localization and secretion of IGF receptors and IGF-II, respectively, as
well as differential sorting of IGFBP-2, -4, and -6 in
polarized HT29-D4 colon carcinoma cells (33), demonstrate the complexity of IGF-mediated epithelial
cell regulation. Because nothing is known about the
synthesis of IGFBP proteases in epithelial cells, the
Madin-Darby canine kidney (MDCK) cell line was
used as a model to examine and characterize secreted IGFBP proteases. Here we report that MDCK
cells secrete different proteases cleaving IGFBP-2 to
-6. Proteases cleaving IGFBP-4 and -6 are delivered
preferentially to the basolateral side.
Address for reprint requests and other correspondence: T.
Braulke, Univ. of Hamburg, Children’s Hospital—Biochemistry,
Martinistr. 52, D-20246 Hamburg, Germany (E-mail: braulke@uke.
uni-hamburg.de).
The costs of publication of this article were defrayed in part by the
payment of page charges. The article must therefore be hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
insulin-like growth factor-binding proteins and proteases;
polarized sorting; disintegrin metalloprotease
http://www.ajpendo.org
0193-1849/01 $5.00 Copyright © 2001 the American Physiological Society
E1221
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
Shalamanova, Liliana, Bernd Kübler, Jens-Gerd
Scharf, and Thomas Braulke. MDCK cells secrete neutral
proteases cleaving insulin-like growth factor-binding protein-2 to -6. Am J Physiol Endocrinol Metab 281: E1221–E1229,
2001.—Proteolysis of insulin-like growth factor-binding proteins (IGFBPs) may be an important mechanism to regulate
IGF availability and IGF-independent functions of IGFBPs.
We analyzed the secretion of IGFBP proteases in MadinDarby canine kidney (MDCK) cells. The results showed that
several specific proteases were secreted, cleaving IGFBP-2 to
-6 at neutral pH. The proteolytic activity against IGFBP-6
differed at least from IGFBP-5 protease activity in its sensitivity both to IGF-II and to the hydroxamic acid-based disintegrin metalloprotease inhibitor, as well as serine protease
inhibitors. During partial purification steps, the serine protease inhibitor-sensitive fraction with IGFBP-6 protease activity was separated from fractions characterized by the
presence of a 30-kDa disintegrin immunoreactive band.
Whereas the IGFBP-4 and -6 proteases are predominantly
secreted across the basolateral membrane, the majority of
IGFBPs are sorted to the apical medium from filter-grown
cells. These studies indicate that the side-specific secretion of
several distinct IGFBP proteases with partially overlapping
IGFBP specificities may be another level in the regulation of
IGF-dependent epithelial functions.
E1222
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
MATERIALS AND METHODS
AJP-Endocrinol Metab • VOL
RESULTS
Secretion of IGFBP proteases. When conditioned media from MDCK cells were tested for IGFBP protease
activity at neutral pH using 125I-labeled rhIGFBP-1 to
-6 as substrates, no fragmentation of IGFBP-1 was
detected within a 6-h incubation period. During the
cell-free incubation, the 125I-IGFBP-2 was cleaved into
two fragments of 24 and 13.5 kDa, and the 30-kDa
nonglycosylated 125I-IGFBP-3 was cleaved into four
fragments of 25, 21, 15, and ⬃8.5 kDa (Fig. 1). 125IIGFBP-4 was fragmented to 17- and 10-kDa products,
and the 125I-IGFBP-5 was cleaved to a 22-kDa doublet,
16- and 8-kDa peptide fragments, in the presence of
conditioned MDCK medium. 125I-IGFBP-6 was almost
completely hydrolyzed, depending on the incubation
time, with a transient appearance of 17-, 10-, and
6-kDa IGFBP-6 fragments (Fig. 1). When the unrelated
62-kDa 125I-ASA was incubated with conditioned
MDCK medium for 14 h at 37°C under conditions
identical to 125I-IGFBP-3, no ASA proteolysis products
were detected (not shown).
Characterization of IGFBP proteases secreted by
MDCK cells. To characterize the IGFBP protease(s)
present in media of MDCK cells in more detail, the
experiments were carried out using 125I-IGFBP-4, -5,
or -6 as representative substrates, as indicated. The pH
optimum of proteolytic activities with 125I-IGFBP-4
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
Materials. Sodium [125I]iodine (carrier free; specific activity 16.9 mCi/␮g iodine), [35S]methionine (100 Ci/mmol), and
prestained protein standard (Rainbow) were purchased from
AmershamPharmaciaBiotech (Freiburg, Germany). Recombinant human (rh)IGF-II from GroPep (Adelaide, Australia)
was iodinated by the chloramine T method (45) to a specific
activity of 60–80 ␮Ci/␮g. Recombinant IGFBP-1 and
IGFBP-2 and -5 were purchased from UBI (Lake Placid, NY)
and GroPep, respectively. Nonglycosylated rhIGFBP-3 was a
generous gift from Drs. A. Sommer and C. Maack (Celtrix,
Santa Clara, CA). rhIGFBP-4 and -6 produced in yeast (20)
were kindly provided by Dr. J. Zapf (University Hospital,
Zurich, Switzerland). Recombinant human arylsulfatase A
(ASA) was a kind gift from Dr. T. Dierks (University of
Göttingen, Germany). The IGFBPs and ASA were iodinated
using IODO-GEN (Pierce Chemical, Rockford, IL) as described (31). The hydroxamic acid-based metalloprotease inhibitor TAPI was prepared at Immunex (Seattle, WA) (29)
and was a kind gift from Dr. S. Rose-John (University of Kiel,
Kiel, Germany). Antibodies directed against the prodomain
(rb 132), the cystein-rich domain (rb 122), and the disintegrin
domain (rb 119) of human ADAM 12 S (25) were kindly
provided by Dr. U. Wewer (University of Copenhagen, Copenhagen, Denmark). Peroxidase-conjugated goat anti-rabbit
IgG came from Dianova (Hamburg, Germany).
