to difference in averages of two hybridizations of receptor-expressing adenovirus-infected cells versus control vector. The METAP2 gene is excluded for lack of ...
[Cancer Biology & Therapy 2:2, 164-170, March/April 2003]; ©2003 Landes Bioscience
Research Paper
The Essential Similarity of TGFβ and Activin Receptor Transcriptional Responses in Cancer Cells ABSTRACT
*Correspondence to: Scott E. Kern, M.D.; Sidney Kimmel Comprehensive Cancer Center; Cancer Research Building 451; 1650 Orleans Street; The Johns Hopkins University School of Medicine; Baltimore, Maryland 21231 USA; Fax: 410.614.9705; E-mail: sk@jhmi.edu.
rd
Received 01/23/03; Accepted 01/27/03
rib ut io n.
Sidney Kimmel Comprehensive Cancer Center and Department of Oncology; The Johns Hopkins Medical Institutions; Baltimore, Maryland USA
The binding of activin and TGFβ to their respective receptors initiates signals that are carried by common intermediates (Smad proteins) to induce transcriptional activation of downstream genes. Mutations in tumors indicate that both receptor types convey tumorsuppressive signals, among other biologic roles, but their respective sets of transcriptional targets (transcriptomes) and the shared degree of transcriptome similarity are not well explored in these cells. Transcriptome changes were analyzed by gene expression profiling after expression of constitutively active activin type I (ALK4m) and TGFβ type I (ALK5m) receptors and by variation of Smad4 expression in cancer cells. Eleven of 15 previously reported TGFβ downstream genes were confirmed to be responsive to TGFb and activin receptors in cancer cells. Expression profiling detected eight of these 11, as well as 13 new Smad4dependent transcripts. Although Smad4-dependent CDKN1A/p21 induction represents the sole known effector of TGFβ and activin tumor-suppressor effects, many downstream genes have not yet been evaluated for a suppressive role. A high similarity of TGFβ and activin responses among the known and new transcriptional target genes indicated an essential redundancy of the two related inputs. This similarity helps relate the mutations seen in both receptor systems and their Smad mediators in human cancers.
is t
Byungwoo Ryu Scott E. Kern*
e.
Pancreatic cancer, Activin, transforming growth factor beta, Signal transduction, Gene expression profiling
N
KEY WORDS
ot
fo
Previously published online as a CB&T “Paper in Press” at: http://landesbioscience.com/journals/cbt/
©
20 03
La
nd
es
Supported by NIH grant CA68228.
164
ie
INTRODUCTION
io
sc
The Smad signal transduction mechanism is initiated by ligands binding at the cell surface to cognate receptors. This initiates a signal chain mediated by Smad proteins, resulting in transcriptional activation of individual genes (reviewed in refs.1–4). Differential use of the receptor-activated Smads (r-Smads) defines two distinctive types of Smad pathways, respectively activated by the ligands, TGFβ and activin or by bone morphogenic proteins (BMPs). Smad4 is considered to be a common mediator of both types, based on functional and biochemical studies. For examples, Smad4 is required for activation of downstream genes and of reporter genes transactivated through the Smad-binding element (SBE).5-9 The importance of TGFβ and activin-activated Smad signaling in cancer derives in part from observations of the growth inhibitory and apoptotic effects of TGFβ, activin, and Smad activation in cells.5,10-13 More convincingly, tumor-suppressive qualities are evidenced by a high frequency of mutational inactivation in MADH4 (Smad4, DPC4) in pancreatic cancer and occasional mutations of MADH4 and MADH2 (Smad2, JV18-1) in other cancer types.14-17 Mutations in type I and type II receptors for both activin and TGFβ confirm that both receptor systems initiate tumor-suppressive signals.18-22 Compared to TGFβ, there has been