(Received for publication, December 24, 1996, and in revised form, February 19, 1997)
From the Departments of Thoracic/Head and Neck
Medical Oncology and ¶ Thoracic and Cardiovascular Surgery,
University of Texas M. D. Anderson Cancer Center, Houston, Texas 77030 and the § Department of Pediatrics, University of
Pennsylvania, Philadelphia, Pennsylvania 19104
Retinoids, including retinol and retinoic acid
derivatives, maintain the normal growth and differentiation of human
bronchial epithelial cells. The signaling pathways through which
retinoids mediate these effects have not been defined. Insulin-like
growth factor binding protein-3 (IGFBP-3) and the transforming growth factor- (TGF-
) gene family (
1-3) were examined as potential components of the retinoid signaling pathway in normal human bronchial epithelial cells. All-trans-retinoic acid (t-RA) increased
the levels of TGF-
2 and IGFBP-3 mRNA and of secreted TGF-
and
IGFBP-3 proteins. An antagonist of retinoic acid receptor-
,
LG100629, abrogated the increase in TGF-
2 and IGFBP-3 mRNA
levels induced by t-RA. t-RA increased IGFBP-3 mRNA levels
transiently from 1 to 6 h, and subsequently a sustained increase
began at 72 h, which coincided with the appearance of active
TGF-
in the media. Treatment with TGF-
2 increased IGFBP-3
mRNA levels, but treatment with latency-associated peptide, which
inactivates secreted TGF-
, did not abrogate the effect of t-RA on
IGFBP-3 expression. These findings provide evidence that t-RA increased
TGF-
2 and IGFBP-3 expression through an retinoic acid
receptor-
-dependent pathway, and the increase in IGFBP-3
expression by t-RA did not require activation of the TGF-
pathway by
autocrine or paracrine mechanisms.
Retinoids control normal tracheobronchial epithelial growth and differentiation. Rodents that are deprived of vitamin A develop squamous metaplasia in the tracheobronchial epithelium, and normal epithelial differentiation is restored by vitamin A supplementation (1, 2). In tissue culture, human bronchial epithelial (HBE)1 cells undergo squamous differentiation with a variety of agents, and all-trans-retinoic acid (t-RA) inhibits this process (3-7). HBE cells treated with t-RA develop mucociliary features in collagen gels (3, 8). Grown in monolayer cultures, retinol-treated HBE cells undergo growth arrest with no evidence of morphologic differentiation (9).
Retinoids are ligands for the retinoic acid receptors (RAR-, -
,
and -
) and retinoid X receptors (RXR-
, -
, and -
), which form RAR-RXR heterodimers and RXR homodimers and are transcriptionally activated by ligand binding (reviewed in Ref. 10). In bronchial epithelial cells, RAR-
is expressed at high levels and has been shown to activate growth inhibitory pathways (11, 12). The signaling
pathways activated by RAR-
that mediate growth inhibition in normal
HBE cells have not been defined. Retinoids increase the expression of
transforming growth factor-
(TGF-
) family members (13-22).
TGF-
is secreted as a latent complex and converted to an active
form, and it signals through a heteromeric complex of the type I and
type II receptors (23). Activation of the TGF-
pathway has been
implicated in the effects of retinoids on cellular growth and
differentiation (13-22). In addition, the secretion of insulin-like
growth factor binding protein-3 (IGFBP-3) is enhanced by retinoid
treatment (24-33, 35). IGFBP-3 is one of a family of seven IGFBPs
(36-39). IGFBP-3 has been proposed to inhibit cell growth by reducing
IGF bioavailability, by altering the responsiveness of the IGF-I
receptor to IGF-I, and by mechanisms independent of the IGF-I receptor
(40, 41). IGFBP-3 expression is also increased by TGF-
2 treatment in
breast cancer cells and has been implicated in the growth inhibitory
effects of TGF-
2 (42). These findings support the notion that
TGF-
and IGFBP-3 actions are connected through a common retinoid
signal transduction pathway.
In this study, we examined the regulation of IGFBP-3 and TGF- gene
family expression by t-RA in normal HBE cells. We demonstrated that
t-RA increased the levels of TGF-
2 and IGFBP-3 mRNA and of
secreted TGF-
and IGFBP-3 proteins. These events were inhibited by a
retinoid that functions as an RAR-
antagonist. Treatment with
TGF-
2 increased IGFBP-3 mRNA levels, demonstrating a linkage of
IGFBP-3 with TGF-
2 signaling pathways. However, the addition of
latency-associated peptide (LAP), which inactivates secreted TGF-
,
did not abrogate the effect of t-RA on IGFBP-3 expression. These
findings provide evidence that t-RA increased TGF-
2 and IGFBP-3
expression through an RAR-
-dependent pathway, and the increase in IGFBP-3 expression by t-RA did not require activation of
the TGF-
pathway by autocrine or paracrine mechanisms.