Blot analysis. 125I-labeled IGF-II ligand blot analysis was
performed according to Hossenlopp et al. (16), with slight
modifications. Briefly, conditioned media of 0.35 ml were
mixed with 0.82 ml of ice-cold 96% ethanol and kept on ice for
1 h. Precipitated probes were solubilized and separated by
SDS-PAGE (12.5% acrylamide) under nonreducing conditions. After electrotransfer and blocking in 1% fish gelatin,
the nitrocellulose membranes were probed with 125I-IGF-II
and exposed to X-ray films (X-Omat AR, Kodak). For immunoblotting, the membranes were blocked with 5% nonfat
skim milk and probed with the rb 119, 122, and 132 antisera
(dilution 1:100–1:500) and anti-rabbit IgG coupled to horseradish peroxidase (dilution 1:15,000). Reactive bands were
visualized by the SuperSignal enhanced chemiluminescence
detection system (Pierce Chemical) and exposure to X-ray
films.
Cell culture. MDCK cells were maintained in Dulbecco’s
modified Eagle’s medium (DMEM) supplemented with 10%
(vol/vol) fetal calf serum (FCS), 2 mM glutamine, 4.5 g/l
(wt/vol) glucose, and penicillin/streptomycin. The MDCK
cells were routinely grown on 35-mm dishes (Greiner, Germany) or, to obtain polarized cell monolayers, on 24-mm
Transwell (Costar, Cambridge, MA) (0.4-␮m pore size) polycarbonate filters, as described recently (5). Serum-free medium conditioned for 24–72 h in the presence or absence of
IGF-II (50 nM) was obtained as described earlier (4).
Northern blot analysis. The isolation of total RNA from
MDCK cells, radiolabeling by random priming of cDNA
probes for human IGFBPs, and the Northern blot hybridization were carried out as described previously (35).
IGFBP protease assay. Fifty microliters of conditioned media or 5–20 ␮l of column fractions dialyzed against 20 mM
Tris 䡠 HCl, pH 7.4, containing 10 mM NaCl, were incubated
with 125I-labeled IGFBPs (5,000–10,000 cpm) at 37°C for
6–18 h. When indicated, protein inhibitors were included.
After solubilization, the samples were subjected to SDSPAGE (12.5% acrylamide) and visualized by autoradiography, as described previously (24), or by phosphorimaging
(Cyclone, Packard, Meriden, CT).
Protease purification. Three hundred milliliters of conditioned MDCK medium were sequentially precipitated with
30 and 45% ammonium sulfate. The latter precipitate was
dissolved in 3 ml 20 mM Tris buffer, pH 7.4, containing 10
mM NaCl (TBS) and was desalted by Sephadex G25 column
chromatography (PharmaciaBiotech, 1.0 ⫻ 35 cm). Two-milliliter fractions were collected at a flow rate of 0.8 ml/min,
and 50-␮l aliquots were assayed for IGFBP-6 protease activity. Active fractions were pooled and loaded onto a 2-ml
DEAE-Sephadex column equilibrated with TBS. The column
was washed with the same buffer (1 ml/min) until absorbance
(280 nm) returned to the baseline. The proteins were eluted
with a stepwise gradient of 0.1, 0.25, 0.5, 0.75, and 1.0 M
NaCl in 20 mM Tris buffer, pH 7.5. One-milliliter fractions
were collected, dialyzed against TBS, and tested for IGFBP-6
protease activity. The protease-containing fractions were
pooled (fractions 7 and 8 to pool I, fractions 9–12 to pool II,
and fractions 13–15 to pool III), dialyzed against 50 mM Tris
buffer, pH 7.4, and loaded onto a hydroxyapatite column
(PharmaciaBiotech; 1.0 ⫻ 1.0 cm). After the column had been
washed, the proteins were eluted with a two-step gradient of
150 and 300 mM KPi in 50 mM Tris buffer, pH 7.4.
Other methods. Hexosaminidase activity was determined
as described earlier (40). To examine the secretion of total
newly synthesized proteins, cells were labeled with [35S]methionine (80 ␮Ci/ml) for 4.5 h. Aliquots (20%) of the apical or
basolateral media were precipitated with 1 ml of ice-cold
acetone at ⫺20°C for 24 h. After centrifugation, the pellets
were dried and solubilized for SDS-PAGE under reducing
conditions. In parallel, aliquots of media were used to determine the total radioactivity incorporated in secreted proteins
by trichloroacetic acid precipitation. The radiolabeled
polypeptides visualized by autoradiography or phosphorimaging were quantified by densitometric scanning (HewlettPackard Scan Jet 4c/T and the Advance Image Data Analyzer
programme, Raytest, Straubenhardt, Germany).
