Tumor cell‐specific inhibition of MYC function using small molecule inhibitors of the HUWE1 ubiquitin ligase

1With over 660,000 new cases each year, colorectal cancer is the most common gastrointestinal malignancy (Jemal et al, ). Sequence analysis of tumor genomes shows that each tumor harbors multiple mutations that alter the function of central signaling pathways, which control the growth of colon epithelial cells (CancerGenomeAtlasNetwork, ). Predominant among these are the WNT and the RAS signal transduction pathways, and recurrent genetic alterations in both pathways occur in the majority of colorectal cancers (CancerGenomeAtlasNetwork, ). Comparison with gene expression profiling shows that enhanced expression and function of the MYC oncoprotein, a downstream effector of both WNT- and RAS-dependent signal transduction, is a common denominator of a vast majority of colon tumors (CancerGenomeAtlasNetwork,; van de Wetering et al, ). Deletion of MYC ablates tumorigenesis in mouse models of colorectal cancer demonstrating that MYC function is essential for colorectal tumorigenesis (Sansom et al, ).

2MYC is a transcription factor that establishes a gene expression program characteristic of colon epithelial stem cells (Dang,; van de Wetering et al, ). Like many other transcription factors, MYC is rapidly turned over by the ubiquitin/proteasome system (Gregory & Hann,; Welcker et al, ). At least three ubiquitin ligases, SKP2, HUWE1 (HECTH9/MULE/ARF-BP1), and FBXO28, are also required for MYC function (Adhikary et al,; Cepeda et al,; Kim et al,; von der Lehr et al, ). The use of mutants of MYC, in which groups of lysines have been replaced by arginines, suggests that MYC itself needs to be ubiquitinated and possibly degraded to regulate transcription (Adhikary et al,; Zhang et al, ). It is also possible that MYC recruits ubiquitin ligases to degrade repressor proteins that antagonize MYC function. One example for this is the identification of HDAC2, a histone deacetylase that associates with the MXD/MAD complex (Laherty et al, ) as a substrate for the HUWE1 ligase (Zhang et al, ).

3In human and mouse tumor cells, MYC binds to target promoters either as part of a binary activating complex with a partner protein, MAX, or as a ternary repressive complex that contains in addition the zinc finger protein MIZ1; the balance of both complexes at each promoter determines the transcriptional response to MYC (Eilers & Eisenman,; Walz et al, ). HUWE1 associates with MYC, the related N-MYC protein, and with MIZ1 and ubiquitinates all three proteins (Adhikary et al,; Li et al,; Yang et al, ). Ubiquitination by HUWE1 degrades both N-MYC and MIZ1 and restricts N-MYC function in vivo (Zhao et al, ). In contrast, HUWE1 has only weak effects on the turnover of MYC (Adhikary et al,; Zhao et al, ). One possibility is that HUWE1 assembles K63-linked ubiquitin chains on MYC that do not target the protein for degradation (Adhikary et al, ). Alternatively, lysines that are critical for degradation of MYC may be buried (e.g., by complex formation with MIZ1) and therefore inaccessible to HUWE1 in vivo (Kim et al, ).

4The effects of HUWE1 depletion or deletion on the function of MYC proteins have been characterized in several biological contexts: Deletion of HUWE1 enhances N-MYC levels in embryonic stem cells cultured in the absence of leukemia inhibitory factor and expands neuronal cell population in the cerebellum (D'Arca et al,; Zhao et al, ). In skin papillomas, deletion of HUWE1 stabilizes MIZ1 and enhances repression of Cdkn2b (p15INK4b) and Cdkn1a (p21CIP1) by the MYC/MIZ1 complex, correlating with enhanced tumorigenesis (Inoue et al, ). In contrast, depletion of HUWE1 in several human tumor cells arrests proliferation and inhibits expression of MYC-activated target genes (Adhikary et al, ). Here, we used both shRNAs and small molecule inhibitors to explore the role of HUWE1 as a regulator of MYC function in human colon cancer cells. The aim of this study was to test the hypothesis that inhibition of HUWE1 is feasible to control MYC activity in this tumor entity and to explore the underlying mechanistic basis.

5To test whether HUWE1 is required for growth of colon cancer cells in culture, we generated retroviruses that constitutively express shRNAs targeting HUWE1. We identified two shRNA sequences that strongly reduced expression of HUWE1 as determined by RQ-PCR and immunoblotting of lysates of infected cells (Fig 1A, left and middle panel). Expression of either shRNA suppressed the clonogenic growth of Ls174T colon carcinoma cells, which depend on MYC for proliferation (van de Wetering et al, ), relative to control-infected cells (Fig 1A, right panel). Depletion of HUWE1 also suppressed the growth of three other colon carcinoma cell lines that we tested (Supplementary Fig S1A and B). To rule out that the effects on colony formation were due to variations in infection efficiency, we generated Ls174T cells that express one of the two shRNAs targeting HUWE1 in a doxycycline-inducible manner (Fig1B). Consistent with the results obtained upon constitutive expression of shHUWE1, addition of doxycycline led to a depletion of HUWE1 and strongly suppressed colony formation (Fig1C). These results were confirmed with a second doxycycline-regulated shRNA targeting HUWE1 (shHUWE1-3) (Fig1B and C).

