DAXX drives de novo lipogenesis and contributes to tumorigenesis

Cancer cells exhibit elevated lipid synthesis. In breast and other cancer types, genes involved in lipid production are highly upregulated, but the mechanisms that control their expression remain poorly understood. Using integrated transcriptomic, lipidomic, and molecular studies, here we report that DAXX is a regulator of oncogenic lipogenesis. DAXX depletion attenuates, while its overexpression enhances, lipogenic gene expression, lipogenesis, and tumor growth. Mechanistically, DAXX interacts with SREBP1 and SREBP2 and activates SREBP-mediated transcription. DAXX associates with lipogenic gene promoters through SREBPs. Underscoring the critical roles for the DAXX-SREBP interaction for lipogenesis, SREBP2 knockdown attenuates tumor growth in cells with DAXX overexpression, and DAXX mutants unable to bind SREBP1/2 have weakened activity in promoting lipogenesis and tumor growth. Remarkably, a DAXX mutant deficient of SUMO-binding fails to activate SREBP1/2 and lipogenesis due to impaired SREBP binding and chromatin recruitment and is defective of stimulating tumorigenesis. Hence, DAXX’s SUMO-binding activity is critical to oncogenic lipogenesis. Notably, a peptide corresponding to DAXX’s C-terminal SUMO-interacting motif (SIM2) is cell-membrane permeable, disrupts the DAXX-SREBP1/2 interactions, and inhibits lipogenesis and tumor growth. These results establish DAXX as a regulator of lipogenesis and a potential therapeutic target for cancer therapy.


Fig. S1. DAXX and lipogenic genes are highly expressed in breast cancer.
(a) A gene expression heatmap of the indicated genes in normal breast tissues and breast cancer samples based on a TCGA breast cancer dataset (TCGA, Nature, 2012;PMID: 23000897).
(b) The mRNA expression levels for DAXX and SREBP2 in normal breast tissues and tumors of the indicated breast cancer subtypes based on an analysis of the TCGA breast cancer dataset as in a.
(c) Cholesterol levels in normal control and breast tumors of the indicated subtypes based on a published dataset (Tang X. et al., 2014;PMID: 25091696). Data are presented as mean values ± SEM.
(d) Genomic features of DAXX and the indicated genes in the lipid metabolism pathways. The percentages of breast tumor samples with the gain (mRNA upregulation or copy number gains), and the loss of gene functions (mRNA downregulation or copy number losses) are shown.
(e) A heatmap of relative mRNA levels of DAXX along with select genes in the lipid metabolism pathways. The red and blue groups refer to a high level (red) or low level (blue) of mRNA expression of the indicated genes according to combined expression scores in an individual tumor sample.
(f) A Kaplan-Meier plot of the correlation between gene expression levels of the selected genes in panel e (the red and blue groups) and patient survival time based on data of TNBC patients in the TCGA BC datasets.
(g) A Kaplan-Meier plot of TNBC patient survival based on DAXX mRNA levels in the METABRIC dataset.
(h) Kaplan-Meier plots of recurrence-free survival (RFS) for TNBC patients based on the DAXX mRNA expression levels using two distinct microarray probes. The "best cutoff" option was selected in the analysis. The plot is generated at kmplot.com.

Fig. S2. DAXX and lipogenic genes are upregulated in BC brain metastases.
The mRNA levels of the indicated genes in the lipogenesis pathways were analyzed based on the microarray dataset GSE14020. The p-values were obtained via the two-sided student-t test. The box of the box plot represents the interquartile range (IQR) --the range of the middle 50% of the data. The top of the box represents the upper quartile (75th percentile) of the data, and the bottom of the box represents the lower (25th percentile quartile) of the data. The center line represents the median value of the data set. The upper whisker is either the maximum value in the data set or the value that is 1.5 times the IQR above the upper quartile, whichever is lower. The lower whisker is either the minimum value in the data set or the value that is 1.5 times the IQR below the lower quartile, whichever is higher. If there are any values in the data set that fall outside the range between the lower whisker and the upper whisker, they will be displayed as individual points in the box plot. (a) Gene set enrichment analyses (GSEA) of cholesterol biosynthesis genes downregulated due to DAXX KD or DSM OE (left and right) or upregulated due to DAXX WT OE in (center) MDA-MB-231 cells. The KEGG cholesterol biosynthesis geneset was used for the GSEA plots. GSEA defines an enrichment score (ES) via a weighted Kolmogorov-Smirnov statistic, and determines p-value of the ES using an empirical phenotype-based permutation test procedure.
(b) MDA-MB-231 cells stably transfected with a control vector (Control), a DAXX shRNA (KD), the wt DAXX, or the DSM mutant expression vector were cultured in the presence of serum or serum-starved for 24 hours. The cells were then fixed and stained with an anti-FASN polyclonal antibody (red) and counter stained with DAPI for visualizing nuclei (blue). The cells were imaged using a fluorescence microscope. All images were captured with the same duration of light exposure for the red or blue channel.
(c) Immunoblotting analysis of cell extracts of the four MDA-MB-231-derived cell lines with antibodies against the indicated proteins.
(d) Immunoblotting analysis of cell extracts of the H1299 cells expressing tetracycline (Tet)-inducible wt DAXX in the presence of control (DMSO) or Tet with antibodies against the indicated proteins.   (a & b) MDA-MB-231, MDA-MB-468 (breast cancer) and HCT116 (colon cancer) cells were fractionated into cytoplasmic and nuclear fractions, which were analyzed in immunoblotting for SREBP2 (Cayman antibody) and SREBP1 (ProteinTech antibody). Note that the mature form (M) of SREBP2 (~55 kDa) and SREBP1 (between 55 and 70 kDa) were predominantly detected in the nuclear fraction. FL: the full-length (precursor) form of SREBP1/2.
(c) Whole cell lysates of the prostate cancer cell line R1-AD1 (a subline of CWR-R1) and breast cancer cell line Hs578T were subjected to immunoblotting using the Cayman antibody as above. Two larger bands of SREBP2 (likely the precursors, denoted with • and ••, respectively) were detected. FL: full-length (precursor) SREBP2. M: mature SREBP2.
(b) Mature FLAG-SREBP1a (Addgene 26801) or SREBP2 (Addgene 26807) were co-transfected with the indicated GFP-DAXX constructs in 293T cells. The transfected cells were treated with 5 µM Z-VAD. At 24h after transfection, the total lysates of transfected cells were subjected to IP with the anti-FLAG mAb (M2, Millipore-Sigma F1804). Immunoblotting of the immunoprecipitates and the total cell extracts (input) was conducted using a rabbit anti-DAXX (Bethyl Laboratories, A301-352A) or anti-FLAG antibody (Cell Signaling Technology, #14793). The GFP-DAXX and endogenous DAXX are denoted.

