TGF-β1–induced expression of human Mdm2 correlates with late-stage metastatic breast cancer (original) (raw)
TGF-β1 regulates p53 stability. To examine the ability of TGF-β1 to regulate p53 stability, we utilized HCT116 colorectal cells. HCT116 cells are deficient in DNA mismatch repair and TβRII (21). HCT116:3-6 cells were produced by reintroduction of chromosome 3 into cells by microcell fusion, which restored TβRII protein expression and activity (22). HCT116 and HCT116:3-6 cells were used as a model system to analyze the effect of the TGF-β1 pathway on the regulation of p53. Treatment of HCT116 cells with TGF-β1 for 48 and 72 hours did not alter p53 levels (Figure 1A). Conversely, decreased levels of p53 were observed at 24 and 48 hours in HCT116:3-6 cells treated with TGF-β1 (Figure 1A). To determine whether this loss was due to Hdm2-mediated destabilization of p53, HCT116:3-6 cells were treated with TGF-β1 for 24 and 48 hours and a Western blot was prepared to examine higher molecular weight forms of p53 that would be consistent with ubiquitination. The data in Figure 1B show formation of higher molecular weight bands of p53 in HCT116:3-6 cells after TGF-β1 treatment. To demonstrate that ubiquitin was conjugated to p53, HCT116:3-6 or HCT116 p53–/– cells were pretreated with the proteasome inhibitor MG132, followed by TGF-β1 exposure. We observed that immunopurified p53 from HCT116:3-6 cells was conjugated with ubiquitin in response to TGF-β1 treatment by probing for ubiquitin (Figure 1C). In addition, we also transiently transfected HA-tagged ubiquitin (HA-Ub) into HCT116:3-6 cells and treated them with TGF-β1 and in the absence or presence of MG132 (Figure 1D). These data further confirm that p53 was conjugated with ubiquitin in response to TGF-β1 exposure.
p53 destabilization by TGF-β1 stimulation. (A) HCT116 and HCT116:3-6 cells were treated with vehicle or 10 ng/ml of TGF-β1 for 48 and 72 hours. Cellular extracts were prepared for Western blot, and p53 and Ku70 (internal control) levels were detected. (B) Western blot of p53 laddering, an indication of ubiquitination (Ub) (arrows) in cellular extracts of HCT116:3-6 cells treated with 10 ng/ml of TGF-β1 for 0, 24, and 48 hours. (C) HCT116, Hct116 p53–/–, and HCT116:3-6 cells were treated with 10 ng/ml of TGF-β1 with or without the proteasome inhibitor MG132 (30 μM). Cellular extracts were immunoprecipitated with control IgG or p53 antibodies. Precipitates were separated on an SDS-PAGE gel, and a Western blot was prepared. The left side was probed for ubiquitin. The right side was probed for p53. (D) HCT116:3-6 cells were transfected with HA-Ub. Cells were then treated with MG132 (30 μM) with or without TGF-β1 (10 ng/ml); cellular extracts were immunoprecipitated with p53 or IgG antibodies and prepared for Western blot analysis of HA. (E) Western blot of p53 in extracts isolated from HCT116:3-6 cells treated with TGF-β1 (10 ng/ml) with or without Nutlin3 (10 μM).
Ubiquitination of p53 requires an E3 ubiquitin ligase. There are several E3 ubiquitin ligases that may conjugate ubiquitin to p53, including Cop1, Mule, Pirh2, and Mdm2 (human and murine) (23–27). Through overexpression and knockdown studies, Mdm2 (or human homolog Hdm2) is the best-known E3 ubiquitin ligase for p53. The regulation of p53 stability requires the induction of HDM2 gene expression by p53 in response to genotoxic stress (28). Hdm2 associates with p53 via van der Waal forces within the first 52 amino acids of p53; this site serves as a docking site for subsequent conjugation of ubiquitin to the carboxyterminal lysines of p53 (29). To determine whether Hdm2 played a role in the destabilization of p53 in response to TGF-β1, we used Nutlin3, a small molecule inhibitor that sterically inhibits the association of p53 to Hdm2 by binding the hydrophobic pocket in the amino terminus of Hdm2 (30). The data in Figure 1E demonstrate that 10 μM Nutlin3 protected p53 from degradation mediated by the TGF-β1 signaling pathway. Thus, the observed stabilization of p53 by Nutlin3 implicates Hdm2 in the destabilization of p53 in response to TGF-β1.