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
E1223
and -6 as substrates was estimated between 6.5 and
7.4 (not shown). By the addition of 50 nM IGF-II to
the conditioned medium from MDCK cells, neither
the proteolysis of 125I-IGFBP-4 nor that of 125IIGFBP-5 was affected, whereas the protease activity
cleaving 125I-IGFBP-6 was efficiently inhibited in
vitro under these conditions (Fig. 2). Time course
experiments, however, showed that the final degradation of IGFBP-6 was not prevented by IGF-II,
whereas the rate of degradation was prolonged (not
shown). To classify the proteases catalyzing the
cleavage of the IGFBPs according to their inhibitor
profile, media from MDCK cells were incubated with
125
I-IGFBP-5 or -6 in the presence or absence of
the metalloprotease inhibitors 1,10-phenanthroline
(Phe, 10 mM) and TAPI (0.1 mM) and the serine
protease inhibitors 3,4-dichloroisocoumarin (0.2
mM) and aprotinin (Apr, 0.3 ␮M). The proteolysis of
125
I-IGFBP-5 was hardly affected by the inhibitors
tested with the exception of the formation of the
8-kDa IGFBP-5 fragment, which was inhibited by
TAPI and aprotinin (Fig. 3). In contrast, 125IIGFBP-6 proteolytic activity in conditioned medium
was almost completely inhibited by Phe and TAPI
with the exception of the formation of 4–8% of the
17-kDa fragment, whereas the serine protease inhibitors impaired the proteolysis of 125I-IGFBP-6 moderately or weakly (not shown for Apr). These data
suggest that 1) different IGF-sensitive IGFBP proteases are present in the medium of MDCK cells and
Fig. 2. Effect of insulin-like growth factor II (IGF-II) on the proteolysis of IGFBP-4, -5, and -6 in conditioned media
from MDCK cells. 125I-labeled IGFBP-4, -5 and -6 were incubated with 50 ␮l of nonconditioned (Co) or conditioned
media (72 h) for 8 h at 37°C at neutral pH. Aliquots of conditioned media were incubated with 125I-IGFBPs in the
presence (⫹) or absence (⫺) of 50 nM IGF-II, followed by SDS-PAGE and autoradiography. Experiments were
carried out 1–2 times with a single medium of ⱖ3 different MDCK cell cultures.
AJP-Endocrinol Metab • VOL
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 1. Proteolysis of insulin-like growth factor-binding proteins (IGFBPs) by conditioned media from MadinDarby canine kidney (MDCK) cells. 125I-labeled IGFBP-1 to -6 were incubated with 50 ␮l of nonconditioned (1) or
72-h-conditioned media from MDCK cells (2) at 37°C and pH 7.4 for 6 h (IGFBP-1 and -5) or 12 h (IGFBP-2, -3, -4,
and -6). Samples were analyzed by SDS-PAGE and autoradiography. Experiments were carried out 1–2 times with
a single medium of ⱖ3 different MDCK cell cultures.
E1224
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
ered in fractions of pool I eluted at 150 mM KPi (I-1).
The loss of IGFBP-6 protease activity might be due to
the dialysis of the fractions eluted or the separation of
a cofactor required for proteolysis. Strong IGFBP-6
proteolytic activity was found in fractions of pool III
also eluted at 150 mM KPi (III-1), but not in fractions
eluted at 300 mM KPi (III-2) (Fig. 4B). Because the
IGFBP-6 protease activity in conditioned media was
inhibited by the disintegrin metalloprotease-specific
inhibitor TAPI, the fractions eluted from the hydroxy-
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 3. Effects of protease inhibitors on IGFBP-5 and IGFBP-6 protease activity in MDCK medium. Aliquots of 48-h-conditioned MDCK
cell medium (25 ␮l) were incubated with 125I-IGFBP-5 or -6 (5,000–
9,000 counts/min) in the absence of inhibitors (⫺) for 10 h at 37°C.
1,10-Phenanthroline (Phe 10 mM), hydroxamic acid-based metalloprotease inhibitor (TAPI, 0.1 mM), 3,4 dichloroisocoumarin (DCI; 0.2
mM), or aprotinin (Apr; 0.3 ␮M) was added as indicated. Reaction
products were separated by SDS-PAGE and visualized by autoradiography. As a control (Co), 125I-IGFBP-5 or -6 was incubated in
nonconditioned medium under identical conditions. The autoradiogram of 1 representative experiment of 4–6, testing various protease
inhibitors, is shown. The percentage of intact IGFBP-5 or -6 after the
incubation was determined by densitometry and is listed below each
lane.
2) more than one protease with a distinct inhibitor
profile may be involved in the degradation of a specific IGFBP.
For further characterization, the IGFBP-6 protease
was partially purified. A fraction containing protease
activity with 125I-IGFBP-6 as substrate was precipitated from conditioned MDCK medium with ammonium sulfate and was bound to DEAE-Sephacel.
IGFBP-6 protease activity was detected in fractions
eluted with 0.5 M NaCl (fractions 7 and 8) and with 1.0
M NaCl (fractions 13–15) (Fig. 4A). The active fractions were pooled (pools I and III, respectively), as well
as the inactive fractions 9–11 (pool II), and were applied separately to a hydroxyapatite column. No proteolytic activity was found in the flow through, and
only a weak 125I-IGFBP-6 protease activity was recovFig. 4. Partial purification of IGFBP-6 protease. A: ammonium sulfate precipitates of conditioned medium from MDCK cells were
dissolved, dialyzed, and applied on a DEAE-Sephacel column. Bound
proteins were eluted by a stepwise NaCl gradient (top). Dialyzed
aliquots of every second fraction were tested for IGFBP-6 proteolytic
activity, followed by SDS-PAGE and autoradiography (bottom). Underlined fractions were pooled separately (pools I, II, and III). B:
after dialysis, pools I, II, and III were applied separately to an
hydroxyapatite column. Bound proteins were eluted at 150 mM (1) or
300 mM (2) KPi. Dialyzed aliquots of these fractions were incubated
with 125I-IGFBP-6 for 6 h at 37°C and analyzed by SDS-PAGE and
autoradiography. Co, control incubation of 125I-IGFBP-6 with buffer.