6We injected Ls174T cells expressing both inducible shRNAs targeting HUWE1 subcutaneously in immunocompromised mice. Addition of doxycycline to the drinking water suppressed expression of HUWE1 and strongly decreased BrdU incorporation in subcutaneous tumors (Fig 1D). Addition of doxycycline retarded (shHUWE1-2; Supplementary Fig S2A) or abrogated (shHUWE1-3) subcutaneous tumor growth (Fig1E and Supplementary Fig S2B). We also tested the effect of depletion of HUWE1 in an orthotopic setting, in which Ls174T cells stably expressing luciferase and doxycycline-inducible shHUWE1-3 were transplanted into the cecum of immunocompromised mice. Tumor growth was monitored twice a week by luciferase-based in vivo imaging. Out of 12 grafted mice, six developed a primary tumor in the colon. Half of these mice were left untreated, resulting in outgrowth of the primary tumor and their subsequent dissemination to the peritoneum, lymph nodes, liver, and lung. Addition of doxycycline strongly suppressed the growth of tumors in this orthotopic setting (note the logarithmic scale) and suppressed the formation of metastases (Fig1F; data for individual mice are shown in Supplementary Fig S2C). We concluded that HUWE1 is required for growth and tumor formation of human colon cancer cells.

7To understand the mechanisms underlying these observations, we isolated RNA from pools of Ls174T cells stably expressing shRNA targeting HUWE1. Immunoblots showed that depletion of HUWE1 had no significant effect on steady-state levels of MYC (Fig2A), consistent with previous observations (Adhikary et al, ). RQ–PCR assays revealed that constitutive depletion of HUWE1 downregulated multiple genes that are activated by MYC relative to control-infected cells (Fig2B). In contrast, knockdown of HUWE1 had only marginal effects on expression of a control gene (ACTB) or on expression of two genes, CDKN1A and GADD45A, that are repressed by MYC (Fig2B). Depletion of HUWE1 also strongly increased expression of MUC2, a gene that is induced during terminal differentiation of colon epithelial cells (Fig2B); this is consistent with the observation that MYC suppresses differentiation of Ls174T cells (van de Wetering et al, ). Virtually identical changes in gene expression were observed in Ls174T cells expressing a doxycycline-inducible shRNA targeting HUWE1 (Fig2C). To explore the changes in gene expression elicited by depletion of HUWE1 in an unbiased manner, we performed microarray analyses and found that depletion of HUWE1 led to upregulation of 492 and downregulation of 250 genes, respectively (cut-off: fold change 2; P < 0.01). Gene set enrichment analysis (GSEA) showed that multiple sets of MYC-activated genes were downregulated upon depletion of HUWE1, and this was statistically highly significant (Fig2D, upper panel) (Subramanian et al, ). In contrast, sets of MYC-repressed genes were not significantly affected, arguing that HUWE1 is required for activation, but not repression by MYC (Fig2D, lower panel). The sets of MYC-activated genes include many target genes of MYC, like ODC1 or HSPE1, that are not regulated in a cell cycle-dependent manner, demonstrating that downregulation of these genes is not an indirect consequence of the cell cycle arrest induced by depletion of HUWE1. HUWE1 also ubiquitinates p53 and promotes the degradation of p53 (Chen et al, ), but depletion of HUWE1 had no significant effects on steady-state levels of p53 or phosphorylation of ATM or CHK2, in contrast to cells exposed to DNA damage (Fig2E). Furthermore, shRNA-mediated depletion of p53 did not alleviate the effects of shHUWE1 on expression of MYC target genes (Supplementary Fig S3A and B). We concluded that HUWE1 is required for activation, but not repression, of MYC target genes in colon carcinoma cells.

8To identify potential inhibitors of HUWE1, we configured an in vitro assay of HUWE1 activity for high-throughput screening of small molecules, exploiting the fact that the HECT-domain of HUWE1 auto-ubiquitinates (Pandya et al, ). Briefly, the HECT-domain of HUWE1 was biotin-tagged, attached to streptavidin-coated 96-well plates, and incubated with UBA1, UbcH5b, a ubiquitin-conjugating enzyme that supports HUWE1 activity in vitro (Adhikary et al, ), ATP, and MYC-tagged ubiquitin. Auto-ubiquitination was detected using a europium-labeled MYC antibody and appropriate detection reagents (Fig3A). Screening of 840,243 compounds resulted in 2,765 hits that inhibited HUWE1 activity, yielding an initial hit rate of 0.33%. After hit confirmation in repeat experiments, we identified inhibitors of UBA1 or UbcH5b by a thioester assay that measures the covalent binding of UbcH5b to ubiquitin in the presence of UBA1: Compounds active in this assay were eliminated. Furthermore, dose responses were determined in both HUWE1 and NEDD4 auto-ubiquitination assays, and compounds inhibiting NEDD4 auto-ubiquitination were eliminated (Fig3B and unpublished observations). From these experiments, we selected two compounds (BI8622 and BI8626) that inhibited HUWE1 with IC50 values of 3.1 μM (BI8622) and 0.9 μM (BI8626), respectively (Fig3B and C). HECT-domains of multiple ligases auto-ubiquitinate in in vitro assays containing both UBA1 and UbcH5b (M. Gmachl, unpublished observation). These assays were used to analyze the specificity of the identified inhibitors. We found that neither compound inhibited the activity of other HECT-domain ubiquitin ligases in these assays, arguing that they are specific inhibitors of HUWE1 (Fig3C). Attempts to co-crystallize compound/HUWE1 complexes failed due to the very high solubility of the HECT-domain of HUWE1 (M. Gmachl, unpublished observation).