Normal HBE cells were cultured
from bronchial mucosal biopsy samples taken from fresh surgical
specimens as monolayer cultures on standard plasticware in keratinocyte
serum-free medium (Life Technologies, Inc.) containing bovine pituitary
extract (BPE) and epidermal growth factor as described previously (11).
Mink lung epithelial cells (MLECs) stably transfected with a luciferase reporter plasmid containing a truncated plasminogen activator inhibitor
type I promoter (16) were a gift from Drs. Irene Nunes and Daniel
Rifkin (Department of Cell Biology and Kaplan Cancer Center, New York
University Medical Center, New York, NY). MLECs were cultured in
Dulbecco's modified Eagle's medium (Life Technologies, Inc.)
containing 10% fetal calf serum as described previously (16). t-RA was
purchased from Sigma. The RAR- antagonist LG100629 (43) was obtained
from Ligand Pharmaceuticals, Inc. (La Jolla, CA), and LAP was provided
by R & D Systems, Inc. (Minneapolis, MN). Recombinant TGF-
2 was
purchased from Genzyme, Inc. (Cambridge, MA). Actinomycin D and
cycloheximide were purchased from Sigma.
Total cellular RNA was prepared from
normal HBE cells, electrophoresed (20 µg/lane) on a 1% agarose gel
containing 2% formaldehyde, transferred to a nylon membrane
(Zetaprobe, Bio-Rad), hybridized to an
[-32P]dCTP-labeled cDNA probe, washed, and
autoradiographed as described previously (11). cDNAs for TGF-
1,
-
2, and -
3 (44-46) were obtained from Dr. Rik Derynck
(University of California, San Francisco, CA), and the IGFBP-3 cDNA
(47) was obtained from Dr. William Wood (Genentech, Inc., San
Francisco, CA).
Normal
HBE cells were seeded on 10-cm plates (105 cells/plate),
treated with t-RA for different time periods or with medium alone, and
conditioned medium samples were collected simultaneously 144 h
after cell seeding. For the medium sample that represents the 0-24-h
time point, treatment with 106 M t-RA in
BPE-free medium (to eliminate exogenous TGF-
) was begun 120 h
after cell seeding, and an aliquot was collected at 144 h. For the
sample that represents 24-48 h, treatment was begun 96 h after
cell seeding, replaced at 120 h with BPE-free medium containing
t-RA, and an aliquot was collected at 144 h. For the sample that
represents 48-72 h, treatment was begun 72 h after cell seeding
and replaced at 120 h with BPE-free medium containing t-RA, and an
aliquot was collected at 144 h. For the sample that represents
72-96 h, treatment was begun 48 h after cell seeding and replaced
at 120 h with BPE-free medium containing t-RA, and an aliquot was
collected at 144 h. For the sample that represents 96-120 h,
treatment was begun 24 h after cell seeding and replaced at
120 h with BPE-free medium containing t-RA, and an aliquot was
collected at 144 h. For the control (untreated) medium sample, no
t-RA was added, the medium was changed to BPE-free medium at 120 h, and an aliquot was collected at 144 h. Conditioned medium samples were transferred to three separate wells on a six-well plate
seeded with MLECs that were stably transfected with the TGF-
-responsive luciferase vector (105 cells/well).
MLECs were treated for 16 h, and then cytosolic luciferase
activity was measured as described previously (11). Luciferase values
reflect TGF-
bioactivity in the conditioned medium. To examine
bioactivity that reflects total (active plus latent) TGF-
levels,
MLECs were treated with conditioned medium previously heated to
80 °C for 5 min, which activates latent TGF-
(21). The results
represent the means and standard deviations of luciferase activities
from three identical wells, and luciferase activities were corrected
for normal HBE cell numbers determined at the end of t-RA treatment.
The presence of t-RA in the conditioned medium did not appreciably
alter luciferase activity in MLECs (data not shown).
Conditioned medium samples were collected from
t-RA-treated normal HBE cells as described for the TGF- bioactivity
assay. Aliquots determined on the basis of equal cell number were
concentrated 1:10 with Centriprep 10 filters (Amicon, Inc., Beverly,
MA), loaded on a 1% SDS, 12% polyacrylamide gel, blotted to a
nitrocellulose membrane (BAS-83, Schleicher & Schuell, Inc.,
Burlingame, VT), incubated for 1 h at room temperature with an
anti-IGFBP-3 affinity purified monoclonal antibody (Diagnostic Systems
Laboratories, Webster, TX), and detected with the enzyme-linked
chemiluminesence assay (ECL kit, Amersham Corp.). For WLB, the membrane
was incubated with a 2 × 106 cpm mixture of
125I-labeled IGF-I and IGF-II and exposed to film as
described previously (48).
