MYCMI-7: A Small MYC-Binding Compound that Inhibits MYC: MAX Interaction and Tumor Growth in a MYC-Dependent Manner

Deregulated expression of MYC family oncogenes occurs frequently in human cancer and is often associated with aggressive disease and poor prognosis. While MYC is a highly warranted target, it has been considered “undruggable,” and no specific anti-MYC drugs are available in the clinic. We recently identified molecules named MYCMIs that inhibit the interaction between MYC and its essential partner MAX. Here we show that one of these molecules, MYCMI-7, efficiently and selectively inhibits MYC:MAX and MYCN:MAX interactions in cells, binds directly to recombinant MYC, and reduces MYC-driven transcription. In addition, MYCMI-7 induces degradation of MYC and MYCN proteins. MYCMI-7 potently induces growth arrest/apoptosis in tumor cells in a MYC/MYCN-dependent manner and downregulates the MYC pathway on a global level as determined by RNA sequencing. Sensitivity to MYCMI-7 correlates with MYC expression in a panel of 60 tumor cell lines and MYCMI-7 shows high efficacy toward a collection of patient-derived primary glioblastoma and acute myeloid leukemia (AML) ex vivo cultures. Importantly, a variety of normal cells become G1 arrested without signs of apoptosis upon MYCMI-7 treatment. Finally, in mouse tumor models of MYC-driven AML, breast cancer, and MYCN-amplified neuroblastoma, treatment with MYCMI-7 downregulates MYC/MYCN, inhibits tumor growth, and prolongs survival through apoptosis with few side effects. In conclusion, MYCMI-7 is a potent and selective MYC inhibitor that is highly relevant for the development into clinically useful drugs for the treatment of MYC-driven cancer. Significance: Our findings demonstrate that the small-molecule MYCMI-7 binds MYC and inhibits interaction between MYC and MAX, thereby hampering MYC-driven tumor cell growth in culture and in vivo while sparing normal cells.


Introduction
Deregulated expression of the MYC family of oncogenes/transcription factors MYC, MYCN, and MYCL (here collectively referred to as "MYC") occurs in more than half of all cancers, and is often strongly associated with aggressive tumors, resistance to therapy, and poor prognosis (1,2). MYC is a key player in most or all proliferative signaling networks where it acts as a central "hub" for interactions with other proteins, such as the heterodimerization partner MAX, which is required for specific binding to "E-box" regulatory elements in the target gene promoters, thereby regulating transcription (3)(4)(5)(6). Attempts to find small molecules that interfere with the MYC:MAX or MYC:MAX:DNA interaction, or favor MAX:MAX interaction have been made previously (7)(8)(9)(10)(11)(12)(13)(14)(15)(16).
Although promising, these attempts have not yet resulted in molecules that have reached the clinic, sometimes due to low potency, poor or unclear target selectivity, and/or inadequate bioactivity in vivo (reviewed in refs. [17][18][19][20][21], warranting further efforts to discover and characterize MYC-inhibitory molecules with potential for clinical development. Recently, based on the dominant-negative Omomyc approach, cell-penetrating peptides that bind MYC and blocks MAX binding have been developed (22), and are now in phase I/II clinical trials (NCT04808362). Most likely, it will be beneficial to develop many different strategies to successfully combat this multifunctional target in future cancer treatment. In addition, combinatorial use of anti-MYC drugs with different mechanism of action could potentially synergize and give less side effects by lowering the concentrations of each drug.
Using a cell-based MYC:MAX interaction inhibitor screen, we previously identified several compounds that target the MYC:MAX protein-protein interaction in cells, named MYCMIs. Three of these, MYCMI-6, MYCMI-11, and MYCMI-14, have been characterized previously (8). In the same screen, the compound MYCMI-7 was identified, which in contrast to MYCMI-6, MYCMI-11, and MYCMI-14, also downregulates MYC protein expression and therefore potentially has higher therapeutic efficacy. Here, we characterized the molecule MYCMI-7 in more detail and found that it is a MYC-binding compound that inhibits the MYC:MAX protein interaction in cells, and increases MYC protein turnover. MYCMI-7 induces apoptosis in a MYC-dependent manner in tumor cells but not in normal cells at single-digit micromolar concentrations and exhibits potent tumor growth-inhibitory activity in vivo in mouse models of MYC-driven leukemia, breast cancer, and neuroblastoma.

