Smad3 reduces susceptibility to hepatocarcinoma by sensitizing hepatocytes to apoptosis through downregulation of Bcl-2

Abstract

In the liver, derangement of TGF-beta signaling is associated with an increased incidence of hepatocellular carcinoma (HCC), but the mechanism is not clear. We report here that forced expression of a major TGF-beta signaling transducer, Smad3, reduces susceptibility to HCC in a chemically induced murine model. This protection is conferred by Smad3's ability to promote apoptosis by repressing Bcl-2 transcription in vivo through a GC-rich element in the Bcl-2 promoter. We also show that the proapoptotic activity of Smad3 requires both input from TGF-beta signaling and activation of p38 MAPK, which occurs selectively in the liver tumor cells. Thus, Smad3 enables the tumor suppression function of TGF-beta by serving as a physiological mediator of TGF-beta-induced apoptosis.

Figure 1. Expression of Smad3 transgenes in hepatocytes

A: Schematic illustration of Smad3, Smad3SD, and Smad3Δc transgenes. B: Dox-controllable expression of Smad3 transgenes in livers of LAP/S3, LAP/S3SD, and LAP/S3Δc doubly transgenic mice as analyzed by RT-PCR amplification of total liver RNA with primers denoted by arrows in A. Transgenic animals were sacrificed 2 weeks after dietary Dox withdrawal. C: Western analysis of Smad3 transgenes in liver extracts. Fold increase of transgenic relative to endogenous Smad3 level (LAP-tTA control mice) is indicated. D: Immunohistochemistry staining of Smad3 (brown) in liver sections. Both endogenous and transgenic wild-type Smad3 as well as Smad3Δc showed exclusive cytoplasmic staining in resting hepatocytes, whereas Smad3SD exhibited strong nuclear staining. Scale bar, 60 µm.

Figure 2. Ectopic expression of Smad3 protects liver from chemically induced carcinogenesis

A: Percentage of mice developed at least one liver tumor at the end of 6th (n = 9) or 9th (n = 10) month after DEN injection. A separate graph shows the percentage of mice with at least one malignant carcinoma at 9th (n = 10) month after DEN injection. B: Average number of lung tumors per animal at 9th month after DEN injection (n = 10). C-F: H&E staining of liver sections. No macroscopic tumor was detected in livers from LAP/S3 and LAP/S3SD mice. Scale bar, 5 mm. G-J: Higher magnification of liver sections in C–F showing eosinophilic hepatocellular adenomas found in LAP-tTA, LAP/S3, and LAP/S3SD mice and a hepatocellular carcinoma in a LAP/S3Δc mouse. Note the size of adenomas from LAP/S3 and LAP/S3SD livers is much smaller than that from LAP-tTA or LAP/S3Δc livers. Arrowheads denote the tumor margin. Scale bar, 50 µm. K and L: Density of average number of tumors per cm² liver area (K) and percentage of tumor in total liver area (L) measured in liver sections. M: Density of average number of tumors per cm² liver area. Dox was either taken out of the diet when mice were 3 weeks of age (Dox off, 3W), taken out of the diet 6 months after DEN/Pb treatment (Dox off, 6M), or kept in the diet at all times (Dox on). All analyses in this figure were carried out 9 months after DEN injection except where otherwise indicated. In B, K, L, and M, each cohort consisted of ten mice. *p < 1 × 10⁻⁵. Error bars indicate mean ± standard deviation.

Figure 3. Smad3 promotes apoptosis of liver tumor cells in vivo

A: Immunostaining of active TGF-β1 (green) in normal and neoplastic tissues of LAP-tTA control liver. Cell nuclei were counterstained with DAPI (blue). Scale bar, 10 µm. B: Thymidine incorporation assay of primary hepatocytes in the presence (Dox on) or absence of Dox (Dox off). C: Immunohistochemistry staining of the proliferation marker Ki67 (brown) in normal and neoplastic tissues of LAP-tTA liver. Scale bar, 100 µm. D: Quantification of Ki67-positive cells in normal and neoplastic liver tissues. E: TUNEL assay of apoptotic cells (green) in tumors. Cell nuclei were counterstained with DAPI (blue). Scale bar, 100 µm. F: Quantification of TUNEL-positive cells in normal and neoplastic liver tissues. In D and F, the results represent mean values of 12 tumors, except in the case of LAP/S3SD, in which the mean value was derived from four tumors. Error bars indicate mean ± standard deviation.

Figure 4. Smad3 enhances the responsiveness of hepatocytes to Fas-mediated apoptosis

A: Liver function of transgenic mice (n = 4) measured as serum alanine aminotransferase (ALT) activity (n = 4). B: Quantification of TUNEL-positive cells in liver sections of transgenic mice (n = 4) after Fas activation. Mean values compiled after counting five magnification (20×) fields for each liver are shown. C: Kaplan-Meyer survival curve of each transgenic strain after Fas activation (n = 10). D: H&E staining of liver sections after Fas activation. Scale bar, 250 µm E: Western analysis of caspase cleavage in liver extracts. Arrowheads denote cleaved products of caspases. F: Western analysis of phospho-Smad3 and total Smad3 in liver extracts. G: Immunohistochemistry staining of Smad3 (brown) in liver sections 3 hr after Jo-2 injection. Scale bar, 50 µm. Note that wild-type Smad3 (endogenous or transgenic) in LAP-tTA and LAP/S3 accumulated in the nucleus after Fas was activated, while total Smad3 in LAP/S3Δc mice was still retained in the cytoplasm of hepatocytes. Error bars indicate mean ± standard deviation.