Normal HBE cells were seeded
at a density of 105 cells/10-cm plate and treated for
120 h with media alone, 106 M t-RA
alone, and 10
6 M t-RA in combination with LAP
at concentrations of 1, 10, 100, 200, and 500 ng/ml. At 120 h, the
cells were trypsinized and stained with trypan blue, and the total
viable cell number was counted by using an hemocytometer.
We examined the regulation of IGFBP-3 and TGF-1, -
2, and
-
3 mRNA levels in normal HBE cells during 10
6
M t-RA treatment by Northern analysis (Fig.
1). An increase in TGF-
2 mRNA levels was detected
between 6 and 12 h; this increase continued through 120 h.
TGF-
1 was expressed constitutively with no change during treatment.
TGF-
3 mRNA was not detected (data not shown). IGFBP-3 mRNA
was expressed in a bimodal pattern. A transient increase was observed
at 6 h, and a sustained increase appeared at 72 h of
treatment. Examination of earlier time points revealed that increased
IGFBP-3 mRNA was detectable at 1 h of 10
6
M t-RA treatment (Fig. 2A). The
increase in IGFBP-3 mRNA at 6 h was inhibited by treatment
with actinomycin D or cycloheximide (Fig. 2B), demonstrating
the necessity of both gene transcription and protein synthesis for
activation of this retinoid signaling event.
We investigated the contribution of RAR--dependent
signaling pathways to the increased TGF-
2 and IGFBP-3 mRNA
levels induced by t-RA. Normal HBE cells were treated for 120 h
with 10
8 M t-RA and different doses of
LG100629 (43), which is a synthetic retinoid that functions as an
RAR-
antagonist, and total cellular RNA was prepared for Northern
analysis. In combination with 10
6 M t-RA,
LG100629, which must be in molar excess to inhibit t-RA-induced activation of RAR-
, was toxic to normal HBE cells (data not shown). LG100629 abrogated the increase in TGF-
2 and IGFBP-3 expression induced by 10
8 M t-RA (Fig.
3). TGF-
1 mRNA levels, which did not change with t-RA treatment, were not altered by LG100629. LG100629 also inhibited the increase in IGFBP-3 expression observed at 6 h of t-RA
treatment (data not shown).
The effect of t-RA on TGF- bioactivity in conditioned medium was
investigated. Normal HBE cells were treated with 10
6
M t-RA, and conditioned medium samples that reflect defined
24-h periods (0-24, 24-48, 48-72 h, etc.) of t-RA treatment were
collected as described under "Experimental Procedures." MLECs that
are stably transfected with a reporter plasmid containing a truncated
plasminogen activator inhibitor type I promoter (16) were treated with
conditioned medium samples for 16 h and subjected to luciferase
assays to determine relative levels of TGF-
bioactivity in the
conditioned media. Total TGF-
(active plus latent) bioactivity was
measured by heating the conditioned medium samples to convert latent
TGF-
into its active form. Luciferase activities increased during
t-RA treatment (Fig. 4), demonstrating increases in
total and active TGF-
. The increase in active TGF-
occurred
between 48 and 72 h of treatment.
The effect of t-RA on IGFBP-3 protein secretion was examined in
conditioned medium samples collected at 24-h intervals (as described
for the MLEC luciferase assay) by performing immunoblot and WLB (Fig.
5). IGFBP-3 was detected in the media of untreated cells
(t = 0) by WLB, and increased IGFBP-3 levels, first
detected by immunoblot at 24 h, occurred with t-RA treatment.
Because the increase in active TGF- protein in the media coincided
with the increase in IGFBP-3 mRNA levels at 72 h, we examined whether IGFBP-3 expression is responsive to activation of TGF-
2 signaling pathways. Normal HBE cells were treated with 5 ng/ml TGF-
2, which has been shown to increase IGFBP-3 expression in human
breast cancer cells (42), and Northern analysis was performed at
72 h, revealing increased IGFBP-3 mRNA (Fig.
6). The contribution of TGF-
signaling pathways to
the t-RA-induced increase in IGFBP-3 mRNA was explored by treatment
with 10
6 M t-RA combined with varying doses
of LAP, which noncovalently associates with active TGF-
in the
extracellular space, converting TGF-
into its inactive form.
Examination of conditioned medium samples collected on day 5 revealed
that LAP abrogated the increase in TGF-
activity induced by
10
6 M t-RA (Fig. 7),
confirming its inhibitory effect on extracellular TGF-
activity.