Cell Culture
The following cell lines were purchased from the ATCC repository dur- Mutu, and U-937 were kept in RPMI supplemented with 10% FBS and 1% penicillin/streptomycin. In addition, the medium for HO15. 19 and primary REFs contained 1% sodium pyruvate. U2OS-MYCER cell lines were cultured in phenol-red free DMEM supplemented with 10% estrogen-free FBS and 1% penicillin/streptomycin and treated with 100 nmol/L 4-hydroxytamoxifen (4-OHT; Sigma-Aldrich) to activate MYCER. HEK293 cell lines were established to stably express both full-length MYC and MAX fused to split Guassia luciferase fragments GLuc1 and GLuc2, respectively. To induce TOP2 degradation in HCT116TOP2A_mAID cells, the culture medium was replaced with DMEM containing 500 μmol/L freshly prepared auxin (Indole-3-acetic acid, Sigma). All cells used were Mycoplasma free and kept at 37°C and 5% CO 2 . Cells were used for a maximum of 10 passages after collection or thawing. All cell lines After another 17 hours, the cells were harvested, and luciferase activity measured using the Dual Luciferase Kit (Promega) in a Berthold Lumat LB9501 or OmegaFluostar (BMG Labtech). For further details, see Supplementary Data.

In Situ Proximity Ligation Assay
The in situ proximity ligation assay (isPLA) has been described previously (8). Briefly, cells were grown on collagen-coated chamber slides (Falcon), treated with compounds, and then washed twice with PBS and fixed in ice-cold methanol for 5-15 minutes at room temperature. Slides were washed in PBS with 0.05% Tween 20 and incubated in blocking buffer after which isPLA was performed using the Duolink In Situ PLA Kit (Sigma-Aldrich) according to the manufacturer's protocol. DNA was stained with DAPI. Incubation with primary antibodies were performed at +4°C overnight. Images were taken using an Axiovert 200M inverted microscope (Zeiss) and fluorescent dots were quantified using semiautomated analysis in ImageJ (http://imagej.net) and averaged to the number of dots per cell. Antibodies used are listed in Supplementary Data.
Briefly, cells were crosslinked with 1% formaldehyde on ice for 6 minutes. Nuclear chromatin was sonicated on ice to fragments from 0.3 kb to 0.5 kb. Nuclear chromatin equivalent to 2.5 × 10 7 cells was immunoprecipitated with 2 μg antibody. For studies of protein turnover, cells were treated with 100 μg/mL cycloheximide to block protein synthesis for 2 hours, followed by chase. Quantification of western blots was done by Image J analysis.

Surface Plasmon Resonance
The surface plasmon resonance (SPR) experiments were performed at 25°C using a Biacore T200 (GE Healthcare) instrument kindly provided by SciLifeLab Solna. An amino coupling procedure was used to immobilize protein on a CM5 sensor chip (GE Healthcare). Sensorgrams were generated by subtraction of the reference (blank immobilized) surface. For further details, see Supplementary Information.

RNA-Sequencing Analysis
Libraries for RNA sequencing were prepared using the TruSeq Stranded Total RNA kit with RiboZero (Illumina) and sequenced on two lanes of the HiSeq 2500 platform with a single-end 1 × 51 setup and the HiSeq Rapid SBS v2 chemistry. Demultiplexed .fastq files were aligned to the human GRCh37 reference genome using Tophat v 2.0. After alignment, .bam files from two separate flowcell lanes were merged using samtools. Raw read counts per gene were then generated using htseq-count v0.6.1. Differential expression analysis comparing the two DMSO-treated to the two MYCMI-7-treated samples was performed using the R/Bioconductor DESeq2 package v1. 26.0 (Bioconductor v3.10, R v3.6.1), Following differential expression analysis, all genes were ranked according to P adj value and log fold change. Gene-set enrichment analysis was then performed using GSEA software with the Hallmarks (H) and curated (C2:CGP) MSigDB gene sets; v6.2.

Mouse Tumor Models
All animal protocols in these studies were approved by the ethical committee for animal experiments of northern Stockholm (N47/14, N241/15 and N231/14) and of Uppsala (C41/14). Mice were maintained under pathogen-free conditions according to guidelines of the animal facility at MTC, Karolinska Institutet, or at AKM, Karolinska University Hospital. Drug toxicity in the mice was evaluated by examining changes in body weight, changes in behavior, and overall wellness, loss of fur coat, breathing, mouse activity, and body posture following drug injections, food intake, as well as the histologic effect on the livers of treated mice.
For the AML tumor model, MYC+BCL-XL expressing, GFP + leukemic cells were isolated from leukemic mice as described previously (28). A total of 3 × 10 5 cells were injected into each recipient C57BL mouse after irradiation (600 rad).
AML-like leukemia initiation was confirmed via flow cytometry and Giemsa staining, at day 8 after transplantation. MYCMI-7 cells were then administrated intraperitoneally daily at a dose of 12.5 mg/kg. Liver, spleen, and bone marrow were extracted at day 11, 15, and endpoint for measurement. In addition, body weight of mouse in each group was also collected every fourth day until the endpoint of the experiment.
For description of IHC and immunofluorescence, see Supplementary Data.