AD, aliquots of applied sample on DEAE-Sephacel. C: aliquots of the
same fractions shown in B, eluted from hydroxyapatite column, were
solubilized under nonreducing conditions and analyzed by SDSPAGE and anti-disintegrin immunoblotting with the rb 119 antibody. Elution of proteins from the DEAE- or hydroxyapatite column
in a second purification protocol with linear NaCl or KPi gradients,
respectively, did not result in better separation of IGFBP-6 protease
activities.
AJP-Endocrinol Metab • VOL
281 • DECEMBER 2001 •
www.ajpendo.org
E1225
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
apatite column were tested by immunoblotting with rb
119 antibodies directed against the disintegrin domain
of a disintegrin and metalloprotease (ADAM) 12. Under denaturating conditions, a single immunoreactive
band at 30 kDa was found exclusively in I-1 and I-2 but
not in the pool III-eluted fractions (Fig. 4C). When the
fivefold concentrated I-1 and I-2 fractions were analyzed by SDS-PAGE and silver staining, in addition to
the major BSA band, several weak bands of 88, 105, 40,
and 34 kDa were detected, but no polypeptide of 30
kDa (not shown). In the initial fraction used for purification, a second disintegrin immunoreactive band at
35 kDa was detected. ADAM 12 S is the only soluble
member of this protease family known so far that is
composed of four domains forming an ⬃68-kDa mature
glycosylated protease after cleavage of the 25-kDa
prodomain (25). To examine whether the 30-kDa disintegrin immunoreactive band is a fragment of ADAM
12 S, aliquots of the fraction applied on the DEAEcolumn (AD) and of the fractions eluted with low salt
concentration (I) were immunoblotted with antibodies
against the disintegrin domain (rb 119), the cysteinerich domain (rb 122), and the prodomain (rb 132) of
human ADAM 12 S. The staining pattern of the two
fractions with the three antibodies shown in Fig. 5 is
partially consistent with the expected sizes of fulllength ADAM-12 S (⬃92 kDa) or the mature protease
(⬃65 kDa). All polypeptide bands of smaller molecular
mass indicate the presence of truncated fragments.
Because these fragments were found in fractions that
lost IGFBP-6 protease activity, it is likely that they
present inactive protease polypeptides. Thus the 30kDa immunoreactive band may consist of the disintegrin domain (⬃10 kDa), the cysteine-rich domain (⬃9
kDa), and at least 11 kDa of the metalloprotease domain. Larger molecular mass complexes of ADAM 12 S
with other proteins, as detected with the rb 122 antibody, have been described (25). Because the crossAJP-Endocrinol Metab • VOL
reactivity of the anti-human ADAM 12 S antibodies
with the polypeptides secreted by MDCK cells is unclear, the significance of the data is speculative.
When the IGFBP-6 protease activity was measured
in the III-1 fraction in the presence and absence of
protease inhibitors, the proteolytic activity was inhibited by neither Phe nor by TAPI but was almost completely blocked by the serine protease inhibitors DCI
and Apr (Fig. 6), indicating the separation of the
IGFBP-6 metalloprotease from the serine protease. Of
interest, after incubation of 125I-IGFBP-6 with the
fraction III-1, a prominent fragment of 10 kDa was
formed compared with the continuing proteolytic cleavage in the unfractionated medium.
Polarized sorting of IGFBP-6 protease. To analyze
whether newly synthesized IGFBP proteases are selec-
Fig. 6. Effects of protease inhibitors on partially purified IGFBP-6
protease-containing fraction. Aliquots of the partially purified fraction III-1 eluted from hydroxyapatite column (see Fig. 5) were incubated with 125I-IGFBP-6 for 8 h at 37°C in the presence or absence
(⫺) of Phe (10 mM), TAPI (0.1 mM), DCI (0.2 mM), or Apr (0.3 ␮M).
Reaction products were separated by SDS-PAGE and visualized by
phosphorimaging. Densitometric evaluation revealed the percentage
of remaining intact IGFBP-6 given below each lane. The experiment
was carried out twice with similar results.
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 5. Immunoblot analysis of MDCK
medium fractions with antibodies
against different ADAM 12 S domains.
Aliquots of MDCK medium applied on
the DEAE-ion exchange column (AD)
and of the fraction eluted from the column with 0.35 M NaCl (I) were separated by SDS-PAGE under nonreducing
conditions, transferred to nitrocellulose,
and analyzed with antibodies against
the disintegrin domain (rb 119), the cysteine-rich domain (rb 122), and the
prodomain (rb 132) of ADAM 12 S. Immunoreactive polypeptides were visualized by peroxidase-conjugated secondary
antibodies and enhanced chemiluminescence.
E1226
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
tively directed to the apical or basolateral membrane in
polarized MDCK cells, apical and basolateral media
conditioned for 24 h were collected from cells grown on
filter inserts 4 days after reaching confluence. Incuba-
Fig. 8. Polarized secretion of proteins. A: filter-grown MDCK cells were labeled with [35S]methionine ([35S]Met) for
4.5 h at 37°C. Twenty percent each of the apical (a) and basolateral (b) media, corresponding to 18,000 and 10,000
counts/min TCA-insoluble radioactivity, respectively, were precipitated with acetone, separated by SDS-PAGE
(10%), and visualized by phosphorimaging. This experiment was repeated twice with identical results. *Distinct
proteins secreted into apical medium; 4, proteins preferentially secreted to one side. B: ␤-hexosaminidase activity
was measured in duplicates in apical and basolateral media from 2 cell cultures in parallel. The experiment was
repeated 3 times with MDCK cells after different passages, resulting in variations of the absolute activity values
but with a reproducible ratio of activities in the apical and basolateral media. Enzyme activity in the apical medium
is given as a percentage (means ⫾ SD) of activity in the basolateral medium (5–18 mU/ml and 24 h). C: media from
filter-grown MDCK cells were collected after 24 and 48 h from the apical (lanes 1 and 3) or basolateral side (lanes 2 and
4) and analyzed by SDS-PAGE (12.5% acrylamide), 125I-IGF-II ligand blot, and autoradiography. Positions of molecular
mass marker proteins are indicated. A representative blot of 6 independent experiments is shown.