considerably less study of activin for cancer-related biologic properties. A high similarity of TGFb and activin signals suggest that both ligands originate related tumor-suppressive effects. The ligands and their heterodimeric receptors are evolutionarily closely related. Both share the use of r-Smads 2 and 3 with the co-Smad 4 and there is similarity in their activation of DNA response elements to cause to gene transcription.23 For example, the activin-responsive DNA element binds Fast1 protein to transactivate genes in response to both activin and TGFβ and can cooperate with Smad signaling.24 TGFβ-responsive reporters are also activin-responsive.7,24 Despite the similar mechanism of signaling used by activin and TGFb, an extended comparison of the sets of genes (the transcriptomes) activated by both receptor systems is not available. The identification of the downstream target genes of SMAD signaling would facilitate such a comparison and would contribute to a better understanding of the biologic roles of
B
TGFβ Type β transforming growth factor SBE Smad-binding element ALK4m Constitutively active mutant form of type I activin receptor ALK5m Constitutively active mutant form of type I TGFβ receptor
nc
ABBREVIATIONS
Cancer Biology & Therapy
2003; Vol. 2 Issue 2
THE ESSENTIAL SIMILARITY OF TGFβ AND ACTIVIN RECEPTOR TRANSCRIPTIONAL RESPONSES IN CANCER CELLS
the activin and TGFβ ligands. Their functional roles include regulation of proliferation, differentiation, migration and apoptosis of multiple cell types seen in multiple biologic systems, mediated by the transcriptional activation, most likely, of some known, and a larger number of uncharacterized, downstream target genes. There are a number of other genes known to be up-regulated by TGFβ, however, for most their functional roles are likely unrelated to tumor-suppression activities (table 1 and refs. 25,26). Although CDKN1A/p21 (and CDKN2B/p15 with much less support) have been proposed to mediate the property of tumor suppression,27,28 the other downstream genes potentially responsible for ligand/receptor/ Smad-dependent tumor-suppression are largely unexplored, as judged from the limited literature. Investigators have experienced some difficulties that have impaired such explorations. For example, some experimental studies of TGFβ are consistent with the existence of Smad-independent pathways, including those dependent on ras, mitogen- and stress-activated kinase pathways (MAPKs and SAPKs).29-37 In expression profiling of the TGFβ transcriptome, a disappointing paucity of distinct transcriptional targets was encountered due to widespread and non-robust changes in the cellular transcriptome in response to TGFβ.26 In the current study, we overcame the practical difficulties of ligand administration by employing a receptor-specific activation of Smad signaling, through use of constitutively active mutant type 1 receptors, ALK4m (activin ACVR1B receptor) and ALK5m (TGFβ receptor). We compared the transcriptomes induced by TGFβ and activin receptors using newly identified and known genes and assessed the Smad-dependence of the transcriptional responses.
MATERIALS AND METHODS Adenovirus Construction. The pAdEasy system was a gift of Dr. Bert Vogelstein. Vectors encoding constitutively active mutant forms of activin (ALK4m) and TGFβ (ALK5m) type 1 receptors, pCMV5B-ActRIB/HA and pCMV5-TbR1/HA, were gifts of Dr. Jeff Wrana. Adenoviruses expressing Alk5m and Alk4m proteins were constructed using the AdEasy system as described.38 In brief, the shuttle vectors (pAdTrack-CMV-ALK5m and pAdTrack-CMV-ALK4m) carrying the receptor sequences were created by insertion of ALK5m and ALK4m open-reading frames at the KpnI and XbaI restriction sites of the pAdTrack-CMV vector. Generation of recombinant adenoviral plasmids by homologous recombination in E. coli was conducted