9To test the efficacy of both compounds in tissue culture, we initially confirmed observations that HUWE1 ubiquitinates and degrades MCL1 in response to DNA damage (Zhong et al, ). Consistent with these published data, steady-state levels of MCL1 decreased rapidly upon UV irradiation of U2OS cells and depletion of HUWE1 both enhanced steady-state levels of MCL1 and retarded the decrease upon UV irradiation (Supplementary Fig S4A). Furthermore, ectopically expressed HUWE1 ubiquitinated MCL1 (Supplementary Fig S4B) and ubiquitination of MCL1 increased after UV irradiation (Supplementary Fig S4C) in a HUWE1-dependent manner (Supplementary Fig S4D). These data confirmed that degradation of MCL1 upon UV irradiation is a valid read-out of HUWE1 activity. Incubation of HeLa cells with either HUWE1 inhibitor abolished ubiquitination of MCL1 induced by ectopically expressed HUWE1 (Fig3D). For BI8622, we performed multiple ubiquitination assays in the presence of different concentrations of inhibitor; this yielded an IC50 value of 6.8 μM, similar to that obtained in vitro (Supplementary Fig S4E). Both compounds retarded the degradation of MCL1 in response to UV irradiation to the same extent as depletion of HUWE1 (Fig3E). Furthermore, both compounds induced accumulation of TopBP1 (Fig3F), another substrate of HUWE1 (Herold et al, ). Consistent with the observed effects in response to depletion of HUWE1, neither compound led to an accumulation of p53 in Ls174T cells (Supplementary Fig S5A). We concluded that both BI8622 and BI8626 inhibit HUWE1 in cells.

10Incubation with HUWE1 inhibitors suppressed colony formation of Ls174T cells with estimated IC50 values of 8.4 μM (BI8622) and 0.7 μM (BI8626), respectively, which is in good agreement with the IC50 values for inhibition of HUWE1 activity (Fig4A). Both inhibitors also suppressed the growth of three additional colon carcinoma cell lines that we tested with minor variations in sensitivity (Supplementary Fig S5B). Partial shRNA-mediated depletion of HUWE1 led to a moderate, but highly significant reduction in IC50 value for growth inhibition by BI8622, arguing that growth inhibition is an on-target effect of the inhibitor (Supplementary Fig S6). FACS assays of propidium iodide-stained cells showed that both compounds led to a decrease in the percentage of cells in the S and G2 phases and an accumulation of Ls174T cells in the G1 phase of the cell cycle, but had only marginal effects on apoptosis (Fig4B). Combining these data with a growth curve (Fig4C) enabled us to calculate the length of each phase of the cell cycle (Supplementary Fig S7A). Incubation with either compound retarded passage of Ls174T cells through all phases of the cell cycle, with the effect being strongest for G1. In contrast to colon carcinoma cells, HUWE1 is not required for proliferation of embryonic stem (ES) cells and deletion of HUWE1 instead delays the decrease of N-MYC levels that occurs upon removal of leukemia inhibitory factor (LIF) from the medium (Zhao et al, ). Consistent with these data, chemical inhibition of HUWE1 had marginal effects on the growth and cell cycle distribution of ES cells as well as expression of MYC target genes in the presence of LIF and retarded the decrease in N-MYC levels when LIF was withdrawn (Fig4D–F and Supplementary Fig S7B). Furthermore, inhibition of HUWE1 in crypt cultures of normal intestinal epithelial cells had no effect on expression of MYC target genes (Fig4G). We concluded that both compounds inhibit MYC-dependent transactivation in colon cancer cells, but not in normal intestinal epithelial cells and embryonic stem cells. In vitro assays revealed that both compounds are unstable in the presence of microsomes (Supplementary Fig S7C). Measurements of compound levels in serum after intraperitoneal injection in mice showed that neither compound accumulated to high levels and both were rapidly cleared after injection, precluding a more detailed in vivo analysis of the efficacy of these compounds (Supplementary Fig S7D).