AJP-Endocrinol Metab • VOL
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 7. Polarized secretion of IGFBP proteases. Medium was collected from the apical (a) or basolateral (b) side of filter-grown MDCK
cells conditioned for 24 h. Aliquots of the media (50 ␮l) were tested
for protease activity against 125I-IGFBP-4 and 125I-IGFBP-6, followed by SDS-PAGE and autoradiography. As a control (Co), 125IIGFBP-4 or -6 was incubated in nonconditioned medium under
identical conditions. Autoradiograms of 1 representative experiment
of 2 (IGFBP-4) or 4 (IGFBP-6) are shown. Aliquots contain 12.2 (a)
and 4.9 mU/ml (b) ␤-hexosaminidase activity secreted within 24 h.
tion of 125I-IGFBP-4 with apical medium revealed that
64% of the IGFBP-4 remained intact compared with
only 28% of the 125I-IGFBP-4 after incubation in the
basolateral medium (Fig. 7). Similarly, the majority of
IGFBP-6 proteolytic activity was found in the same
basolateral medium (Fig. 7), whereas 54% of the
IGFBP-6 remained intact in the apical medium. These
data indicate that proteases cleaving both IGFBP-4
and -6 are secreted preferentially to the basolateral
side. For comparison, filter-grown MDCK cells labeled
with [35S]methionine for 4.5 h secrete more newly
synthesized proteins to the apical than to the basolateral medium (⬃1.8-fold) as measured by TCA-insoluble
radioactivity. The composition and the ratio of labeled
proteins secreted into the apical and basolateral media
were different and polypeptide dependent (Fig. 8A).
Thus, when ␤-hexosaminidase activity was determined, a 2.6-fold (range 2.3–3.1, n ⫽ 6) higher activity
was measured in the apical than in the basolateral
medium (Fig. 8B). In addition, conditioned media from
the apical and basolateral sides were analyzed by ligand blotting with 125I-labeled IGF-II. In both media,
one band with an estimated molecular mass of 28 kDa
and a prominent band at 25 kDa were detected, showing an approximate 4.2-fold higher abundance in the
apical (range 3.9–4.7; estimated by densitometry) than
in the basolateral medium (Fig. 8C). These data indicate that the presence of proteases cleaving IGFBP-4
and -6 as substrate in the basolateral medium is spe-
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
cific and not due to loading differences. The identity of
the 28- and 25-kDa IGFBPs could not be determined,
because none of the tested antibodies raised against
various human or rodent IGFBPs cross-reacted with
canine IGFBPs in immunoblots, and no hybridization
of RNA from MDCK cells with any of the cDNAs probes
specific for human IGFBP-1 to -6 was observed (not
shown).
DISCUSSION
AJP-Endocrinol Metab • VOL
rather unlikely that these acid-activated proteases
play a role in regulation of the extracellular IGFBP-6
level. Data from mouse fibroblasts partially deficient
for several of the lysosomal enzymes indicate that the
acid-activated proteases may play a role in degradation
of endocytosed IGFBP-3 (3).
The data presented here demonstrate that at least
the proteases cleaving IGFBP-4 and -6 are secreted
preferentially to the basolateral medium of polarized
MDCK cells. In contrast, the activity of the lysosomal
␤-hexosaminidase and the abundance of IGFBPs in the
apical medium are ⬃2.6 and 4 times higher, respectively, than in the basolateral medium. It has been
reported that, in the human colon adenocarcinoma cell
line Caco-2, a majority of mannose 6-phosphate-containing lysosomal enzymes accumulate in the apical
medium (41). This polarized distribution results from
selective receptor-mediated uptake of lysosomal enzymes from the basolateral surface, followed by transcytotic delivery to the apical side (41). It is still unclear whether the low IGFBP level in basolateral
medium results from increased proteolysis. The addition of serine protease inhibitors or TAPI increases, if
anything, the abundance of IGFBPs in the apical
rather than in the basolateral medium (unpublished
results); however, the stability of the TAPI in hydrous
solution at 37°C is low, and long-term toxic effects on
cultured cells cannot be excluded (S. Rose-John, personal communication). Recently, Remacle-Bonnet et al.