Statistical Analysis
Analysis of the probability of a cell line with "high MYC" or "low MYC" mRNA/protein levels to respond to MYCMI-7 among the NCI-60 cancer cell lines was tested with the binomial exact test. The analysis was carried out in R (v. 3.3.3; R Foundation for Statistical Computing), at a level of significance α = 0.05. The Kaplan-Meier survival curves in the animal studies were evaluated with log-rank test using GraphPad Prism. The rest of the data were analyzed with two-tailed paired Student t tests (GraphPad Prism).

Data Availability Statement
The RNA-seq data in this study are publicly available in Gene Expression Omnibus (GEO) at GSE197062. The other data generated in this study are available within the article and its Supplementary Data files or upon request from the corresponding author.  1B). In contrast, MYCMI-7 did not have any significant effect on homodimerization of the bZip protein GCN4 (Fig. 1C). Although MYCMI-7

MYCMI-7 Inhibits the MYC:MAX Interaction in Cells, Binds MYC In Vitro, and Blocks MYC Function
shows structural similarities to ellipticine, the latter was inactive at the same concentration ( We concluded that MYCMI-7 binds MYC in vitro, and that it rapidly, strongly, and selectively inhibits MYC:MAX interaction and MYC's association with chromatin in cells at low micromolar concentrations.

MYCMI-7 Increases MYC Protein Turnover
We next studied the expression levels of MYC and MAX after MYCMI-7 treatment. The steady-state level of the MYC protein decreased drastically in MCF7 cells after 17-hour treatment with 5 μmol/L MYCMI-7, while treatment with 5 μmol/L ellipticine had no effect ( Fig. 2A). Considering the structural similarities between the two compounds, we increased the dose to 10 μmol/L, but ellipticine still only had a slight effect on MYC protein level. Also, DMSO alone had a some effect on MYC expression at 10 μmol/L, although this is not   To map regions of MYC involved in MYCMI-7-mediated MYC turnover, different MYC mutants were utilized in transient transfection assays (Fig. 2I). This analysis showed that the C-terminus, but not the N-terminus, of MYC was required for turnover. Further mapping of the C-terminal part showed that the DNA-binding basic region was necessary for MYCMI-7-mediated MYC turnover ( Fig. 2J-L).