Induction of HDM2 gene expression by TGF-β1. It has been reported that an increase in Hdm2 levels causes the formation of Hdm2 homodimers, which in turn function to destabilize p53 (31). Therefore, it was necessary to further characterize levels of Hdm2 protein in response to TGF-β1. HCT116:3-6 and HCT116 cells were treated with TGF-β1, and whole-cell extracts were isolated between 0 and 72 hours. Treatment of HCT116 cells with TGF-β1 did not alter the levels of Hdm2 over 72 hours, while HCT116:3-6 cells showed a dramatic induction in Hdm2 levels with a concomitant decrease in p53 over the same time course (Figure 2A).
TGF-β1 increases Hdm2 mRNA and protein levels. (A) Western blot analysis of Hdm2 and p53 in HCT116 and HCT116:3-6 treated with TGF-β1 (10 ng/ml) for 0, 24, 48, and 72 hours. (B) Vaco400 cells with control plasmid (Vaco400:neo) or expressing the functional TGF-β1 receptor II (Vaco400:RII) cells were treated with vehicle (veh) or TGF-β1 (10 ng/ml) for 24 or 48 hours. Western blot was prepared from the extracts, and Hdm2, p53, and tubulin were detected. (C) HCT116 and HCT116:3-6 were transfected with the HDM2 promoter-luciferase reporter construct and renilla expression plasmid; then cells were treated with TGF-β1 (10 ng/ml) or vehicle. After 48 hours, extracts were prepared for analysis. Fold induction represents vehicle to TGF-β1 treatments and error bars represent SD generated from the mean. (D) Real-time PCR was performed on HCT116:3-6 cells treated with vehicle or TGF-β1 (10 ng/ml) at 6 or 24 hours.
We next examined another colorectal carcinoma cell line, Vaco400, which is also deficient in mismatch repair and lacks a functional TβRII. Control vector (Vaco400:neo) or Vaco400 cells that stably express TβRII (Vaco400:RII) were used to determine whether the observations in HCT116:3-6 were exclusively dependent on TβRII. Both Vaco400:neo and Vaco400:RII cells were treated with TGF-β1 for 24 or 48 hours. The data in Figure 2B are similar to the results from the isogenic HCT116 model system, whereby stimulation with TGF-β1 caused an increase in Hdm2 levels and a subsequent decrease in p53 only in cells that possess the TβRII. Both HCT116 and Vaco400 cell systems provided reproducible data of increased Hdm2 protein levels and the subsequent loss of p53.
These experiments led us to examine the mechanism by which Hdm2 was increased. In order to determine whether it was a transcriptional event, a HDM2 reporter assay was used to determine whether the P2 promoter of HDM2 was activated in response to TGF-β1 stimulation. The HDM2 reporter was constructed by linking the _HDM2_-P2 promoter upstream of a luciferase reporter gene. In addition to reporter assays, we examined changes in HDM2 mRNA levels by real-time PCR to determine whether the endogenous gene was induced in response to TGF-β1 treatment. The HDM2 promoter was reproducibly induced in response to TGF-β1 treatment in TβRII-reconstituted HCT116:3-6 cells, as determined by reporter assay (Figure 2C) and by real-time PCR (Figure 2D). These experiments demonstrate that TGF-β1 increased Hdm2 levels through a transcription-dependent event.