(33) reported on preferential sorting of IGFBP-2 and -4
to the basolateral side, whereas IGFBP-6 is primarily
delivered to the apical surface of polarized enterocytelike HT29-D4 human colonic carcinoma cells. However,
the IGFBP protease activities are not determined in
the media of HT29-D4 cells, and the molecular mechanism underlying the distinct pattern in IGFBP abundance in media from polarized cells is not known. It is
believed that the transport of newly synthesized proteins to the basolateral side represents a default pathway for exocytosis, whereas the sorting to the apical
side is signal dependent, occurring in the trans-Golgi
network of MDCK cells (28, 34). Sorting signals for
directing proteins in polarized cells to the apical or
basolateral surface are best characterized for membrane proteins. For MDCK cells, it has been well documented that glycosylphosphatidylinositol-anchored
proteins are specifically targeted to the apical plasma
membrane (24). On the other hand, tyrosine and dileucine-containing signals have been identified in the
cytoplasmic domains of membrane proteins, which mediate sorting to the basolateral cell surface (19) and
might be followed by a subsequent transcytotic delivery to the apical membrane. Furthermore, both Nlinked and O-linked glycosylations of proteins, as well
as linear glycosaminoglycan chains, appear to contain
apical sorting information (21, 37, 43) recognized by
molecules with lectin activity (13). However, there is
evidence that the sorting mechanism for individual
signals may vary considerably among different epithelial cell types (19). This might explain the discrepancies between IGFBP expression in the media of polar-
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
The present study shows that MDCK cells secrete
neutral proteases that cleave IGFBPs either to fragments of defined sizes, like IGFBP-2, -4, and -5, or to
small peptides like IGFBP-6, hardly detectable under
the conditions used. Inhibitor studies and data from
the partial purification of the proteases using IGFBP-5
and -6 as substrates indicate that at least two classes of
proteases, metallo- and serine proteases, are secreted
by MDCK cells with different substrate specificities.
The IGFBP proteases in the conditioned medium separated by ion exchange chromatography into two fractions are inhibited either by serine protease inhibitors
or by metalloprotease inhibitors such as 1,10-phenanthroline and the hydroxamic acid-based inhibitor
TAPI. The latter has been shown to be a potential
inhibitor of tumor necrosis factor-␣-converting enzyme
(TACE), a member of the disintegrin metalloproteases
(1, 29). The presence of several immunoreactive
polypeptides, including a 30-kDa protein band in
TAPI-sensitive IGFBP-6 protease fractions detected
with specific antibodies directed against the disintegrin and cysteine-rich domain of ADAM 12 S, suggests
the involvement of disintegrin metalloproteases in
IGFBP-6 proteolysis. These metalloproteases have
been reported to contribute to the proteolysis of
IGFBP-3, -4, and -5 in human pregnancy serum and in
cleavage of IGFBP-3 by placental trophoblasts (17, 22).
Recently, the direct interaction with IGFBP-3 and the
proteolysis of IGFBP-3 by recombinant ADAM 12 S
have been shown (38). Whether disintegrin metalloproteases are involved in the proteolysis of IGFBPs in
MDCK cells remains to be demonstrated and requires
the sequencing of the 30-kDa immunoreactive polypeptide and the analysis of the ADAM expression pattern.
The observed protective effect of IGF-II on the proteolysis of IGFBP-6 may be due to the preferential binding of IGF-II to IGFBP-6 compared with IGFBP-4 or -5,
resulting in conformational changes that prolong the
degradation rate rather than a direct effect on the
protease.
Whereas the identity of proteases cleaving IGFBP-6
is still unknown and remains to be determined,
IGFBP-6 proteolytic activity was also detected in acidified conditioned media from NIH-3T3 and HaCaT human keratinocytes (7, 27). However, the pH optimum,
the complete degradation, and the inhibitor profile
suggest that different acid-activated proteases, including the lysosomal protease cathepsin D, might be involved in IGFBP-6 proteolysis, either directly or indirectly by activation of IGFBP-6 proteases. Thus, it is
E1227
E1228
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
This work was supported by Deutsche Forschungsgemeinschaft
SFB 402 A6 (B. Kübler and T. Braulke), SFB 402 A5 (J.-G. Scharf),
and Graduiertenkolleg 336 (L. Shalamanova).
REFERENCES
1. Black RA, Rauch CT, Kozlosky CJ, Peschon JJ, Slack JL,
Wolfson MF, Castner BJ, Stocking KL, Reddy P, Srinivasan S, Nelson N, Boiani N, Schooley KA, Gerhart M, Davis
R, Fitzner NJ, Johnson RS, Paxton RJ, March CJ, and
Cerretti DP. A metalloproteinase disintegrin that releases tumour-necrosis factor-alpha from cells. Nature 385: 729–733,
1997.
2. Braulke T, Claussen M, Saftig P, Wendland M, Neifer K,
Schmidt B, Zapf J, von Figura K, and Peters C. Proteolysis
of IGFBPs by cathepsin D in vitro and in cathepsin D-deficient
mice. Prog Growth Factor Res 6: 265–271, 1995.
3. Braulke T, Dittmer F, Götz W, and von Figura K. Alteration
in pancreatic immunoreactivity of insulin-like growth factor
(IGF)-binding protein (IGFBP)-6 and in intracellular degradation of IGFBP-3 in fibroblasts of IGF-II receptor/IGF-II-deficient
mice. Horm Metab Res 31: 235–241, 1999.
4. Braulke T, Götz W, and Claussen M. Immunohistochemical
localization of insulin-like growth factor binding protein-1, -3
and -4 in human fetal tissues and their analysis in media from
fetal tissue explants. Growth Regul 6: 55–65, 1996.
5. Breuer P, Körner C, Böker C, Herzog A, Pohlmann R, and
Braulke T. Serine phosphorylation site of the 46-kDa mannose
6-phosphate receptor is required for transport to the plasma
membrane in Madin-Darby canine kidney and mouse fibroblast
cells. Mol Biol Cell 8: 567–576, 1997.
6. Busby WH, Nam TJ, Moralez A, Smith C, Jennings M, and
Clemmons DR. The complement component C1s is the protease
that accounts for cleavage of insulin-like growth factor-binding
protein-5 in fibroblast medium. J Biol Chem 275: 37638–37644,
2000.