MYCMI-7 Reduces Tumor Cell Growth and Viability in a MYC-Dependent Manner and Downregulates the MYC Pathway
To address whether MYCMI-7 affects cell growth in a MYC-dependent manner, we utilized the immortalized Rat1 fibroblasts with different MYC status mentioned above; the HO15.19 MYC-null cells derived from parental TGR1 cells, and HOMyc3, which are HO15.19 cells reconstituted with MYC (34). Proliferation/viability of cells expressing MYC declined drastically at low concentrations of MYCMI-7, with an average growth inhibition of 50% (GI 50 ) around 2 μmol/L as measured by WST-1 assay (which measures metabolic activity in cells), while the MYC-null cells were unaffected even at concentrations of 10 μmol/L (Fig. 3A), demonstrating that the effect of MYCMI-7 was MYC dependent. In contrast, ellipticine reduced growth of all three cell clones in a dose-dependent manner, irrespective of MYC status ( Supplementary   Fig. S3A). When HO15.19 cells were exposed to MYCMI-7 concentrations higher than 10 μmol/L, a marked decline in viability was observed, suggesting that off-target effects started to appear above this concentration ( Supplementary Fig. S3B).
To investigate whether MYCMI-7 affects growth of typical MYC-driven tumor cells, we utilized a panel of childhood neuroblastoma cells with or without MYCN-amplification as well as a set of Burkitt's lymphoma cell lines, which carry MYC translocations. Treatment with MYCMI-7 reduced tumor growth and viability in all the neuroblastoma cell lines, but the effect was significantly stronger in the MYCN-amplified cases (Fig. 3B). Note that non-MYCN-amplified neuroblastoma cells express MYC, albeit at a lower level than MYCN in amplified lines (8). The efficiency of MYCMI-7 toward MYCNamplified neuroblastoma cells was even stronger in three-dimensional (3D) cultures, with GI 50 in the nanomolar range ( Fig. 3C; Supplementary Fig. S3C).
To investigate whether the levels of MYC expression in tumor cells correlate with growth-inhibitory response to MYCMI-7, we utilized GI 50 and and combined with MYC protein data obtained from Novartis proteome scout SymAtlas Project (https://proteomescout.wustl.edu/proteins/ 52581/expression) or elsewhere in the literature as described previously (8).
The cell lines were categorized as "responsive" or "less responsive" to MYCMI-7 based on average logarithmic GI 50 values, as well as the categories "higher MYC" and "lower MYC" based on higher or lower than average MYC mRNA and/or high protein levels. There was a significant correlation between the response to MYCMI-7 and the MYC mRNA/protein levels among the 60 tumor cell lines (Fig. 3D). Ninety-six percent of the cells with high MYC mRNA/protein level were responsive to MYCMI-7 and 73% of the cells with low MYC levels were less responsive (Fig. 3D). This indicates that cells with high MYC levels are more likely to respond to MYCMI-7 treatment than cells with low MYC levels.
To investigate whether MYCMI-7 could inhibit MYC-induced oncogenic transformation of normal cells together with H-RAS, normal REFs were transfected with MYC + RAS vectors. Formation of transformed foci as well as the ability of MYC + RAS-transformed REFs to form colonies in semi-solid medium was strongly inhibited by treatment with MYCMI-7 ( Supplementary Fig. S3E and S3F).
To study the impact of MYCMI-7 on global gene expression, we performed RNA-seq analysis in MCF7 breast carcinoma cells after treatment with 5 μmol/L MYCMI-7 for 24 hours. Gene-set enrichment analysis (GSEA) of differentially expressed genes showed a downregulation of the MYC and E2F target genes (Fig. 3E, left and middle). In contrast, upregulated genes were enriched in pathways associated with inflammatory signaling via NFκB, which is consistent with the reported suppressive action of MYC on immune signaling (Fig. 3E, right; refs. 35,36). To further document the impact of MYCMI-7 on MYC's transcriptional activity, we utilized U2OS cells expressing a MYCestrogen receptor (MYCER) fusion protein, which is regulated by 4-OHT (24). Treatment with MYCMI-7 significantly reduced 4-OHT-induced expression of CR, RGS, CAMKV, and nucleolin (Fig. 3F), which all previously have been characterized as direct MYC target genes (37,38).
Taken together, these results suggest that MYCMI-7 reduces tumor cell growth/viability in a MYC-dependent manner and downregulates the MYC pathway.

MYCMI-7 Induces Growth Arrest and Apoptosis in Malignant Cells and only Growth Arrest in Normal Cells
We next studied the effect of MYCMI-7 on the cell cycle utilizing P493-6 cells with regulatable MYC (31). Cells synchronized in G 0 -G 1 by downregulating MYC with doxycycline were allowed to reenter the cell cycle by doxycyclinewithdrawal while treated with MYCMI-7 or DMSO. Compared with DMSO, MYCMI-7-treated cells showed a higher G 1 -S ratio, but also strongly induced cell death, as evidenced by the high proportion of sub-G 1 cells (Fig. 4A, left). In contrast, MYCMI-7 induced G 1 arrest but not cell death in normal REFs ( Fig. 4A, right). Furthermore, MYCMI-7 induced apoptosis in immortalized TGR1 and MYC-reconstituted HO15.19 (HOMyc3), but not in MYC knockout HO15.19 Rat1 cells (Fig. 4B, left), demonstrating that MYCMI-7-induced apoptosis is MYC dependent. To investigate whether MYCMI-7 induced apoptosis also in tumor cells, we utilized A375 melanoma cells. Titration using these cells showed that MYCMI-7 reduced viability and inversely induced apoptosis in a dose-dependent manner with an GI 50 in the nanomolar range (Fig. 4C). Looking at different normal cells, we found that MYCMI-7 did not induce apoptosis in normal REFs or normal human peripheral blood lymphocytes (Fig. 4B, middle and right). Normal human dermal fibroblasts (NHDF) treated with MYCMI-7 for 3 days showed a lower overall cell growth as determined by cell count and resazurin assay (which is a redox indicator that undergoes colorimetric change mediated by dehydrogenase enzymes in metabolically active cells) when normalized to exponentially growing DMSO-treated cultures at low micromolar concentrations, while cell viability was unaffected. This suggests that MYCMI-7 induced growth arrest without killing the cells (Fig. 4D and E). Also, in primary normal human epidermal melanocytes (NHEM), MYCMI-7 reduced metabolic activity, but did not induce cell death ( Fig. 4F and G). In