Figure 5. Dual requirements of Smad3 and p38 MAPK in TGF-β-induced apoptosis

A: Immunostaining of active TGF-β1 in mock or Fas-activated liver sections as in Figure 3A. Scale bar, 10 µm B: ELISA assay of DNA fragmentation (Cell Death Detection Kit, Roche) as a function of apoptosis in primary hepatocytes isolated from transgenic mice in the presence or absence of Dox, and/or TGF-β. C: Specific requirement of Smad3 in TGF-β-induced apoptosis in primary hepatocytes. The specificity of Smad siRNAs is shown in the Western analysis of siRNA-transfected hepatocytes in the left panel. NS, nonsilencing control siRNA. D: Immunohistochemistry staining of phospho-p38 in liver sections. Note that the weak brown staining of phospho-p38 MAPK is concentrated in the interstitial stellate cells in nontumor tissue, whereas it is highly induced in hepatocytes in the tumors. Scale bar, 50 µm. E: Requirement of p38 MAPK but not JNK, Erk, Akt, or ROCK in TGF-β-induced apoptosis in primary hepatocytes. Cell death was monitored by ELISA assay as in B. Error bars indicate mean ± standard deviation.

Figure 6. Smad3 downregulates Bcl-2 expression

A: Western analysis of different Bcl-2 family members and Akt in liver extracts from mice treated with or without Jo-2. B: RT-PCR analysis of Bcl-2 and Bcl-xL mRNA in liver extracts of A. RNA from two different livers was used for each group. C: Immunohistochemistry staining of Bcl-2 in liver tumors after 9 months of DEN/Pb treatment. Representative data of four tumor samples in each cohort are shown. Scale bar, 100 µm. D: Western analysis of endogenous Bcl-2 protein levels in retrovirus-infected SK-Hep-1 cells expressing LPCX vector, FLAG-tagged Smad3, or Smad3Δc. Cell lysates were collected after TGF-β treatment for 24 hr. Expression of FLAG-tagged Smad3 or Smad3Δc is shown in the middle panel. Endogenous GAPDH was used as the loading control (bottom panel). *Nonspecific band. E: Suppression of TGF-β-induced apoptosis in Hep3B and SK-Hep-1 cells by adenovirus-mediated expression of Bcl-2. Expression of Bcl-2 in the Hep3B and SK-Hep-1 cell lysates was confirmed by a Western analysis in right panel. Note that TGF-β treatment did not affect the level of exogenous Bcl-2 protein. Error bars indicate mean ± standard deviation.

Figure 7. Smad3 represses Bcl-2 transcription

A: Luciferase assays of different Bcl-2 promoter reporters in transfected Hep3B cells. One day after transfection, the cells were treated with TGF-β for 20 hr. B: Luciferase assays of the Bcl-2 P2 (−1278) promoter reporter in transfected Hep3B or SK-Hep-1 cells as in A. C: Luciferase assays of the Bcl-2 P2 (−1278) promoter reporter in the presence of nonsilence control (NS), Smad2, or Smad3 siRNA. The siRNAs were cotransfected with reporter DNAs, and 2 days after transfection, the cells were treated with TGF-β for 20 hr. D: Western analysis of Bcl-2 expression in SK-Hep-1 cell lysates after 2 days of the siRNA transfection. Error bars indicate mean ± standard deviation.

Figure 8. Smad3 specifically represses Bcl-2 transcription via a GC-rich Smad binding element

A: Schematic representation of Bcl-2 promoters. Nucleotide coordinates are anchored to the Bcl-2 translational initiation site. The conserved GC-rich element is underlined, and the Drosophila Mad binding consensus and inactivating mutations are shown. B: Mapping of TGF-β-responsive element in the Bcl-2 promoter region in transfected Hep3B cells. TGF-β treatment lasted for 20 hr, and the promoter activity was presented as the ratio of luciferase activity between TGF-β-treated and nontreated samples. C: Reversal of the TGF-β-mediated transcription repression of Bcl-2 −802 reporter by Smad3-specific siRNA. Note that neither TGF-β nor Smad3 siRNA had any effect on Bcl-2 −802m reporter, in which the GC-rich element was mutated. D: Electrophoresis mobility shift assay for specific interaction between the purified GST-Smad3 DNA binding domain (Smad3NL) and DNA oligos containing the GC-rich element. GST and GST-Smad2NL were used as controls. Competition by a 50-fold excess of unlabeled wild-type but not mutant (m or m2 as in A) DNA probe, and supershift by anti-Smad3 antibody indicate specific binding of Smad3NL to the DNA. DNA-protein (shift) and antibody-induced supershift (ss) complexes are marked. E: ChIP assay for binding of Smad3 to the Bcl-2 promoter in vivo in the Fas-activated liver tissues. Animals were treated with Jo-2 antibody for 3 hr prior to being sacrificed. PCR primers for amplifying Bcl-2 promoter are specified in the Experimental Procedures. F: ChIP assay for binding of Smad3 to the Bcl-2 promoter in vivo in primary hepatocytes. Prior to TGF-β treatment for 1 hr, cells were cultured in the absence of Dox for 24 hr to allow for the expression of Smad3 transgenes. Error bars indicate mean ± standard deviation.