7. Claussen M, Buergisser D, Schuller AG, Matzner U, and
Braulke T. Regulation of insulin-like growth factor (IGF)-binding protein-6 and mannose 6-phosphate/IGF-II receptor expression in IGF-IL-overexpressing NIH 3T3 cells. Mol Endocrinol 9:
902–912, 1995.
8. Claussen M, Kübler B, Wendland M, Neifer K, Schmidt B,
Zapf J, and Braulke T. Proteolysis of insulin-like growth
factors (IGF) and IGF binding proteins by cathepsin D. Endocrinology 138: 3797–3803, 1997.
AJP-Endocrinol Metab • VOL
9. Claussen M, Zapf J, and Braulke T. Proteolysis of insulin-like
growth factor binding protein-5 by pregnancy serum and amniotic fluid. Endocrinology 134: 1964–1966, 1994.
10. Clemmons DR. Role of insulin-like growth factor binding proteins in controlling IGF actions. Mol Cell Endocrinol 140: 19–24,
1998.
11. Cohick WS and Clemmons DR. Regulation of insulin-like
growth factor binding protein synthesis and secretion in a bovine
epithelial cell line. Endocrinology 129: 1347–1354, 1991.
12. Conover CA and De Leon DD. Acid-activated insulin-like
growth factor-binding protein-3 proteolysis in normal and transformed cells. Role of cathepsin D. J Biol Chem 269: 7076–7080,
1994.
13. Fiedler K and Simons K. Characterization of VIP36, an animal lectin homologous to leguminous lectins. J Cell Sci 109:
271–276, 1996.
14. Fowlkes J and Freemark M. Evidence for a novel insulin-like
growth factor (IGF)-dependent protease regulating IGF-binding
protein-4 in dermal fibroblasts. Endocrinology 131: 2071–2076,
1992.
15. Giudice LC, Farrell EM, Pham H, and Rosenfeld RG. Identification of insulin-like growth factor-binding protein-3
(IGFBP-3) and IGFBP-2 in human follicular fluid. J Clin Endocrinol Metab 71: 1330–1338, 1990.
16. Hossenlopp P, Segovia B, Lassarre C, Roghani M, Bredon
M, and Binoux M. Evidence of enzymatic degradation of insulin-like growth factor-binding proteins in the 150K complex
during pregnancy. J Clin Endocrinol Metab 71: 797–805, 1990.
17. Irwin JC, Suen LF, Cheng BH, Martin R, Cannon P, Deal
CL, and Giudice LC. Human placental trophoblasts secrete a
disintegrin metalloproteinase very similar to the insulin-like
growth factor binding protein-3 protease in human pregnancy
serum. Endocrinology 141: 666–674, 2000.
18. Jones JI and Clemmons DR. Insulin-like growth factors and
their binding proteins: biological actions. Endocr Rev 16: 3–34,
1995.
19. Keller P and Simons K. Post-Golgi biosynthetic trafficking.
J Cell Sci 110: 3001–3009, 1997.
20. Kiefer MC, Schmid C, Waldvogel M, Schlapfer I, Futo E,
Masciarz FR, Green K, Barr PJ, and Zapf J. Characterization of recombinant human insulin-like growth factor binding
proteins 4, 5, and 6 produced in yeast. J Biol Chem 267: 12692–
12699, 1992.
21. Kolset SO, Vuong TT, and Prydz K. Apical secretion of chondroitin sulphate in polarized Madin-Darby canine kidney
(MDCK) cells. J Cell Sci 112: 1797–1801, 1999.
22. Kübler B, Cowell S, Zapf J, and Braulke T. Proteolysis of
insulin-like growth factor binding proteins by a novel 50-kilodalton metalloproteinase in human pregnancy serum. Endocrinology 139: 1556–1563, 1998.
23. Lawrence JB, Oxvig C, Overgaard MT, Sottrup-Jensen L,
Gleich GJ, Hays LG, Yates JR III, and Conover CA. The
insulin-like growth factor (IGF)-dependent IGF binding protein-4 protease secreted by human fibroblasts is pregnancyassociated plasma protein-A. Proc Natl Acad Sci USA 96: 3149–
3153, 1999.
24. Lisanti MP, Sargiacomo M, Graeve L, Saltiel AR, and
Rodriguez-Boulan E. Polarized apical distribution of glycosylphosphatidylinositol-anchored proteins in a renal epithelial cell
line. Proc Natl Acad Sci USA 85: 9557–9561, 1988.
25. Loechel F, Gilpin BJ, Engvall E, Albrechtsen R, and
Wewer UM. Human ADAM 12 (meltrin alpha) is an active
metalloprotease. J Biol Chem 273: 16993–166997, 1998.
26. Maile LA and Holly JM. Insulin-like growth factor binding
protein (IGFBP) proteolysis: occurrence, identification, role and
regulation. Growth Horm IGF Res 9: 85–95, 1999.
27. Marinaro JA, Hendrich EC, Leeding KS, and Bach LA.
HaCaT human keratinocytes express IGF-II, IGFBP-6, and an
acid-activated protease with activity against IGFBP-6. Am J
Physiol Endocrinol Metab 276: E536–E542, 1999.
28. Matter K and Mellmann I. Mechanisms of cell polarity: sorting
and transport in epithelial cell. Curr Opin Cell Biol 6: 545–554,
1994.
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
ized MDCK cells reported in this study and expression
in intestinal HT29-D4 cell media (33).