MYCMI-7 Does not Induce DNA Damage Signaling at Active Concentrations and Acts Independently of Topoisomerase 2 and p53
Considering the structural resemblance of MYCMI-7 to ellipticine ( Fig. 1A; Supplementary Fig. S4A), which has been described as a DNA intercalator and topoisomerase 2 (TOP2) inhibitor (39), we were concerned about possible effects of MYCMI-7 on TOP2 activity and DNA damage. We first investigated whether MYCMI-7 is able to inhibit TOP2A, which is the main TOP2 isoform in cycling cells (40). First, we utilized an in vitro TOP2A decatenation assay. TOP2 enzymes decatenate kinetoplast DNA (kDNA) converting a network of interlocking DNA rings into individual rings that can be separated by agarose gel electrophoresis. The TOP2 inhibitor doxorubicin started to inhibit TOP2A activity at 3 μmol/L with complete TOP2A inhibition at 10 μmol/L (Supplementary Fig. S4B and S4C), while the TOP1 inhibitor camptothecin has no effect, as predicted. MYCMI-7 exhibited TOP2A-inhibitory activity at concentrations higher than 10 μmol/L with complete TOP2A inhibition at 100 μmol/L. This is consistent with off-target effects starting to appear in MYC knockout cells at higher concentrations than 10 μmol/L ( Supplementary Fig. S3B). Ellipticine had no effect on TOP2A activity at any concentration used (Supplementary Fig.  S4B and S4C). This indicated that MYCMI-7 was able to inhibit TOP2A activity in vitro above 10 μmol/L.
To address whether the antiproliferative effects of MYCMI-7 in cells could be explained by DNA intercalation and/or TOP2-inhibitory activity, we first compared its ability to downregulate MYC mRNA expression (Fig. 5A), which is one well-known effect of DNA intercalators/topoisomerase inhibitors such as doxorubicin and etoposide (41,42). MYC expression was strongly reduced by 1 μmol/L doxorubicin, and also by 5 μmol/L etoposide in MCF7 cells (Fig.  5A), while 5 μmol/L MYCMI-7 had only minor effect on MYC expression at concentrations where it readily induced growth arrest in the cells in agreement with Fig. 2B, suggesting that MYCMI-7 acts through a distinct mechanism. To investigate whether TOP2A is required for MYCMI-7 activity, we made use of HCT116TOP2A_mAID cells with auxin-regulatable destruction of TOP2A (23,43). Pretreatment with auxin to deplete TOP2A had no effect on MYCMI-7-induced inhibition of MYC:MAX interactions in the cells as measured by isPLA ( Fig. 5B and C), nor did auxin treatment of HCT116TOP2A_mAID cells or siRNA-mediated knockdown of TOPA in MDA-MB-231 breast cancer cells affect the growth-inhibitory effect of MYCMI-7 ( Fig. 5D; Supplementary  Fig. S4D), arguing against a mechanistic involvement of TOP2A in the action of MYCMI-7. Furthermore, in contrast to ellipticine and the TOP1 inhibitor camptothecin, which induced p53 expression as expected, 5 μmol/L MYCMI-7 had no effect on p53 expression early (3 hours) and only a minor increase late (24 hours) after treatment in HCT116 cells and in MYCN-amplified Kelly neuroblastoma cells ( Fig. 5E; Supplementary Fig. S4E). We also investigated the p53 and DNA damage response (DDR) response in A375 melanoma cells, which are more sensitive to MYCMI-7-induced growth arrest and apoptosis than HCT116 cells (Fig. 4C vs. Fig. 5D). While 5 μmol/L ellipticine strongly induced phosphorylation of ATM in parallel with increased p53 expression, indicative of DDR signaling, treatment with 5 μmol/L MYCMI-7 showed only a slight increase in these markers, and no p53 or p-ATM induction was observed at concentrations of MYCMI-7 that readily induced growth arrest and apoptosis ( and Figs. 4C and 5F). To address whether p is required for the effects of MYCMI-7, we utilized wt and p53-deficient HCT116 cells. Titration of MYCMI-7 in these two cell lines showed no difference with respect to growth response (Fig. 5G). Western blot analysis showed that MYC expression was significantly downregulated upon 5 μmol/L MYCMI-7 treatment both in pproficient and p-deficient HCT116 cells, although to a somewhat lesser extent in p-deficient cells ( Supplementary Fig. S4F and S4G). Ellipticine did not affect MYC expression at 5 μmol/L, but downregulated the MYC level significantly by around 50% at 10 μmol/L in p wt cells, while it was unaffected in p-deficient cells.
In conclusion, although MYCMI-7 can inhibit TOP2A in vitro at higher micromolar concentrations, there were very little signs of such effects in cells at relevant concentrations, as evidenced by the lack of DDR signaling, p53 induction, or MYC mRNA reduction. Furthermore, the anti-MYC activity of MYCMI-7 was not dependent on TOP2A or p53.

