DTX1 physically interacts with ICN1 and induces ubiquitination and proteasome-mediated degradation of ICN1. (A) COL cells transduced with DTX1 or empty vector were treated with 5mM EDTA for 15 minutes to increase NOTCH1 cleavage and total ICN1 level. Immunoprecipitation (IP) of DTX1 from DTX1-transduced cells yielded both DTX1 and ICN1, as shown by immunoblot (IB). (B) Endogenous DTX1 was knocked down with either of 2 specific shRNA constructs (1) or (2); Immunoblot (IB) shows ICN1, HES1 and DTX1 protein from shRNA transduced cells compared to vector (V). (C) Histograms represent matrigel invasion of the COL cells shown in figure 1B. 1×10⁴ COL cells with DTX1 shRNAs were seeded in each transwell and invasion was quantified after 48 hours; histograms shown the mean of three wells, and error bars show standard deviation. (*, p<0.05). (D) HA-ICN1 or DTX1, alone or in combination, were transiently overexpressed in 293T cells for 48 hours. After treatment with proteasome inhibitor PS341 (0.1μM) for 6 hours or no treatment, total cell lysates were prepared. Immunoblots (IB) of ICN1, DTX1 and actin in the whole cell extracts are shown (upper panels). After immunoprecipitation of ICN1 with anti-HA beads, ubiquitination of ICN1 was measured using anti-ubiquitin antibody (lower panel). Ubiquitination level of ICN1 was quantified by densitometry (bottom panel); histograms represent relative ubiquitination level.

Dominant negative HES1 upregulates DTX1 expression. (A) The HES1-AB reporter is activated by CSL/NOTCH and inhibited by HES1. 293T cells were cotransfected with HES1-AB reporter (firefly luciferase) and a Renilla luciferase control together with various dosages of HES1 and/or dnHES1 as shown in the figure. Relative luciferase activity was measured 48 hours after transfection. (**, p<0.005) (B) PCR and (C) quantitative PCR analysis of changes in HASH1 and DTX1 expression induced by dnHES1. Osteosarcoma cells OS 187, COL and KRIB were transduced with empty vector (V) or dnHES1 (dnH). Gel image (B) shows the PCR amplification of HASH1, DTX1 and actin; histograms (C) represent relative Q-PCR results, normalized to a value of 1 for the vector-only cells. (D) Histograms represent matrigel invasion of the same cells described in 4B and 4C. Analysis was performed as in figure 1E. (**, p<0.005. *, p<0.05). V: empty vector control. dnH: dnHES1.

HES1 suppression increases DTX1 promoter activity through the N-box motif. (A) Immunoblot (IB) shows the impact of shRNA to HES1 or of dnHES1 on DTX1 and HES1 expression in OS187. Left panel: V indicates vector control; 5, 6, 7 and 8 indicate four distinct shRNAs specific for HES1. Actin serves as a loading control. Right panel: V indicates vector control; dhH indicates dominant negative HES1. Anti-HES1 antibody can detect both HES1 and dnHES1. (B) Wild type (WT) or N-box mutated (mutN) DTX1 promoter reporter construct (Renilla luciferase) was cotransfected with a firefly luciferase control into OS187 cells stably expressing empty vector, shHES1 or dnHES1. Relative luciferase activity was determined after 48-hr incubation. (*, p<0.05. **, p<0.005). (C) Histograms represent matrigel invasion of OS187 cells stably expressing empty vector or HES1-specific shRNA (clone 5 and 6). Analysis was performed as in figure 1E (*, p<0.05).

DTX1 negatively regulates NOTCH1/HES1 signaling and suppresses invasion of osteosarcoma cells. Immunoblot (A) and PCR (B) analysis of DTX1, NOTCH1/ICN1, HES1 and actin in OS187 cells stably expressing vector control or DTX1 gene. To confirm that the band identified by the western blot for ICN1 in fact represents the cleaved intracellular fragment of Notch1, we cultured OS 187 cells in 10 or 50 μM compound E, a gamma secretase inhibitor (GSI), and showed that the cleaved protein is reduced with GSI treatment (Supplemental figure 1). (C) Histograms represent quantitative PCR analysis of NOTCH1, HES1 and DTX1 from the same cells. C_t values were normalized to actin; vector control is defined as a relative expression of 1 (*, p<0.05). (D) Histograms represent luciferase emission from a CSL reporter (i.e., NOTCH-responsive) in OS 187 cells transduced with DTX1 or empty vector (*, p<0.05). (E) Histograms represent matrigel invasion of OS187 cells stably expressing empty vector, DTX1 or DTX1 and HES1. 1×10⁴ stably transduced OS187 cells were seeded in each transwell and cell invasiveness was quantified by the invasive cell number after 48 hours incubation. Average of three wells, with error bars showing standard deviation, is shown (*, p<0.05).