using adenoviral plasmid pAdEasy-1 and shuttle vectors (pAdTrack-CMV-ALK5m, pAdTrack-CMV-ALK4m, and pAdTrack-CMVControl) linearized with PmeI in electrocompetent E. coli BJ5183 cells. The resulting adenoviral plasmids were termed pAdALK5m, pAdALK4m, and pAdCon. The high-titer viral stocks were generated in mammalian 293 cells, aliquotted into small volumes, and stored at -80˚C until use. Cell Lines Used. MADH4 wild-type pancreatic cancer cell lines, Panc-1, MiaPaCa1, Su86.86, PL-45, and the MADH4-null BxPc3, were obtained (except for PL-45) from American Type Culture Collection (ATCC), maintained in growth medium (either DMEM or RPMI 1640 (Invitrogen), as recommended by ATCC, supplemented 10 % FBS (HyClone), 100 units/ml of penicillin, and 100 µg/ml of streptomycin) at 37˚C in 5% CO2. The MADH4-null breast cancer cell line, MDA-MB-468*, has been described; it has a stably integrated chimeric gene encoding a tamoxigenresponsive modified murine estrogen receptor (MER) ligand binding domain12,39 fused to the Smad4 C-terminus (Smad4-MER). It was maintained in Leibovitz’s L-15 (Invitrogen), supplemented as above. Reporter Assay. In order to determine the cell lines optimally TGFβresponsive, Panc-1, MiaPaCa1, Su86.86, and PL-45 cells were plated on 6-well cluster dishes at a density of 2x105 per well and transfected with 0.5 mg of p6SBE-Luc,8 which contains Smad consensus palindromic binding sites. In all experiments, 0.5 mg of β-galactosidase expression vector
www.landesbioscience.com
Figure 1. Overexpression of constitutively active activin and TGFβ type I receptors and RNA harvesting-time was optimized. (A) Equalization of pAd-ALK4m and pAd-ALK5m adenovirus infection based on reporter assay. Panc-1 cells, in which luciferase reporter gene controlled by six Smad binding elements is stably integrated, were infected with various volumes of adenovirus. After 24 hours infection, cells are harvested and reporter activity was measured. The means of three independent experiments were shown. Bars, standard error of the mean. (B and C) Time courses of ALK4m and ALK5m induced SERPINE1/PAI1 and JUNB gene transactivation. 40 ml of pAdALK4m and 200 ml of pAdALK5m were used to infect Panc-1 cells. Total cellular RNA was isolated on the indicated time-points and RT-PCR performed. As examples, the amplified product was separated electrophoretically and visualized by ethidium bromide fluorescence for SERPINE1 gene (B) and relative fold induction of JUNB transactivation was monitored in real-time (C)
Cancer Biology & Therapy
165
THE ESSENTIAL SIMILARITY OF TGFβ AND ACTIVIN RECEPTOR TRANSCRIPTIONAL RESPONSES IN CANCER CELLS
Table 1
INITIAL ANALYSIS
OF
KNOWN TGFβ TESPONSIVE GENES† Panc-1
Genes
MDA-MB-468*
Function
TGFβ
ALK5m
ALK4m
4-OHT
Serine (or cysteine) protease inhibitor, blood coagulation (PAI1)
+++
+++
+++
++
JUNB
Proto-oncogene, transcription regulator from Pol II promoter
+++
+++
++
FN1
Cell adhesion & migration (fibronectin)
++
NC
NC
+
CDKN2B
Cyclin-dependent kinase inhibitor, negative control of cell proliferation (p15)
NC
NC
NC
NC
CDKN1A
Cyclin-dependent kinase inhibitor, negative control of cell proliferation (p21)
+
++
++
+
MADH7
Inhibitor of SMAD pathway (SMAD7)
+
+
+
+
Growth factor (TGFβ1)
+
NC
NC
NC
SPP1
Bone & blood vessel ECM protein (osteopontin)
+
+
+
TIEG
TGFb-inducible early growth response, cell proliferation, transcription factor
+
+
+
SERPINE1
TGFB1
MMP2
Breakdown of ECM
+
+
+
Connective tissue metabolism (biglycan)
+++
+++
+++
BHLHB2
Basic helix-loop-helix domain containing class B2, transcription factor (DEC1)
+
+
+
PDGFB
Growth factor (PDGFβ)
+
+
+
CTGF
Ligand for α-2 macroglobulin receptor, wound healing, growth factor
NA
NA
NA
PLAU
Serine protease (urokinase plasminogen activator)
NC
NC
NC
BGN
NA
NA
†Relative fold-induction of previously known TGFβ-responsive genes; +++: ≥8-fold; ++: ≥4 to