MYCMI-7 Is a Potent Inhibitor of Ex Vivo Growth of Primary Patient-Derived Glioblastoma and AML Tumor Cells
Next, we investigated the efficacy of MYCMI-7 using primary patient tumor samples in culture. An ex vivo screen of cells derived from primary glioblastoma tumor biopsies of 42 patients, representing different subtypes of glioblastoma (proneural, neural, classical, and mesenchymal; ref. 44), was performed in 2D cultures. MYCMI-7 was very potent with GI 50 in the submicromolar range, and did not seem to discriminate between subtypes ( Fig. 6A; Supplementary Table   S1). Furthermore, we found no significant correlation between MYC mRNA levels and MYCMI-7 response in this case (Supplementary Table S1). For growth was monitored by the WST-1 assay. One representative biological experiment out of three, performed in triplicates is shown. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. The statistical analysis was performed using t test.
future studies, it will be important to also collect data on MYC protein levels to include in such correlations. MYCMI-7 also showed potent dose-dependent inhibition of growth of four primary patient-derived AML cell cultures as well as of established AML/CML cell lines, with GI 50 ranging from 0.15 to 1.3 μmol/L, and were in all cases more efficient than the MYC:MAX inhibitor 10058-F4, the bromodomain inhibitor JQ1, and cisplatin (the latter used in one case; Fig. 6B and C).

MYCMI-7 Reduces Tumor Burden in a MYC-Driven AML Mouse Model
Encouraged by the results in established tumor cell lines, primary patientderived tumor cells, and normal cells, we next decided to apply MYCMI-7 in vivo in mice. First, we performed a pharmacokinetic study of the behavior of the molecule in healthy mice. Analysis by mass spectrometry of plasma samples collected at 1, 2, 4, and 24 hours after intraperitoneal (i.p.) injection of MYCMI-7 at a concentration of 6.25 mg/kg body weight showed an estimated half-life of 1.5 hours for the compound in plasma ( Supplementary  Fig. S5A).
We next investigated potential antitumor effects of MYCMI-7 in vivo. First, we utilized a MYC/BCL-X L -driven AML mouse tumor model (28,45). Purified AML cells from spleens of moribund mice placed in culture were highly sensitive to MYCMI-7 treatment, with an GI 50 less than 1 μmol/L ( Supplementary  Fig. S5B). After tail vein injection of the purified AML cells into sublethally irradiated syngeneic recipient mice, leukemic blasts were first observed in blood smears at day 8, at which point the mice were treated with 12.5 mg/kg body weight MYCMI-7 daily by intraperitoneal injection. This higher but still tolerated dose was chosen due to the high turnover rate of the compound in plasma ( Supplementary Fig. S5A). The mice were then sacrificed according to scheme illustrated in Fig. 7A. GFP-positive leukemic cells were hardly detectable by flow cytometry in the bone marrow at day 11, but reached around 4% at day 15 in vehicle-treated mice, while still undetectable in the MYCMI-7-treated mice, (Fig. 7B, top and middle). At the end point (20 ± 4 days), bone marrow of vehicle-treated mice consisted of around 40% leukemic cells, which was significantly less, at around 10%, in MYCMI-7-treated mice (Fig. 7B, bottom). Similar results were obtained from the spleen (Supplementary Fig. S5C). Interestingly, spleens of MYCMI-7-treated mice retained a more normal histologic spleen structure compared with the collapsed structure of vehicle-treated mice (Fig. 7C). Furthermore, MYC expression was strongly reduced in MYCMI-7-compared with vehicle-treated animals in leukemic cells, as determined by IHC, suggesting that MYCMI-7 reaches its target in vivo ( Fig. 7C and D). Western blot analysis showed an increased expression of cleaved caspase-3 but also of H3K9me3 in leukemic spleens of mice treated with MYCMI-7 compared with vehicle (Fig. 7E), suggesting concurrent induction of both apoptosis and senescence. Despite this, there was no significant difference in mouse survival between the treatments (Supplementary Fig. S5D). Furthermore, there were no signs of severe side effects of MYCMI-7 treatment; all mice remained healthy and retained their weight over time of the experiment (Supplementary Fig. S5E, see Materials and Methods for mouse safety parameters applied).
In conclusion, MYCMI-7 treatment delayed onset of AML, decreased tumor burden, reduced MYC expression, and induced apoptosis and senescence markers in leukemic cells, but did not improve overall survival. One should, however, bear in mind that this is an extremely aggressive mouse tumor model.