We propose that a majority of proteases cleaving
IGFBP-4 and -6 in epithelial MDCK cells are delivered
to the basolateral side and might be involved in release
of IGFs by proteolysis of exogenous IGFBP 䡠 IGF complexes derived from the cellular environment, e.g.,
mesenchymal cells. Basolateral IGFBP proteolysis is a
prerequisite to guarantee the access of IGFs to cell
surface IGF receptors. A small percentage of IGFBP
proteases might be delivered to the apical side via an
indirect pathway, which might control the steady-state
polarized distribution of IGFBPs. In studies with epithelial MDBK and Caco-2 cell lines, differences in the
synthesis and secretion of distinct IGFBPs and in their
mode of regulation have been reported, which might be
necessary for the maintenance of the proliferative state
and/or the initiation differentiation of these cells (11,
46). Further studies are required to identify the molecular mechanism of selective IGFBP sorting and to
evaluate the physiological significance in polarized
MDCK cells.
IGFBPS AND IGFBP-6 PROTEASE IN MDCK CELLS
AJP-Endocrinol Metab • VOL
37. Scheiffele P, Peranen J, and Simons K. N-glycans as apical
sorting signals in epithelial cells. Nature 378: 96–98, 1995.
38. Shi Z, Xu W, Loechel F, Wewer UM, and Murphy LJ. ADAM
12, a disintegrin metalloprotease, interacts with insulin-like
growth factor-binding protein-3. J Biol Chem 275: 18574–18580,
2000.
39. Thrailkill KM, Quarles LD, Nagase H, Suzuki K, Serra DM,
and Fowlkes JL. Characterization of insulin-like growth factor-binding protein 5-degrading proteases produced throughout
murine osteoblast differentiation. Endocrinology 136: 3527–
3533, 1995.
40. Von Figura K. Secretion of beta-hexosaminidase by cultured
human skin fibroblasts. Kinetics, effect of temperature, cell
density, serum concentration and pH. Exp Cell Res 111: 15–21,
1978.
41. Wick DA, Seetharam B, and Dahms NM. Biosynthesis and
secretion of the mannose 6-phosphate receptor and its ligands in
polarized Caco-2 cells. Am J Physiol Gastrointest Liver Physiol
277: G506–G514, 1999.
42. Xu S, Savage P, Burton JL, Sansom J, and Holly JM.
Proteolysis of insulin-like growth factor-binding protein-3 by
human skin keratinocytes in culture compared with that in skin
interstitial fluid: the role and regulation of components of the
plasmin system. J Clin Endocrinol Metab 82: 1863–1868, 1997.
43. Yeaman C, Le Gall AH, Baldwin AN, Monlauzeur L, Le
Bivic A, and Rodriguez-Boulan E. The O-glycosylated stalk
domain is required for apical sorting of neurotrophin receptors in
polarized MDCK cells. J Cell Biol 139: 929–940, 1997.
44. Zapf J. Physiological role of the insulin-like growth factor binding proteins. Eur J Endocrinol 132: 645–654, 1995.
45. Zapf J, Walter H, and Froesch ER. Radioimmunological determination of insulin-like growth factor I and II in normal
subjects and in patients with growth disorders and extrapancreatic tumor hypoglycemia. J Clin Invest 68: 1321–1330, 1981.
46. Zhang Y, Wick DA, Seetharam B, and Dahms NM. Expression of IGF-II and IGF binding proteins in differentiating human
intestinal Caco-2 cells. Am J Physiol Endocrinol Metab 269:
E804–E813, 1995.
281 • DECEMBER 2001 •
www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.5 on June 18, 2017
29. Mohler KM, Sleath PR, Fitzner JN, Cerretti DP, Alderson
M, Kerwar SS, Torrance DS, Otten-Evans C, Greenstreet
T, Weerawarna K, Kronheim SR, Petersen M, Gerhart M,
Kozlosky CJ, March CJ, and Black RA. Protection against a
lethal dose of endotoxin by an inhibitor of tumour necrosis factor
processing. Nature 370: 218–220, 1994.
30. Parker A, Gockerman A, Busby WH, and Clemmons DR.
Properties of an insulin-like growth factor-binding protein-4
protease that is secreted by smooth muscle cells. Endocrinology
136: 2470–2476, 1995.
31. Parker KC and Strominger JC. Localization of the sites of
iodination of human ␣2-microglobulin: quaternary structure implications for histocompatibility antigens. Biochemistry 22:
1145–1153, 1983.
32. Pommier GJ, Garrouste FL, El Atiq F, Roccabianca M,
Marvaldi JL, and Remacle-Bonnet MM. Potential autocrine
role of insulin-like growth factor II during suramin-induced
differentiation of HT29-D4 human colonic adenocarcinoma cell
line. Cancer Res 52: 3182–3188, 1992.
33. Remacle-Bonnet M, Garrouste F, El Atiq F, Marvaldi J,
and Pommier G. Cell polarity of the insulin-like growth factor
system in human intestinal epithelial cells. Unique apical sorting of insulin-like growth factor binding protein-6 in differentiated human colon cancer cells. J Clin Invest 96: 192–200, 1995.
34. Rindler MJ and Traber MG. A specific sorting signal is not
required for the polarized secretion of newly synthesized proteins from cultured intestinal epithelial cells. J Cell Biol 107:
471–479, 1988.
35. Scharf JG, Schmidt-Sandte W, Pahernik SA, Koebe HG,
and Hartmann H. Synthesis of insulin-like growth factor binding proteins and of the acid-labile subunit of the insulin-like
growth factor ternary binding protein complex in primary cultures of human hepatocytes. J Hepatol 23: 424–430, 1995.
36. Scharf JG, Schmidt-Sandte W, Pahernik SA, Ramadori G,
Braulke T, and Hartmann H. Characterization of the insulinlike growth factor axis in a human hepatoma cell line (PLC).
Carcinogenesis 19: 2121–2128, 1998.
E1229