11To test whether the compounds inhibit transactivation of MYC, we infected Ls174T cells with retroviruses expressing either control shRNA or shRNA targeting HUWE1 and incubated pools of stably infected cells with either compound or DMSO as control for 24 h. Both inhibitors reduced the expression of several MYC target genes in control cells, but had no effect in HUWE1-depleted cells (Fig5A). Furthermore, inhibition of HUWE1 resulted in a strong increase in expression of MUC2 (Fig5B). Microarray analyses showed that both compounds led to down- and upregulation of multiple genes (BI8622: 2,267 up, 2,295 down; BI8626: 2,796 up, 2,923 down; cut-off: fold change 2; P < 0.01). Gene set enrichment analysis showed that both compounds downregulated virtually identical gene sets (Fig5C). In contrast, upregulated genes were contained in a much smaller number of previously known gene sets and these differed between compounds (Fig5C). This suggested that the downregulation of genes most likely corresponds to the on-target activity of either chemical. Direct comparisons showed that the effects of both compounds on gene expression were highly similar to depletion of HUWE1 (Fig5D). Importantly, GSEA showed that expression of multiple sets of MYC-activated genes was significantly repressed upon exposure to either compound, whereas MYC-repressed genes were not consistently affected (Fig5E and F). As described above, analysis of candidate genes showed that both inhibitors have little effect on expression of MYC target genes in HUWE1-depleted cells, arguing that their effect is mediated via inhibition of HUWE1 (Fig5A). A genome-wide analysis of gene expression confirmed that BI8622 had no statistically significant effect on expression of any set of MYC-activated target genes in HUWE1-depleted cells, in contrast to its effects in control cells (Fig5G). Finally, shRNA-mediated depletion of p53 (Supplementary Fig S8, upper panel) had no effect on regulation of MYC target genes by either compound (Supplementary Fig S8, lower panel). Collectively, the data establish that these compounds block MYC-dependent transcriptional activation via inhibition of HUWE1 in colon cancer cells.

12Neither depletion nor chemical inhibition of HUWE1 affected the expression of any member of the MXD family of repressor proteins that antagonize transactivation by MYC or nuclear localization of MYC (Supplementary Figs S9A and S10). Furthermore, inhibition of HUWE1 did not inhibit complex formation of MYC with MAX (Supplementary Fig S9B), raising the question how HUWE1 affects MYC-dependent transactivation. Consistent with published data, ectopic expression of HUWE1 strongly decreased steady-state levels of MIZ1 (Fig6A), a zinc finger protein with which MYC forms a repressive complex and a previously identified target of HUWE1 (Inoue et al,; Walz et al,; Wiese et al,; Yang et al, ). In contrast, ectopic expression of HUWE1 had only moderate effects on steady-state levels of MYC, consistent with the results obtained in response to depletion of HUWE1 (Fig6A). As a result, ectopic expression of HUWE1 increased the relative amount of MYC that is not bound to MIZ1 (Fig6B). Notably, HUWE1 also decreased levels of MIZ1 in the presence of a MYC protein devoid of all lysine residues, suggesting that ubiquitination of MYC itself may not strictly be required for its effects on MYC-dependent gene regulation (Fig6C). Inhibition of HUWE1 blocked the HUWE1-mediated decrease in MIZ1 levels; the decrease was also blocked by incubation of cells with an inhibitor of proteasomal degradation, MG132 (Supplementary Fig S11A). Consistently, depletion or chemical inhibition of HUWE1 increased steady-state levels of MIZ1 in colon carcinoma cells (Fig6D and E). Similarly, depletion or inhibition of HUWE1 increased levels of MIZ1 in mouse keratinocytes (Supplementary Fig S11B and C), in which HUWE1 is known to degrade MIZ1 (Inoue et al, ).

13Consistent with the immunoblot data, ChIP-sequencing experiments showed that inhibition of HUWE1 strongly increased the amount of MIZ1 bound to MYC-bound core promoters (Fig6F–H and Supplementary Fig S12). Inhibition of HUWE1 also led to a very small increase in MYC binding, consistent with observations that complex formation with MIZ1 stabilizes MYC (Salghetti et al, ). Depletion or inhibition of HUWE1 induced a significant decrease in acetylation of histone H3 at the HSPE1 promoter, but not at a control (ACTB) promoter (Supplementary Fig S13A and B). In contrast, inhibition of HUWE1 had only weak effects on expression of direct target genes of MIZ1 (Supplementary Fig S13C) and on histone acetylation at a directly MIZ1-bound promoter (VAMP4) (Supplementary Fig S13B). The data argue that inhibition of HUWE1 shifts the balance on MYC target genes toward repressive MYC/MAX/MIZ1 complexes thereby inhibiting transactivation by MYC. To test whether accumulation of MIZ1 is required for repression of MYC target genes, we inhibited HUWE1 in control cells and in cells that express a doxycycline-inducible shRNA targeting MIZ1 (Fig7A). RQ–PCR assays showed that depletion of MIZ1 alleviated repression of individual MYC-activated target genes (Fig7B). GSEA of RNA sequencing data showed that depletion of MIZ1 did not uniformly relieve repression of MYC-activated target genes by HUWE1 inhibitors; instead, the effect was strongest on target genes encoding ribosomal proteins, in essence abolishing repression by HUWE1 inhibitors (Fig7C and D). Analysis of ChIP-sequencing data confirmed that MIZ1 accumulated at the core promoter of ribosomal genes upon HUWE1 inhibition (Fig7E). The data show that HUWE1 is required for transactivation by MYC at least in part since it counteracts the assembly of repressive MYC/MIZ1 complexes at E-box target genes that are bound by MYC (see Fig7F for a model). We note that histone deacetylase 2, another target that is degraded by HUWE1, has been implicated in transcriptional repression at E-box sites that are transactivated by MYC (see Discussion) (Laherty et al, ). While effects of HUWE1 on total cellular pools of HDAC2 were small under our experimental conditions (Fig6A,C,D,E), it is possible that local degradation of HDAC2 by HUWE1 at MYC-bound E-boxes contributes to transcriptional activation by MYC.