Induction of the Mdm2 gene (human and murine) through its P2 promoter is largely regulated by p53 in response to genotoxic stress. Therefore, it became important to characterize the involvement of p53 in the induction of the HDM2 gene in response to TGF-β1. Two cell lines, 293T cells transformed with large T antigen that predominantly neutralizes p53 and SKOV3 ovarian cancer cells that lack p53 expression, were used to assess whether induction of Hdm2 was independent of functional p53 when cells were subjected to TGF-β1 treatment. Treatment of either 293T or SKOV3 cells with TGF-β1 resulted in increased levels of Hdm2 (Figure 3A). Additionally, titration of a constitutively activated TβRI (caTβRI) plasmid in 293T and SKOV3 cells caused an increase in Hdm2 levels (Figure 3B). Furthermore, utilizing a luciferase reporter assay with the second promoter of HDM2 or the TGF-β1 responsive SBE2X2 (a synthetic promoter consisting of 4 GTCTs with AGAC linkers) as a positive control, we demonstrated that caTβRI mediates induction of HDM2 in 293T, SKOV3, and HCT116:3-6 cells (Figure 3C). Thus, activation of the TGF-β1 pathway by ligand or a constitutively active receptor resulted in induction of the HDM2 gene promoter and increased Hdm2 protein expression independent of p53 activity.
Increased Hdm2 in response to TGF-β1 is independent of p53. (A) 293T and SKOV3 cells were untreated (UT) or treated with vehicle or TGF-β1 (10 ng/ml) for 24 and 48 hours. Cellular extracts were prepared for Western blot analysis of Hdm2 and Ku70. (B) SKOV3 and 293T cells were transiently transfected with 0, 4, or 10 μg of constitutively active TβRI (caTβRI) expression plasmid. Cellular extracts were prepared for Western blot analysis of Hdm2 and α-tubulin. (C) Transient transfection of 293T, SKOV3, and HCT116:3-6 cells with the caTβRI and HDM2 reporter or the SBE2X2 reporter construct. All samples were transfected with a renilla construct. Reporter activity was determined relative to renilla to generate relative activity. Fold induction was determined relative to vehicle control and SD was calculated relative to the mean.
Induction of HDM2 gene expression has been reported in response to stimuli other than genotoxic stress. Transcription factors such as N-Myc, Ets family members, Sp1, and ERα can bind and induce HDM2 gene expression through the second promoter (P2) (32–34). After an in silico survey of the P2 promoter of HDM2 for possible transcription factor–binding elements that are activated in response to TGF-β1, 2 possible binding elements in the promoter conformed to the GTCT Smad3 DNA–binding element (SBE). One binding element (SBE2) was located at nucleotide –245 in the P2 promoter region of HDM2 and a second site (SBE1) was located at nucleotide –585 in the P2 promoter region of HDM2. To explore the possibility of Smad3 inducing HDM2 gene expression, we determined whether Smad2 and Smad3 were active after TGF-β1 exposure at times when Hdm2 levels were elevated. Dimerization of TβRI/RII in response to TGF-β1 formed an active serine/threonine kinase that phosphorylates and activates Smad2 and Smad3. We observed Smad3 activation using a phospho-antibody to Smad3 to probe a Western blot of cellular extracts from 293T, SKOV3, Vaco400:RII, and HCT116:3-6 cells treated with TGF-β1 for 48 hours. As expected, Smad3 activation was not observed in parental HCT116 cells treated with TGF-β1 (Figure 4A). Importantly, Smad3 activation occurred and corresponded with induction of the HDM2 gene. To further explore the involvement of Smads in the activation of HDM2 gene induction, we transiently transfected a dominant negative Smad4 (dnSmad4) in transient reporter assays. Smad4 is a binding partner of Smad3, and a dominant negative form would presumably impede the activation of the HDM2 promoter. Using the HDM2 promoter and SBE2X2 as a control, cells transiently transfected with the respective promoters were treated with or without TGF-β1. As predicted, dnSmad4 blocked induction of the synthetic Smad reporter SBE2X2 and also prevented the activation of HDM2 promoter (Figure 4B). Thus, these data support a role for the Smads in the induction of Hdm2 in response to TGF-β1 treatment.