However, LAP did not measurably alter the effects of t-RA on IGFBP-3
mRNA or secreted protein levels (Fig. 8).
In this study, we demonstrated that t-RA increased IGFBP-3 and
TGF-2 mRNA and protein levels in normal HBE cells. t-RA
increases the expression of IGFBP-3 and TGF-
family members in a
variety of transformed and nontransformed cell lines (13-35). In
contrast to our findings in normal HBE cells, the levels of TGF-
1
and -
2 both increase in t-RA-treated hamster tracheal epithelial cells grown in collagen gels (14), providing evidence that retinoid signaling in bronchial epithelial cells depends upon the culture conditions and is species-specific. Further, we found that t-RA stimulated the conversion of latent TGF-
into its active form, suggesting that t-RA activated a protease that cleaves latent TGF-
in normal HBE cells. This TGF-
activity accounted for a relatively
small fraction of total TGF-
, suggesting that protease activity is
tightly regulated. In bovine epithelial cells, t-RA activates latent
TGF-
through enhanced cell-associated plasmin activity (21), whereas
in HL-60 cells, activation of TGF-
by t-RA occurs through
plasmin-independent mechanisms (16).
We examined the role of RAR-, which is expressed constitutively and
activates growth inhibitory pathways in normal HBE cells (11), in these
retinoid signaling events. An RAR-
antagonist inhibited the effects
of t-RA on TGF-
2 and IGFBP-3 expression. Normal HBE cells exhibited
a bimodal pattern of IGFBP-3 mRNA expression, and the antagonist
inhibited the increase at both time points. Treatment with an RAR-
antagonist also blocked the increase in transglutaminase type II
expression induced by t-RA in rat tracheobronchial epithelial cells
(12). Further, RAR activation increases lGFBP-3 expression in human
ectocervical epithelial cell lines (32). These studies support a role
for RARs in specific retinoid signaling events. Retinoid receptors
regulate gene expression directly by binding to gene promoters or
indirectly by protein-protein interactions with other transcription
factors. We found that transcriptional mechanisms contribute to the
increase in IGFBP-3 expression, and t-RA induced this effect rapidly (1 h). Based on these findings, we will investigate direct binding of
retinoid receptors to the IGFBP-3 gene promoter and its role in the
activation of IGFBP-3 expression by t-RA.
We found increased TGF- bioactivity in the media at 72 h of
t-RA treatment, which was coincident with the increase in IGFBP-3 mRNA levels, and TGF-
2 treatment increased IGFBP-3 expression. Similarly, TGF-
2 treatment increases IGFBP-3 expression in breast cancer cells (42). Based on these findings, we investigated whether
t-RA-induced IGFBP-3 expression was TGF-
-dependent.
Supporting this possibility, retinoid signaling events such as
increased expression of thrombospondin and fibronectin are
TGF-
-dependent in rat prostatic epithelial cells (17).
We found that LAP did not abrogate the increase in IGFBP-3 expression
induced by t-RA, suggesting that the increase in IGFBP-3 mRNA
levels by t-RA did not require activation of the TGF-
2 signaling
pathway through autocrine or paracrine mechanisms. It is possible that
multiple retinoid signaling pathways, including
TGF-
-dependent ones, activate IGFBP-3 expression, and
blocking one retinoid signaling pathway is not sufficient to alter the
effects of t-RA on IGFBP-3 expression. Further, TGF-
may activate
IGFBP-3 expression through intracrine mechanisms, which LAP cannot
inhibit. Supporting this possibility, intracrine signaling has been
reported to be a mechanism of TGF-
1 actions in mammary epithelial
differentiation (34).
Prior work in a variety of cell types has demonstrated the importance
of the TGF- and IGFBP-3 signaling pathways in the biologic effects
of t-RA. Blocking antibodies to TGF-
inhibit the mucous differentiation of hamster tracheal epithelial cells by t-RA and partially abrogate the growth inhibitory effects of t-RA in lymphoma cells, rat prostatic epithelial cells, and keratinocytes (13, 14, 17,
18). In breast cancer cells, t-RA increases IGFBP-3 expression, and
antisense IGFBP-3 oligonucleotides abrogate the growth inhibitory
effects of t-RA (24). In contrast, LAP did not alter the growth
inhibitory effects of t-RA on normal HBE cells (data not shown). These
findings suggest that the role of the TGF-
pathway in cell growth
and differentiation depends on the tissue of origin, culture
conditions, and transformation state of the cells examined. Additional
investigations into the roles of the TGF-
and IGFBP-3 signaling
pathways in the biologic effects of retinoid treatment are
warranted.