14A large body of proof-of-principle experiments suggests that targeting MYC proteins may have considerable therapeutic benefit in human tumors (Felsher & Bishop,; Soucek et al, ). As a consequence, several strategies have been proposed to target MYC or N-MYC using small molecule inhibitors. One strategy is based on a class of small molecules that bind to the leucine zipper of MYC and N-MYC and inhibit heterodimerization with MAX, which is critical for transformation by MYC proteins (Yin et al, ). A second way of targeting MYC is the use of small molecule inhibitors that interfere with MYC expression or stability: Examples are inhibitors of the bromodomain protein BRD4, which is critical for transcription of the MYC promoter, and inhibitors of the Aurora-A kinase that disrupt a stabilizing interaction of Aurora-A with N-MYC (Brockmann et al,; Delmore et al, ). We propose here that targeting the HUWE1 ubiquitin ligase is feasible and allows for a tumor cell-specific inhibition of MYC function.

15We have shown previously that both activating and repressive complexes of MYC can be detected at most MYC-bound promoters in tumor cells (Walz et al,; Wiese et al, ). Previous work had also shown that HUWE1 is a degrading ubiquitin ligase for MIZ1 (Inoue et al,; Yang et al, ). We show here that MIZ1 globally accumulates at MYC-bound promoters upon inhibition of HUWE1, thereby shifting the equilibrium of MYC complexes from activating toward repressive complexes. shRNA-mediated depletion of MIZ1 attenuated, but did not abolish inhibition of MYC-dependent transactivation by HUWE1 inhibition. It is possible that this is due to the fact that the depletion of MIZ1 is not complete. Alternatively, HUWE1 has additional functions and substrates via which it promotes MYC-dependent transactivation, such as ubiquitination of MYC itself (Adhikary et al, ). HUWE1 also ubiquitinates and degrades the HDAC2 histone deacetylase, which associates with the MXD/MAX repressive complex (Laherty et al, ) that antagonizes MYC transactivation, and it is possible that degradation of chromatin-bound HDAC2 is also a critical activity of HUWE1 (Zhang et al, ).

16Notably, depletion of MIZ1 preferentially de-repressed genes that encode ribosomal proteins when HUWE1 was inhibited. Previous work had shown that MIZ1 function is controlled by the ribosomal protein RPL23 that retains nucleophosmin (NPM), a co-activator of MIZ1, in the nucleolus (Wanzel et al, ). Like HUWE1, NPM antagonizes complex formation of MYC with MIZ1 (Herkert et al, ). The data reported here therefore argue that HUWE1 and NPM1 are part of a regulatory circuit that couples transcriptional activation of ribosomal protein genes by MYC to the levels of free ribosomal proteins in the nucleolus. Additional circuits exist that couple MYC function to the level of ribosomal proteins (Dai et al, ).

17The notion that HUWE1 antagonizes assembly of MYC/MIZ1 repressive complexes is consistent with genetic data showing that deletion of HUWE1 enhances MYC/MIZ1-dependent repression during skin carcinogenesis (Inoue et al, ). Importantly, MIZ1-dependent repression of Cdkn1a (encoding p21CIP1) expression is a critical function of MYC in Ras-driven skin tumorigenesis, since deletion of MYC or MIZ1 inhibits Ras-induced tumor formation and upregulates p21CIP1 expression; tumor formation in the absence of MYC or MIZ1 is restored when Cdkn1a is co-deleted (Honnemann et al,; Oskarsson et al, ). We propose therefore that HUWE1 degrades MIZ1 in both colon carcinoma cells and keratinocytes, but whether this promotes or inhibits oncogenesis depends on whether transcriptional activation or repression by MYC is critical for oncogenesis in a given tumor.

18Surprisingly, neither genetic ablation of HUWE1 (Zhao et al, ) nor its inhibition (this report) affects expression of MYC target genes in embryonic stem and normal epithelium cells of the intestine, arguing that targeting HUWE1 may open a significant therapeutic window. One factor contributing to this specificity is that embryonic stem cells express both MYC and N-MYC and they are functionally redundant (Smith et al, ). The association of MIZ1 with N-MYC is weaker than that with MYC (Peukert et al, ) (E. Wolf and M. Eilers, unpublished observations), and comparison of N-MYC-dependent with MYC-dependent gene expression suggests that N-MYC does not repress transcription via MIZ1 in vivo (M. Eilers, unpublished observations). This difference between N-MYC and MYC may explain why N-MYC does not require HUWE1 for transactivation. A second factor contributing to this specificity is the ARF tumor suppressor protein that is expressed in tumor but not in normal cells; ARF inhibits HUWE1, promotes the assembly of MYC/MIZ1 complexes, and inhibits transactivation by MYC (Chen et al,; Herkert et al,; Qi et al, ). Additionally, expression of HDAC2 is high in colon carcinoma cells since it is downstream of the WNT pathway (Zhu et al, ). It is likely, therefore, that the high levels of ARF and HDAC2 enhance the potency of the HUWE1 inhibitors in colon cancer cells and contribute to the specificity of the effects. Collectively, our findings argue that the effects of HUWE1 depletion or inhibition will be dictated by the state of the MYC/N-MYC network and that this will allow specific inhibition of MYC function in colon tumor cells.