Smad activation and induction of the HDM2 promoter. (A) Western blot of p-Smad3 in 293T, SKOV3, Vaco400:RII, HCT116, and HCT116:3-6 cells. Tubulin and Ku70 were used as internal controls for loading. (B) A schematic of 2 putative SBEs designated SBE1 and SBE2 and the p53-binding elements in the P2 promoter of HDM2. Reporter assays were conducted using dominant negative Smad4 (dnSmad4) or control vector only (VO) in transient transfection assays. HCT116:3-6 cells were transfected with an SBE2X2 reporter as a positive control or HDM2 P2 reporter. All samples were transfected with renilla to use as an internal control. Cells were treated with vehicle or TGF-β1 (10 ng/ml). Reporter activity was determined relative to renilla to generate relative activity. Fold induction was determined relative to vehicle control and SD was calculated relative to the mean. (C) Western blot of FLAG-Smad3 and Ku70 that bound to the HDM2 promoter. 293T cells were transfected with FLAG-Smad3 and treated with vehicle or TGF-β1 (10 ng/ml). Nuclear extracts were prepared from the latter mentioned cell treatments. Biotinylated PCR fragments of the HDM2 promoter, SBE1, SBE2, Δ_HDM2_, or a positive control, SBE2X2, were mixed with the nuclear extracts. Biotinylated DNA fragments bound to streptavidin beads were used to purify FLAG-Smad3 and Ku70 (internal control).
We next sought to identify the transcription factors responsible for, and the DNA binding elements involved, in HDM2 gene induction by Smads in response to TGF-β1 stimulation. DNA pulldown analyses using full-length and selected truncations of the HDM2 promoter were amplified by PCR with 5′ biotinylated oligos. Biotin-labeled DNA fragments were gel purified, then bound to streptavidin beads, and nuclear extracts from 293T cells transfected with Flag-Smad3 treated with TGF-β1 were mixed. The full-length promoter region of HDM2 purified Flag-Smad3 (Figure 4C). Truncated promoter regions lacking the SBE2 or the Δ element, lacking both SBE sites, did not purify Flag-Smad3. The promoter fragment that contained the SBE2 element purified Flag-Smad3. The SBE2X2 element was used as a positive control. As an internal control, we probed our blots for Ku70. Ku70 is a DNA repair protein that binds to free ends of double-stranded DNA and can be used to normalize nuclear complexes (35). Thus, Smad3 bound to the SBE2 element in the HDM2 promoter region.
Identification of the ability of Smad3 to bind the HDM2 promoter (SBE2) by DNA pulldown assays led to the determination of whether the Smad3/4 complex could bind the genomic SBE elements in the HDM2 promoter. Smad3 and Smad4 ChIP assays were used on HCT116:3-6 cells treated with TGF-β1 to test for the binding to the SBE1 and SBE2 elements in the HDM2 genomic promoter. As a positive control, we performed ChIP on the p21 promoter, a known target of Smads (8) (Figure 5A). The data in Figure 5A show Smad3 and Smad4 bound to the p21 promoter in response to TGF-β1 exposure. Analysis of the HDM2 P2 promoter shows that Smad3 and Smad4 occupied the promoter region that contained the SBE2 element but not the SBE1 element. These data are consistent with the DNA pulldown experiments (Figure 4C). To definitively demonstrate the requirement of the SBE2 element as a downstream target of the TGF-β1 signaling pathway, we used a deletion construct or mutated the _SBE2_-binding element (SBEM2) in the P2 promoter of HDM2 in transient transfection assays with caTβRI. caTβRI was able to induce the P2 promoter independent of the SBE1 element (Figure 5B). In contrast, using the mutated SBEM2 promoter showed a diminished induction of promoter activity in the presence of caTβRI, similar to the vector control. Thus, the HDM2 gene is regulated by TGF-β1–mediated activation of Smad3/4 activation, leading to elevated Hdm2 protein levels.