MYCMI-7 Reduces Tumor Burden and Increases Survival in Xenograft Tumor Models of Basal-Like Breast Cancer and MYCN-amplified Neuroblastoma
We next studied the effects of MYCMI-7 in a xenograft model of the human breast cancer cell line MDA-MB-231. MDA-MB-231 was chosen because it represents basal-like, triple-negative breast cancer, which is a subgroup of breast cancer with frequent MYC amplification and/or high MYC pathway activity (2).
This subgroup, including MDA-MB-231 cells, has also been reported to be vulnerable to MYC depletion (46)(47)(48). The cells were highly sensitive to MYCMI-7 treatment in cell culture ( Supplementary Fig. S6A), with a GI 50 around 1 μmol/L. NOD/SCID mice with established MDA-MB-231 xenograft tumors were treated with 6.25 mg/kg MYCMI-7 or vehicle twice weekly by intratumoral injection. In the solid tumor models with localized disease (in contrast to AML with systemic spread), this route of administration was chosen due to the high turnover rate of the compound in plasma (Supplementary Fig. S5A). After a few days of MYCMI-7 treatment and onwards, tumor growth slowed down considerably compared with vehicle (Fig. 8A), and resulted in a significantly increased survival of the mice (Fig. 8B). Hematoxylin and eosin (H&E) staining of tumor areas showed extensive necrosis/apoptosis in MYCMI-7-treated mice ( Supplementary Fig. S6B). Furthermore, IHC staining of tumors revealed reduced expression of MYC and increased caspase-3 expression in response to MYCMI-7, indicative of apoptosis induction ( Fig. 8C; Supplementary Fig. S6C and S6D). The latter was also supported by increased TUNEL staining ( Fig.  8D; Supplementary Fig. S6E). The tumors were also characterized by reduced proliferation and microvascular density as determined by immunofluorescence staining of Ki67 and CD31 ( Fig. 8D; Supplementary Fig. S6F-H), which are both typical characteristics of MYC inhibition in vivo (8,49). control. At the end points, the last injection was done 3 hours before sacrifice. The statistical analysis was performed using t test.
In biochemical SPR assays, MYCMI-7 was shown to bind directly to the bHLHZip region of MYC with an affinity of approximately 4 μmol/L, which is in good agreement with the cellular GLuc, isPLA and co-IP data (Fig. 1K). By comparison, this is slightly lower affinity than observed for MYCMI-6 (K d = 1.6; 8), but at least a magnitude higher than what was reported for 10074-G5, 10058-F4, and #474 (51) (10), although one should bear in mind that this is not a direct binding assay and therefore may be difficult to compare.
Subsequent to MYC:MAX inhibition, MYCMI-7 decreased the steady-state levels of MYC protein (but not mRNA), at least in part through increased protein turnover (Fig. 2). The exact mechanism behind this is unclear, but has also been observed with some but not all other MYC:MAX inhibitors (9, 10, 51, 54), as well as after deletion of MAX (55) (Fig. 3). This was further supported by the significant correlation between the levels of MYC mRNA/protein and the response to MYCMI-7 within the NCI-60 tumor cell line panel. In addition, RNA-seq data showed that the MYC and E2F pathways were downregulated, while pathways connected to immune response was upregulated in response to MYCMI-7. The latter is expected considering the role of MYC in suppressing immune surveillance (35,36), and is consistent with the recently reported effects of the MYC:MAX inhibitors MYCi361 and MYCi975 (10).
The antitumor cell growth efficacy of MYCMI-7 was in the low single-digit micromolar range in most cases and even lower in 3D cultures, which is well in agreement with its efficacy toward MYC:MAX interaction in cells. This is in a similar range as reported for MYCMI-6, certain analogues to 10074-G5, KJ-Pyr-9, sAJM589, MYCi361, and MYCi975, while it is being much more potent in comparison to the "first-generation" MYC:MAX inhibitors, 10058-F4 and 10074-G5 (8)(9)(10)(11)18). The GI 50 for MYCMI-7 toward patient-derived primary glioblastoma and AML ex vivo cultures were in the range of 150 nmol/L-1.3 μmol/L (Fig. 6), demonstrating an excellent efficacy that shows promise for further investigation.
Importantly, while MYCMI-7 induced cell death/apoptosis in tumor cells, it was not cytotoxic to a range of normal primary cells, including human and murine fibroblasts, human melanocytes, and human peripheral blood lymphocytes, where it instead induced G 1 arrest, indicating that MYCMI-7 is nontoxic to normal cells at active concentrations (Fig. 4). It has been observed previously that inhibition of endogenous MYC by Omomyc in mouse models induced apoptosis in tumor cells, but only a reduction in proliferation in normal tissues, which was reversible upon MYC reactivation (57). This difference in response to MYC inhibition in malignant and normal cells has been described as "oncogene addiction" (1, 4).