19Targeting the ubiquitin system is emerging as a possibility to specifically interfere with the assembly and function of oncogenic networks for tumor therapy, and a general inhibitor of the proteasome, bortezomib, is currently in clinical use (Mahindra et al, ). Proof-of-principle compounds show that it is possible to specifically inhibit individual RING finger and F-box ubiquitin ligase complexes as well as their antagonists, the de-ubiquitinating enzymes (Micel et al, ). Our data provide strong support for the notion that it is possible to specifically target individual HECT-domain ubiquitin ligases using small molecule inhibitors and use this to control the activity of MYC and possibly other transcription factors for tumor therapy.

20A DELFIA format assay was used to measure the covalent association of ubiquitin with HUWE1 (auto-ubiquitination). Reactions were carried out on streptavidin-coated 96-well plates (Roche). 10 μl Assay buffer (50 mM Tris–HCl pH 7.5, 50 mM KCl, 0.5 mM MgCl2, 0.05 mM CaCl2, 0.5 mM DTT, 0.1% Tween-20, 1% DMSO) per well is added with 4% DMSO and inhibitor compounds, followed by 20-min incubation at room temperature with 20 μl enzyme mix [3.7 ng His-UBA1, 70 ng His-UbcH5b, 30 ng biotinylated His-HUWE1 (C-terminal ∼350 amino acids = HECT-domain), 100 ng His-3xMYC-ubiquitin]. After addition of 10 μl 2 mM ATP, samples were incubated for 3 h at room temperature. 200 μl DELFIA wash buffer (Perkin Elmer) was used each for five washing steps. 50 μl Eu-labeled MYC antibody, diluted 1:5,000 in DELFIA assay buffer (Perkin Elmer), was added followed again by five washing steps. Europium fluorescence was measured with 50 μl DELFIA enhancement solution (Perkin Elmer) in a Wallac Victor fluorescence plate reader (Perkin Elmer).

21HCT116, HEK293T, HeLa, HT29, ES, Phoenix, and U2OS cells were cultivated in DMEM (Sigma). Ls174T and SW480 cells were grown in RPMI 1640 (Sigma). All media were supplemented with 10% FBS (Biochrom AG) and 1% penicillin/streptomycin (Sigma). Medium for ES cells contained 15% FBS, 1% NEAA (Gibco), 6 μg/ml LIF, and 0.05 mM β-mercaptoethanol (Sigma). For cultivation of ES cells, plates were coated with 0.1% gelatine. PAM212 cells were grown in Ca2+-free DMEM (Gibco) supplemented with 10% KGM-2 (Lonza), 10% FBS, 1% L-Glutamine (SAFC Biosciences), 0.8% Gentamycin (Lonza). BI8622 and BI8626 were dissolved in DMSO at a stock concentration of 10 mM and used at 10 or 20 μM final concentration unless indicated otherwise. For UV treatment, cells were exposed to UVB (500 J/m2) for 1 min. Transfections were carried out using PEI (Polyethyleneimine, Sigma). Quantification of crystal violet staining was performed by extracting dye with 10% acetic acid and measuring absorbance at 590 nm. IC50 values were calculated using the four parameter logistic equation model.

22Retroviruses expressing an shRNA against HUWE1 were generated by transfection of pRetroSuper, pLKO (TRC consortium) or a pInducer (Meerbrey et al, ) vector together with psPAX.2 and pMD2.G into HEK293T cells. shRNA sequences are shown in Supplementary Table S1. Details on vector constructs used are available upon request. Cells infected with empty vector or vector expressing non-targeting shRNA were used as controls. Generation of MIZ1-depleted cells included a FACS sorting step.

23Mouse experiments were performed in accordance with Swiss guidelines and approved by the Veterinarian Office of Vaud. Subcutaneous xenografts were established by injection of 500,000 cells into the right flank of NOD/SCID/IL2R common gamma chain deficient (cγ−/−) mice under isoflurane anesthesia. Doxycycline (1 mg/ml in drinking water) treatment was started 7 days after engraftment, when tumors became palpable. Tumor progression was monitored by external caliper measurements of the longest (LD) and smallest (SD) diameter of the subcutaneous tumors. For orthotopic xenografts, mice were anesthetized with 90 mg/kg ketamine and 14.5 mg/kg xylazine, and the cecum was exteriorized by laparotomy. 500,000 Ls174T cells were suspended in 15 μl 1/1 matrigel/medium and injected into the cecum. Tumor growth was followed by in vivo imaging of luciferase activity using an IVIS imaging system (Xenogen). For in vivo imaging, mice were injected i.p. with 150 mg/kg luciferine (Biosynth) and imaged for 1–60 s under isoflurane anesthesia. Quantification was performed with LivingImage software (Xenogen). Doxycycline (1 mg/ml in drinking water) treatment was started 14 days after engraftment when the mice showed an abdominal luciferase signal. To assess cell proliferation, mice were injected i.p. with 1 mg BrdU (Sigma) 1 h prior to sacrifice and tumor dissection.