Binding of Smad3 and Smad4 to the HDM2 SBE2 in vivo. (A) HCT116:3-6 cells were treated with vehicle or TGF-β1 (10 ng/ml) for 48 hours. Smad3, Smad4, or IgG isotype control antibodies were used to immunoprecipitate Smads binding to the HDM2 promoter. Oligo sets were used to PCR amplify the SBE1, SBE2, or the p21 promoter SBE (a positive control for Smad3/4 binding) and resolved on an agarose gel. (B) HCT116:3-6 and 293T cells were transfected with caTβRI and either the full-length P2 _HDM2_-promoter driving luciferase or a deletion construct missing SBE1. Reporter activity was determined relative to renilla to generate relative activity. Fold induction was determined relative to vehicle control, and SD was calculated relative to the mean. (C) Cells were transfected with caTβRI and the P2 HDM2 promoter driving luciferase or a mutant whereby the SBE2 site was mutated. A renilla expression construct was used as an internal control for normalization. SEM was calculated as in B. (D) Wound healing assay using HCT116:3-6 cells, scored in a double-blind assay. Bar graph represents the distance between confluent cells after stimulation with TGF-β1 (10 ng/ml), Nutlin3 (10 μM), or TGF-β1 (10 ng/ml) and Nutlin3 (10 μM). (E) Western blot analysis of p53 in HCT116:3-6 cells treated with TGF-β1 (10 ng/ml) with or without Nutlin3 (10 μM) over a 72-hour time course. SD was calculated relative to the mean.
Biological effects of TGF-β1/HDM2/p53. Clinically, the last event for permissive migration/metastatic nature of a tumor is loss of p53 function. Since induction of Hdm2 by TGF-β1 can decrease p53 levels, we determined whether preventing binding of Hdm2/p53 in cells using Nutlin3 would have a physiological effect. HCT116:3-6 cells were grown to confluency, and then some cells were removed by scraping to produce a “wound.” Wound-healing assays were scored in a double-blind method analyzing several areas of the wound. Treatment of cells with TGF-β1 stimulated the cells to migrate and close the wound, while coincubation with Nutlin3 reversed TGF-β1–mediated migration. Mock or Nutlin3 treatment alone did not dramatically affect cell migration (Figure 5D and Supplemental Figure 1C; supplemental material available online with this article; doi:10.1172/JCI39194DS1). Extracts from HCT116:3-6 cells treated with TGF-β1 alone or in conjunction with Nutlin3 were examined by Western blot to assess levels of p53. As expected, Nutlin3 prevented Hdm2-mediated p53 destabilization (Figure 5E) and prevented cell migration mediated by TGF-β1. These data indicate that Nutlin3 can reverse the effects of TGF-β1–mediated downregulation of p53 by Hdm2.
In response to TGF-β1, cancer cells can change biochemically and transition toward a metastatic phenotype (36). Some epithelial cells may adopt a mesenchymal phenotype, which is termed EMT. Murine mammary epithelial cell line, NMuMg, can undergo EMT in response to TGF-β1 in a cell-culture system (37, 38). We used NMuMg cells to determine whether the TGF-β1/p53/Mdm2 axis was engaged to regulate EMT. As NMuMg cells undergo EMT, there is a morphological change to the cell that causes it to have a “fibroblast-like appearance.” Figure 6A indicates that after 48 hours of treatment with TGF-β1, the cells begin to transition and acquire a mesenchymal phenotype; by 72 hours, the vast majority of these cells resemble a mesenchymal morphology. To confirm molecularly that these cells had undergone EMT, levels of N-cadherin, Snail, and Vinculin, all markers of mesenchyme, were elevated over the same 72-hour simulation with TGF-β1 (Figure 6B). In addition, confocal microscopy showed that E-cadherin was decreased upon exposure to TGF-β1, and Vimentin and N-cadherin were elevated following 24 and 48 hours of TGF-β1 treatment. As predicted, Smad3 was activated in response to TGF-β1 at 24 hours and increased over the 72-hour period (Figure 6C). Mdm2 levels began to increase at 48 hours and dramatically increased after 72 hours of TGF-β1 treatment (Figure 6D). Increased levels of Mdm2 in response to TGF-β1 corresponded with Smad3 activation and with the mesenchymal phenotype in NMuMg cells (Figure 6A).