Fig. S7. Colocalization of DAXX and SREBP1.
(a and b) DAXX promotes SREBP1 nuclear translocation. MDA-MB-231 cells in medium with 10% serum (a) or without serum (b) were transfected with GFP-DAXX (WT). At 24 hour after transfection, cells were fixed and stained with an anti-SREBP1 antibody (ProteinTech catalog # 14088-1-AP). Colocalization of GFP-DAXX and SREBP1 were denoted with arrows in the indicated area of interest (ROI) in two representative transfected cells. In panel a, cells with high levels of perinuclear SREBP1 signals are pointed with yellow arrows. The nuclear to cytoplasmic signal intensity (N/C) ratio in cells expressing GFP-DAXX (GFP-DAXX+) vs. cells with no GFP-DAXX (GFP-DAXX-) in panel a were calculated based on relative gray levels measured by the Photoshop software. Data are presented as mean values ± SEM (n=14 different cells). The p value is based on unpaired two-tailed t-test. Microscopic scale bar: 10 µm.
(c) Colocalization of endogenous DAXX and SREBP1 in MDA-MB-231 cells in medium containing fetal bovine serum depleted of lipids. Colocalization of DAXX and SREBP1 in the nucleus (white arrows) and cytoplasm (yellow arrows) are denoted in the indicated ROI (scale bar: 10 µm in main image and 2 µm in ROI).
(d) DAXX overexpression increases levels of mature SREBP1. MDA-MB-231 cells with a control vector, WT DAXX cDNA, and SREBP1 shRNA were subjected to Western blotting with the indicated antibodies. The same anti-SREBP1 antibody (ProteinTech catalog # 14088-1-AP) was used for both IF and WB assays. (d) DAXX knockdown reduces acetate-dependent lipogenesis in cancer cells. The luminal estrogen receptorexpressing BC cell lines MCF7 and T47D and the colon cancer cell line HCT116 with a control vector or a DAXX-specific shRNA were subject to de novo lipogenesis assay using [ 14 C] acetate labeling. The radioactivity counts were normalized against total protein levels. Shown are average values (± SEM, n=3 biologically independent samples). The p values were based on two-tailed unequal variance Student's t-test.   (c) The mouse breast cancer E0771 cells stably transduced with a control, or a mouse Daxx shRNA were subjected to RT-qPCR (left). The cells were engrafted into the mammary fat pads of female C57BL/6 mice (n=4). The final tumor weights are plotted (middle). The images of dissected tumors are shown (right).
(d) The mouse prostate cancer TRAMP-C2 cells stably transduced with a control, or a mouse Daxx shRNA were subjected to RT-qPCR (left). The cells were engrafted into male NSG mice subcutaneously (n=4 or 5 tumor-bearing mice) as in (b). The final tumor weights are plotted (middle). The images of dissected tumors are shown (right).
(j) SIM2 inhibits lipogenic gene expression. MDA-MB-231 control and DAXX OE cell lines were untreated (control) or treated with SIM2 (10 µM) (n=3 biologically independent samples). At 24h after treatment, RNA was isolated and subject to qRT-PCR. RNA levels were normalized against that of ACTB. The p values shown are based on t-test.
All data are presented as mean values ± SEM, and the p value are calculated based on unpaired two-tailed t-test.