24Murine intestinal crypt cultures were established from wild-type mice of a mixed background (50% C57Bl6J, 50% S129) as previously described (Sato et al, ). Briefly, intestines were isolated and flushed with PBS. Villi were removed by scraping and remaining tissue cut and washed with PBS. Crypt extraction was carried out in PBS + 2 mM EDTA for 30 min at 4 °C. Isolated crypts were washed with PBS and plated in growth factor reduced Matrigel (BD Biosciences). Crypts were grown in Advanced DMEM (Invitrogen) supplemented with N2 (Invitrogen), B27 (Invitrogen), EGF (Peprotech), Noggin (Peprotech), and R-spondin (Peprotech). Media were replaced every 2 days and crypts passaged weekly. For drug treatments, crypts were grown until they formed budding “organoids” and then treated with 10 μM BI8622 or control treated with 0.1% DMSO for 24 h. Organoid survival was scored at this point based on their microscopic appearance. Cell viability was also determined using a CellTiter-Blue Cell Viability Assay (Promega) according to the manufacturer's instructions.

25For immunoblotting, cells were lysed in RIPA buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% NP-40, 0.5% DOC, 0.1% SDS) containing protease inhibitors (Calbiochem). Cleared protein lysates were separated by SDS–PAGE and transferred to a PVDF membrane (Millipore). Immunoprecipitation was performed by lysing cells in IP buffer (20 mM HEPES pH 7.9, 200 mM NaCl, 0.2 mM EDTA, 1% NP-40, 10 mM NaF, 10 mM Na-Pyrophosphate) containing protease inhibitors. Cell lysates were cleared by centrifugation and immunoprecipitated with indicated antibodies. Protein complexes were recovered with Protein A- or Protein G-coupled sepharose beads (Sigma). Immunoprecipitates were washed three times with IP buffer, boiled in SDS sample buffer, and subjected to immunoblotting. Antibodies are listed in Supplementary Table S2.

26Transfected cells were lysed in 6 M guanidine-HCl, 100 mM phosphate buffer pH 8.0, 10 mM imidazole, 0.4% Triton X-100, 0.125 M NaCl, containing protease inhibitors, 10 μM MG-132 (Sigma), and 5 mM NEM (Sigma). Cell lysates were sonicated and cleared by centrifugation. His-ubiquitin-modified proteins were collected with Ni-NTA-Agarose (Qiagen). After washing, precipitates were boiled in SDS sample buffer supplemented with 200 mM imidazole and subjected to immunoblotting.

27Cells were lysed in ChIP lysis buffer (5 mM PIPES pH 8.0, 85 mM KCl, 0.5% NP-40), and after centrifugation, nuclei were resuspended in ChIP-RIPA buffer (10 mM Tris–HCl pH 7.5, 150 mM NaCl, 1% NP-40, 1% DOC, 0.1% SDS, 1 mM EDTA). DNA was sonicated with a Branson sonifier to obtain DNA fragments ≤ 500 bp. Chromatin was pre-cleared with Protein A- or Protein G-sepharose beads (Sigma, blocked with salmon sperm DNA (Invitrogen) and BSA (Roth)) and immunoprecipitated with indicated antibodies. For ChIP-seq, only BSA was used for blocking. Chromatin/protein complexes were collected by incubation with Protein A- or Protein G-sepharose beads for 6 h. After several washings, chromatin was eluted with 1% SDS/0.1 M NaHCO3 and crosslinking was reverted overnight. DNA was purified with the GeneJet PCR Purification Kit (Thermo Scientific) or by phenol–chloroform extraction and analyzed by RQ–PCR. Percent input was calculated by subtraction of IP or control CT values from input CT values. Primers for RQ–PCR are listed in Supplementary Table S3.

28Libraries for ChIP-sequencing were constructed using the NEBNext ChIP-Seq Library Prep Master Mix Set for Illumina (NEB). Briefly, ChIP DNA was end-repaired, A-tailed, Illumina adaptors were ligated, and the DNA separated on a 2% agarose gel. DNA fragments of 200 bps were excised and purified using a QIAquick Gel Extraction Kit (Qiagen). Size-selected DNA was amplified with 18 PCR cycles, quality-controlled using the Experion chip electrophoresis system (Bio-Rad), and quantified using a picogreen assay. Sequencing of multiplexed DNA libraries was done on an Illumina Genome Analyzer IIx following the manufacturer's instructions. Basecalling was performed using the real-time analysis (RTA) package within the Genome Analyzer Sequencing Control Software (SCS2.10). Demultiplexing and generation of FASTQ files were done with the CASAVA software only considering high quality sequences (PF-cluster).