Characterization of TGF-β1 treatment on cell migration and EMT. (A) NMuMg mammary cells treated with 4 ng/ml of TGF-β1. Photographs were taken at 0, 24, 48, and 72 hours. (B) N-cadherin, Vinculin, Snail, and α-tubulin (loading control) were detected by Western blot of whole-cell lysates from NMuMg cells treated with TGF-β1 (4 ng/ml) over a 72-hour time course. Confocal microscopy of E-cadherin, N-cadherin, and Vimentin in NMuMG cells treated with 4 ng/ml. Original magnification, ×20. (C) Western blot of a time-course activation of p-Smad3 in NMuMg in response to TGF-β1 (4 ng/ml). Loading was normalized to actin. (D) Time-course analysis of Mdm2 induction as measured by Western blot after treatment of NMuMg cells with TGF-β1 (4 ng/ml).
Since Mdm2 was induced upon TGF-β1 treatment and its elevated levels correlated with EMT, we examined one possible mechanism whereby Mdm2 might mediate destabilization of p53, which would be permissive of EMT. If so, then the utilization of Nutlin3 might protect p53 from Mdm2 destabilization. This would reengage p53 activity and possibly affect EMT progression. To test this, NMuMg cells were incubated with vehicle, TGF-β1, Nutlin3, or Nutlin3 and TGF-β1 cells for 48 hours. Treatment with vehicle or Nutlin3 alone did not grossly affect cells after 2 days (Figure 7A). This is consistent with previous reports, as Nutlin3 exposure to these normal mammary epithelial cells does not promote apoptosis and adherent cells are still present (39). While incubation with TGF-β1 alone induced the formation of mesenchymal cells, the combination of Nutlin3 and TGF-β1 resulted in significantly fewer mesenchymal cells. The remaining cells resembled mesenchymal cells, yet did not appear healthy. We confirmed the presence of mesenchymal cells by detecting Vinculin and N-cadherin by Western blot of cells treated with vehicle, TGF-β1, Nutlin3, or TGF-β1 and Nutlin3 together after 24 and 48 hours (Figure 7B). Western blot analysis of p53 levels showed that, as predicted, Nutlin3 rescued TGF-β1–mediated destabilization of p53 (Figure 7B).
Induction of apoptosis in response to TGF-β1 and Nutlin3 exposure. (A) Morphological changes to NMuMg cells after incubation with 4 ng/ml TGF-β1, Nutlin3 (10 μM), or TGF-β1 and Nutlin3. Original magnification, ×20. (B) Western blot of p53, N-cadherin, Vinculin, and Tubulin under the same treatment conditions as A. (C) Survival assay of NMuMg cells. Cell were treated as in A and stained with methylene blue at day 5. Methylene blue was liberated from the cells and quantified by OD595. Percentage of cell death was calculated from control; data represent 3 independent experiments completed in triplicate. SD was calculated relative to the mean. (D) Flow cytometry of NMuMg cells treated as in A and stained for annexin V and PI at 24 and 48 hours. Data represent triplicates of experiments. White bars represent PI negative and annexin V positive, and black represents PI and annexin V positive. SD was calculated relative to the mean. (E) Immunohistochemistry of Hdm2 and p-Smad3 (serine 425) in normal breast tissue, ductal carcinoma, and lobular carcinoma. Original magnification, ×16.
The observation that fewer cells were present with treatment of TGF-β1 and Nutlin3 suggested that cells were either cytostatic or undergoing cell death. It did not appear that cells were cytostatic, as the cells treated with Nutlin3 or TGF-β1 alone did not demonstrate the same outcome as the combination of both TGF-β1 and Nutlin3. To determine the long-term survival effects of Nutlin3 and TGF-β1 cotreatment on NMuMg cells, a colony-forming assay was used. As shown in Figure 7C, staining of cells cotreated with TGF-β1 and Nutlin3 resulted in a 79% decrease in colonies relative to control. In contrast, Nutlin3 or TGF-β1 treatment alone caused a 21% and 58% decrease in colonies, respectively. Control and Nutlin3-treated cells were not statistically different, while treatment with TGF-β1 alone, or in conjunction with Nutlin3, was statistically different as analyzed from 3 independent experiments performed in triplicate (Figure 7C). The data generated in these assays suggest that TGF-β1 alone, and in conjunction with Nutlin3, has an affect on the viability and/or proliferation of cells.
