Full Text (PDF)
The cytokine transforming growth factor β (TGF-β) is involved in the regulation of organismal development, cell differentiation, proliferation, apoptosis, tissue maintenance, and wound healing (1). Perturbations of the TGF-β system are thought to play a central role in the pathogenesis of diseases such as fibroproliferative disorders and cancer. Therefore, it is important to develop experimental tools that allow visualization of TGF-β pathway activation in vitro and in vivo.
TGF-β binds to its specific receptor, TGF-β receptor II (TβRII), which heterodimerizes with and trans-phosphorylates the TGF-β receptor I (TβRI), followed by activation of intracellular signaling cascades. These involve canonical TGF-β signaling via TβRI-dependent phosphorylation of Smad2 and Smad3, which heterodimerize with Smad4, translocate into the nucleus, bind TGF-β regulatory elements, and alter gene expression (2,3). Additionally, TGF-β can activate non-Smad pathways such as MAPK, PI3K/Akt, pp2A/S6 kinase, and Rho/Rock (2,3).
The complex regulation of both TGF-β activation and downstream signal transduction makes it difficult to determine the activation status of the TGF-β pathway. A Smad4-based reporter plasmid has recently been described (4). However, because Smad4 also mediates bone morphogenetic protein (BMP) signaling, this assay lacks specificity. Antibodies specific for phospho-Smad2 have been developed and allow assessment of Smad2 activation by immunohistochemical approaches (5). However, the same approach is problematic for assessing Smad3-mediated signaling in complex situations like co-cultures and tumor activation, where BMP signaling is difficult to exclude because phospho-Smad3 antibodies also recognize the BMP-stimulated phospho-Smad1 and Smad5. A Smad3-based reporter plasmid has been used for reporting signaling in vivo by luciferase, but this approach cannot be used to determine temporospatial distribution of Smad3 signaling at a cellular rather than tissue level (6,7).
TGF-β plays a complex dual role as tumor suppressor and metastasis promoter (8), and the balance between Smad2 and Smad3 or between Smad and non-Smad signaling may be important for determining the outcome. Smad2 has been found to be mutated or deleted in some human cancers, but Smad3 is not, suggesting that these two closely related Smads may play distinct roles in tumorigenesis (9). The role of Smad3 as a potential therapeutic target in cancer and fibroproliferative disorders makes it highly desirable to identify which cells, in cultures or tissues specifically, have active Smad3 signaling and under what circumstances. Here we describe a method that allows assessment of the spatiotemporal pattern of TGF-β signaling via Smad3 in cells and reconstituted tissues on a cellular level.
We have used a lentivirus-based reporter vector system expressing green fluorescent protein (GFP), red fluorescent protein (RFP), or luciferase under the control of a Smad3-responsive element (CAGA)12 (10) to stably infect human and mouse cells. The (CAGA)12 element was inserted between EcoRI and SpeI sites of a Cop-GFP/RFP lentiviral vector (SBI, Mountain View, CA, USA), and virus was generated with these custom made vectors according to the manufacturer's instructions. In brief, 293T cells (3 × 106 cells/80 cm2 growth area) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) overnight, and then transfected with reporter plasmid (2 µg/plate) and packaging plasmid mix (10 µg/plate) using Lipofectamine Plus™ reagent (Invitrogen, Carlsbad, CA, USA). Medium was replaced with low serum medium (0.5% FBS) 12 h later. Virus-containing supernatants were collected after 24 and 48 h and filtered (0.45 µm) to remove cell debris. For lentiviral infection, cells were seeded at a density of 5 × 105 cells/80 cm2 growth area and incubated with lentiviral supernatant (3 mL/plate) supplemented with Polybrene® (4 µg/mL; Sigma-Aldrich, St. Louis, MO, USA) for 24 h.