Because MYCMI-7 has a structural resemblance to ellipticine, which has been described as a DNA intercalator and TOP2 inhibitor (39), we were concerned that MYCMI-7 might have similar activities. MYCMI-7, in contrast to doxorubicin, did not inhibit TOP2A activity until reaching concentrations above 10 μmol/L ( Supplementary Fig. S4). However, when investigating cellular responses to MYCMI-7, it clearly differed from DNA interactors and/or TOP1/2 inhibitors such as ellipticin, doxorubicin, etoposide, camptothecin. In contrast to the latter drugs, MYCMI-7 did neither downregulate MYC mRNA expression, nor induce p53 or phosphorylation of ATM at active concentrations, suggesting that it is not a potent inducer of DNA damage responses (Fig. 5). Furthermore, the growth-inhibitory/apoptotic activity of MYCMI-7 was not dependent on p53 or TOP2A activity, suggesting that the anti-MYC activity of MYCMI-7 is not related to DNA intercalation, TOP2A inhibition, or DDR signaling.
The in vivo potential of MYCMI-7 was investigated in three MYC-driven mouse tumor models: AML, breast cancer, and MYCN-amplified neuroblastoma. Despite having a half-life in plasma of approximately 1.5 hours, MYCMI-7 treatment reduced tumor volume significantly in all three models and increased overall survival in two of three models (Figs. 7, 8), with exception of MYC/BCL-XL-driven AML. The latter is surprising considering the clear effect of MYCMI-7 on tumor load. One should remember that this is an extremely aggressive model (28,45), and we speculate that the rapid expansion of leukemic cells in several organs together with the massive apoptosis (and senescence) of such cells after MYCMI-7 treatment may cause systems collapse leading to death despite effects of treatment. In all three tumor models, MYCMI-7 treatment led to reduced MYC or MYCN expression in tumor tissue. It also led to massive induction of apoptosis/necrosis and reduced tumor cell proliferation, which are typical outcomes when MYC is inactivated in transgenic mouse tumor models with inducible MYC (49,(57)(58)(59), as well as after treatment with MYCMI-6, 10008-F4, and with Mycro3 (8,54,60). In the solid tumors, there was a reduction in microvascularity and signs of increased hemorrhage, which might reflect collapse of tumor vasculature previously observed in the Omomyc tumor model after MYC inhibition (49) and after MYCMI-6 treatment in a MYCN-amplified neuroblastoma xenograft mouse model (8).
Inhibition of tumor cell growth in vivo using mouse tumor models has been reported previously for the MYC:MAX inhibitors MYCMI-6, 10058-F4, KJ-Pyr-9, KSI-3716, Mycro3, MYCi361, and MYCi975 (8, 10-12, 54, 60). In these studies, MYCMI-6 and Mycro3 were shown to reduce activity or expression of MYC in tumor tissue. Importantly, MYCMI-7 treatment was well tolerated by the mice, which did not lose weight after daily systemic treatment of 12.5 mg/kg body weight, and there were no signs of other severe side effects.
In summary, we show that MYCMI-7 is a direct MYC-binding compound that potently and selectively inhibits MYC:MAX interaction and MYC-mediated gene regulation and tumor cell growth in a MYC-dependent manner in culture and in vivo, while sparing normal cells. MYCMI-7 shows similar potency and selectivity as MYCMI-6 in vitro, in cells and in vivo, but in contrast to MYCMI-6, it also reduces MYC protein stability, and therefore potentially has higher therapeutic efficacy. MYCMI-7 together with other MYC inhibitors thereby contributes to a "tool box" with different types of MYC inhibitors that can used not only in MYC research but most importantly can be developed further and potentially could be utilized for treatment of different types of MYC-driven cancers alone or in combination. Considering the significant discrimination of MYCMI-7 between MYCN-amplified and nonamplified neuroblastoma and between the NCI-60 cancer cell lines with "high" versus "low" MYC expression, we anticipate that both MYC amplification and elevated MYC expression, can be used as biomarkers for identification of patients with cancer likely to benefit from MYC inhibitor treatment. MYC family gene amplification is frequent in neuroblastoma, ovarian cancer, basal-like breast cancer, lung, colon, pancreatic, and other cancers, MYC translocation is observed in Burkitt's lymphoma and some other lymphomas, and deregulated MYC expression is frequent in many different types of cancer (2). Hopefully, MYCMI-7 together with other MYC inhibitors can pave the way for the development of clinically relevant anti-MYC therapy for these and other types of cancer in the future.