The limitation of colony forming assays is that one cannot determine whether the cells are cytostatic or undergoing apoptosis in the presence of TGF-β1 for 7 days. To address this question, we performed a short-term study by utilizing flow cytometry analysis of annexin V and propidium iodide (PI) staining to examine the mechanism of cell death. This assay allowed for the segregation of apoptosis versus other cell death mechanisms at the early stages of programmed cell death. NMuMg cells were treated with vehicle, TGF-β1, Nutlin3, or TGF-β1 and Nutlin3 for 24 and 48 hours. In Figure 7D, early (PI negative/annexin V positive) and late apoptotic/necrotic (PI positive/annexin V positive) cells were detected at 24 hours. Similar levels of apoptotic cells were observed in control and Nutlin3-exposed cells. At 48 hours, while TGF-β1 alone induced low levels of apoptotic cells, the combination of TGF-β1 and Nutlin3 resulted in a much more dramatic induction of apoptosis. These data show that Nutlin3 alone had a minimal effect on cell survival, which is consistent with other reports using various cell lines (30, 39, 40). Moreover, although TGF-β1 treatment caused cytostatic responses as well as cell death, the combination of TGF-β1 and Nutlin3 was more effective at inducing cell death than either agent alone.
Further experiments were conducted to determine whether knockdown of mdm2 in NMuMG cells via siRNA would confer results similar to those of Nutlin3. Earlier studies have shown that knockdown of mdm2 in proliferating cells is lethal (41–43). Furthermore, Mdm2–/– mice are embryonic lethal (12, 13). As predicted, knockdown of mdm2 significantly decreased proliferation and quickly caused cell death in our NMuMG cell system (Supplemental Figure 1, A and B). This prevented us from approaching the question of Mdm2 dependence in regard to TGF-β1–induced EMT through a genetic approach. The use of an Mdm2 inhibitor induced much less stress, as shown by the low percentage of cell death (Figure 7C) and minimal apoptosis in response to Nutlin3 (Figure 7D).
Clinical correlation of Smad3 activation and HDM2 levels. Since we mechanistically showed that TGF-β1 activation of Smad3 induced HDM2 gene expression, we next determined whether there was a clinical correlation between Smad3 activation and detection of Hdm2 in human breast cancer patient samples. We examined a total of 45 patient samples: (a) normal tissue from breast reductions; (b) benign tissues; and (c) ductal and lobular carcinomas. We examined the activated Smad3 and Hdm2 levels in tumor tissue and normal tissue (Figure 7E). We found that 65% of the samples were positive for both Smad3 activation and elevated levels of Hdm2 (P = 0.0014) (Table 1). Furthermore, 73% of ductal or lobular carcinomas that presented as extensions to either regional lymph nodes or metastasis to other organ systems costained for Hdm2 and activated Smad3. We did not detect staining of Hdm2 or activated Smad3 in the surrounding stroma or normal epithelia cells, showing the specificity of the detection of these proteins in breast carcinomas. In normal breast tissue, we observed modest staining for Hdm2 in the ducts. This staining may reflect some ductal hyperplasia, which is often observed. In addition, since Hdm2 may be regulated by the estrogen receptor, we stratified the ductal carcinoma samples into estrogen and progesterone receptor positive (ER+/PR+) or negative (ER–/PR–) expression (when the annotation was available). There was no significant difference in ER/PR status and Hdm2 levels (P = 0.26) (Table 1). Thus, selective costaining of activated Smad3 and elevated Hdm2 levels in infiltrative and metastatic breast cancer patient samples strongly supports a role for an activated TGF-β1/Smad3/Hdm2 pathway in metastatic breast cancer.
Clinical samples stained for Hdm2 and p-Smad3