Impact of manipulation of NOTCH/HES1 signaling on DTX1 expression. (A) PCR (left panel) and western blot (right panel) analysis of HES1, DTX1 and actin expression in OS187 cells overexpressing ICN1, dnMAM or HES1 compared to vector control. (B) Quantitative PCR analysis of HES1 and DTX1 in the same cells at 3A. Analysis was performed as in figure 1C. (C) COL cells, which have high endogenous DTX1, were manipulated and analyzed as in 3A. Left panel: PCR analysis. Right panel: western blot analysis. (D) Histograms represent matrigel invasion of the COL cells from 3C (*, p<0.05). Assay and analysis were performed as in figure 1E.

HES1 actively represses the DTX1 through binding to an N-box motif in the DTX1 promoter. (A) Schematic depiction of the 1kb DTX1 proximal promoter in the 5′ untranslated region. One potential HES1 binding site (N-box) was located at -400bp. Arrows represent the position of the PCR primers utilized for the ChIP analysis of the following panels (upper part). Sequence alignment of the DTX1 genes of human, dog and mouse showed the identical N-box in the similar position (lower part). (B) DTX1 promoter reporter (Renilla luciferase) was cotransfected with a firefly luciferase control into 293T cells transiently expressing vector control, full length HES1, a mutant HES1 with the C-terminus deleted (HES1ΔC) or HES1-WRPW mutant (HES1m), which is predicted to have less Groucho/TLE binding. Relative DTX1 promoter activity was determined after 48 hours. (**, p<0.005). (C) Gel mobility-shift analysis showing HES1 binding to the N-box motif in the DTX1 promoter. Nuclear protein extract (3μg) from 293T cells expressing FLAG-HES1 were incubated with biotin labeled probes containing WT N-box (W), mutant N-box (M), and anti-FLAG antibody, either alone or in combination, at room temperature for 15 min. As cold competitors, excess doses (10-fold) of non-biotinylated WT or mutant oligomers were included. Arrows indicate the specific N-box DNA-HES1 complex and the HES1 supershifted complex. (D) Chromatin immunoprecipitation (ChIP) assay: COL cells were transduced with FLAG-tagged HES1. Immunoblot (IB) confirmed expression of HES1 and effective immunoprecipitation (IP) of HES1 with anti-FLAG antibody (upper panels). Sheared chromatin from transfected cells was immunoprecipitated with anti-FLAG antibody or nonspecific IgG as control. Immunoprecipitated DNA was used as template in PCR for HASH1 promoter region (positive control), HASH1 3′ end region (negative control) or DTX1 promoter region (lower panels).

Expression of ICN1, HES1 and DTX1 in osteoblasts and osteosarcoma cells, and model for the regulation of NOTCH signaling by NOTCH target genes HES1 and DTX1. (A) Immunoblots demonstrate expression of ICN1, HES1 and DTX1 in normal human osteoblasts [HOSB] and a panel of human osteosarcoma cell lines. (B) Model: NOTCH1 is cleaved into ICN1 by γ–secretase, and then activates downstream target genes HES1 and DTX1. Inhibitors of γ-secretase can block this step. DTX1 causes negative feedback of ICN1 by targeting ICN1 for ubiquitination and proteasome-mediated degradation. Meanwhile, HES1 causes positive feedback of ICN1 by repressing DTX1 gene expression. HES1 promotes the invasion and metastasis of osteosarcoma. In osteosarcoma cells expressing high HES1 and low DTX1, NOTCH inhibition, such as that caused by GSI, may downregulate HES1, increasing DTX1, which further downregulates ICN1 through ubiquitination, reducing HES1-mediated invasiveness. However, in osteosarcoma cells with high DTX1 and low HES1, NOTCH inhibition may not reduce invasiveness. In fact, reducing DTX1 may stabilize low level ICN1 expression, increasing ICN1 signaling and HES1 expression, causing increased invasion, as happens with COL. Thus the impact of GSI on NOTCH signaling should take into account expression of both HES1 and DTX1.