29For RNA sequencing, total RNA was extracted from cells using the RNeasy Mini Kit (Qiagen) with on-column digestion of DNA. Poly-A+ RNA was isolated from total RNA using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB). Libraries for RNA sequencing were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina (NEB) or the NEBNext mRNA Library Prep Master Mix Set for Illumina (NEB) following the instructions of the manufacturer. Depending on the kit used for sample preparation, DNA libraries were size selected using Agencourt AMPure XP Beads (Beckman Coulter) followed by amplification with 12 PCR cycles or by excision of 250 bp fragments from 2% agarose gels and amplification with 13 cycles of PCR. Amplicon sizes and library quantities were determined using the Experion chip electrophoresis system (Bio-Rad). Libraries were sequenced on an Illumina Genome Analyzer IIx following the manufacturer's instructions. Reads were aligned to the human genome (hg19) with BOWTIE v0.12.8 (Langmead, ) using default parameters. Analysis of the aligned sequence data was done in R/Bioconductor.

30Microarray analyses were performed on a 44K Whole Human Genome Array (G4845A 026652, Agilent), and raw data were generated with the Feature Extraction software v10.1.1.1 (Agilent).

31For ChIP-seq experiments, reads were mapped with BOWTIE v0.12.8 (Langmead, ) to the human genome (hg19) and peaks were called with IgG sample as control using MACS v1.4.2. The keep-dup parameter was adapted depending on the enrichment at the highest peaks. Peaks were annotated to the nearest RefSeq gene (UCSC GoldenPath RefSeq database) with the “closestBed” feature of the Bedtools suite v2.11.2 (Quinlan & Hall, ). Heat maps indicating co-occupancies at transcriptional start sites and corresponding tag density distributions were generated with SeqMiner v1.3.3 in which the strand orientation was taken into account. To calculate recruitment of MYC and MIZ1 after BI8622 treatment, MIZ1-bound promoters (−1 kb to +0.5 kb relative to the TSS) containing consensus E-boxes were selected and the number of tags were counted in a region ±100 bp around the center of the MIZ1 peak. The tool edgeR (Robinson et al, ) was used to determine differential gene expression and conduct statistical inference.

32The P-value was calculated using a one-sample two-tailed Student's t-test with μ = 0. Gene set enrichment analysis (GSEA) was performed using default settings and the C2 gene sets from the MSigDB (www.broadinstitute.org/gsea/msigdb). To compare two GSEA, the normalized enrichment scores (NES) for all repressed gene sets were plotted and gene sets containing > 25% ribosomal protein genes were highlighted. Data are presented as means with standard deviation unless defined differently in the figure legends. P-values were calculated using Student's t-test.

33Activation of MYC is a central driver of colorectal carcinogenesis, and genetic experiments argue that inhibition of MYC would have a major therapeutic benefit for this tumor. MYC proteins are transcription factors that interact with their partner proteins and with DNA via large surfaces, making direct targeting with small molecule inhibitors difficult.

34We had previously shown that transactivation by MYC proteins requires the ubiquitin ligase HUWE1 (HECTH9). We have identified highly specific inhibitors of HUWE1 and found that inhibition of HUWE1 is a feasible approach to inhibit MYC function in a tumor cell-specific manner, since HUWE1 is required to prevent assembly of a repressive complex of MYC with MIZ1 on MYC-activated target genes.

35Here, we establish a new principle that allows inhibiting MYC-dependent transactivation for tumor therapy.

36Human liver microsomes (1 mg/ml, GE Healthcare) were pre-incubated with 10 μM BI8622 or BI8626 in 0.1 M pre-warmed potassium phosphate buffer (pH 7.4) for 5 min at 37 °C in a total volume of 500 μl. To start the reaction, 1 mM NADPH (Sigma) was added and incubated for 0, 5, 7.5, 10, 30, 60 min at 37 °C. Time point 0 min was taken directly after addition of NADPH. At each time point, 50 μl aliquots was taken and the reaction was stopped by addition of 140 μl methanol for subsequent mass spectral analysis. Mice were injected with 10 μl inhibitor compound (10 mM) per gram body weight to achieve a concentration of 100 μM in vivo. Mice were sacrificed after 30, 60, 120, and 240 min, and blood samples were taken. Samples were centrifuged (4 °C; 10 min; 800 g), and plasma supernatant was used to determine the inhibitor concentration.

37For sample preparation, 50 μl of HUWE1 inhibitor-containing samples (blood plasma or microsomal preparations) was mixed with 10 μl standard solution (50 μM BI8622 or BI8626 in methanol), 140 μl methanol, 150 μl chloroform, and 25 μl 2 M NH3. After centrifugation, the supernatant was mixed with 200 μl chloroform and 50 μl water. The suspension was centrifuged, and the lower phase washed twice with 206 μl of methanol/chloroform/water (100/6/100; v/v/v). After centrifugation, 200 μl of the lower phase was evaporated at 60 °C under a stream of nitrogen gas. The residue was resuspended in 40 μl 1% acetic acid and centrifuged. For mass spectral analysis, 35 μl supernatant was mixed with 40 μl methanol. Mass spectral data were obtained using an APEX II FT-ICR mass spectrometer.