We found that a variety of cell lines including mouse embryonic and dermal fibroblasts, mouse mammary cancer cells (4T07), human breast epithelial (MCF10A), and tumor cells (MCF10CA1h, MCF10CA1a) (11,12,13) could be infected with virus constitutively expressing GFP with high efficiency, although GFP expression was generally higher in human as compared with mouse cells (Figure 1A). Constitutive GFP expression in stably infected cells persisted for at least 20 passages and was not altered by stimulation of these cells with TGF-β (data not shown). To investigate specificity for Smad3 signaling versus Smad2 signaling and sensitivity of (CAGA)12-driven GFP expression, we stimulated serum-starved (overnight) HepG2 (0.5% FBS), MCF10A, and MCF10CA1h (0.2% horse serum) with TGF-β (2 ng/mL) alone or 30 min after addition of small-molecule inhibitors, and GFP expression levels were monitored 24 h later. TGF-β-induced GFP expression was inhibited by the ALK5 (TβRI) kinase inhibitor SB431542 (5 µM), but not the structurally related p38 kinase inhibitor SB203589 (10 µM; Figure 1, B and C), the MEK/Erk inhibitor PD98059 (25 µM), or the ROCK inhibitor Y27632 (20 µM; data not shown). TGF-β (2 ng/mL) induced GFP expression in overnight serum-starved [0.5% fetal calf serum (FCS)] Smad2 knockout fibroblasts within 24 h, but not in Smad3 knockout fibroblasts (Figure 1D). These results prove that the construct indeed responds to TGF-β in a Smad3-dependent manner, and does not respond to Smad2- or non-Smad-mediated TGF-β signaling. TGF-β stimulation of GFP expression in overnight serum-starved (0.2% horse serum) MCF10CA1h cells was dose-dependent between 0.02 ng/mL and 5 ng/mL (Figure 1, E and F) for both (CAGA)12-mediated GFP and luciferase expression. However, the GFP reporter gave a better signal-to-noise ratio than the luciferase reporter in the low TGF-β concentration range (Figure 1F), suggesting that the fluorescence approach has advantages for reporting on the low levels of TGF-β that are likely to occur endogenously in vivo. Infection of cells with a (CAGA)12-RFP construct yielded similar results (data not shown), thus allowing us to simultaneously study endogenous Smad3 signaling in more than one cell type. We employed (CAGA)12-GFP-infected MCF10CA1a cells and (CAGA)12-RFP-infected mouse fibroblasts to track temporospatial Smad3 signaling in a co-culture system (12) (Figure 2). 5 × 105 MCF10CA1a cells and 5 × 105 fibroblasts were plated in 10 mL co-culture medium (46.25% DMEM, 46.25% DMEM F12, 5% FBS, 2.5% horse serum) in a 100-mm tissue culture dish and incubated at 5% CO2, 37°C in a humidified atmosphere. In contrast to homotypic cultures in which very little activation of Smad3 was observed, co-cultures showed activated Smad3 signaling in both cell types, tumor cells, and fibroblasts, with a maximum of Smad3-mediated RFP/GFP expression at 2 days (Figure 2A). The ALK5 inhibitor SB431542 (5 µM) but not dimethyl sulfoxide (DMSO) (control) abolished Smad3-driven GFP expression in co-cultures, confirming that TGF-β signaling through this receptor contributes to Smad3 signaling in co-cultures (Figure 2B). We next explored if the reporter system can be used for in vivo applications. (CAGA)12-GFP-infected MCF10CA1a cells (5 × 105 cells) and (CAGA)12-RFP-infected fibroblasts (5 × 105 cells) were coinjected into the mammary fat pad IV/V of athymic nude mice (nu/nu; 8-week-old), and the tumors that developed were excised after 13 days. Tumors were then cut into cross-sections using a razor blade. Sections were flipped onto a coverglass, covered with Dulbecco's phosphate-buffered saline (DPBS) containing Hoechst 33342, and used for confocal microscopy. In this xenograft model, (CAGA)12-mediated GFP and RFP expression was seen in a subset of tumor cells and fibroblasts (Figure 2C). This indicates that this reporter system has the sensitivity to report on endogeneous TGF-β signaling in vivo.
Our data demonstrate that (CAGA)12-driven GFP/RFP expression is a powerful and sensitive tool to observe the spatiotemporal pattern of endogeneous Smad3 signaling at the cellular level not only in vitro, but also in vivo, as for example in different areas of tumors or during tumor cell invasion and metastasis. This approach should greatly enhance our knowledge about activation of TGF-β signaling cascades in complex biological systems.
C.H.S. and A.K.K. contributed equally to this work. A.B.R. died on May 26th, 2006. The authors very much appreciated the help of Tatiana Karpova and James McNally with quantification of images. Smad2 knockout and the corresponding wild-type fibroblasts were a gift from Erwin Bottinger.
The authors declare no competing interests.
