G-quadruplexes (also known as G4-DNA) are tertiary structures formed in nucleic acids by sequences that are rich in guanine. The purine-rich strand in the NHE III(1) region of the Myc promoter forms G-quadruplexes (Simonsson et al., 1998, Figure 1). A number of small molecule ligands, including cationic porphyrins (Ou et al., 2007), quindolines (Pivetta et al., 2008), platinum complexes (Wu et al., 2009), and ellipticine (Brown et al., 2011) were shown to stabilize such G-quadruplexes in the myc gene, thus repressing its transcription. Notably, CX-3543 (Quarfloxin) was initally selected as a binder of the myc G-quadruplex (Chen et al., 2014) and is the only such G-quadruplex stabilizer to have reached clinical trials (entering Phase II trials for neuro-endocrine carcinomas in 2008). However, it was also shown to function by disrupting nucleolin bound to the G-quadruplexes in ribosomal DNA (Brooks and Hurley, 2009, 2010; Neidle, 2016). Since nucleolin binds to myc G-quadruplexes, Quarfloxin thus may also repress myc expression by a more indirect route (Brooks and Hurley, 2009). Development was discontinued by Cylene, although Quarfloxin has been licensed to TetraGene so development may continue.

Whether target specificity can be achieved using this strategy is not clear yet, although attempts are being made to selectively target myc (Felsenstein et al., 2016). According to many, total selectivity may not be necessary as long as the major target is a driver or provides a required function for the cancer cell, as for Myc (Neidle, 2016). Of course, as for some of the other types of inhibitors described here, the extent and severity of off-target side effects are key.

With the initial excitement over the promise of antisense as a tool to promote degradation of target mRNA, myc was first successfully attacked in vitro in multiple cell lines (Prochownik et al., 1988; Sklar et al., 1991, Figure 2A). Following these first successes, INX-3280, a 15-mer phosphorothioate oligonucleotide against the c-myc oncogene, was in Phase I and II clinical trials for the treatment of lymphoma and solid tumors more than a decade ago, but was discontinued in 2002 by Inex. A modified form incorporating a “transmembrane carrier system,” INXC-6295, was abandoned due to resource constraints.

With a slightly different strategy, AVI BioPharma (now Sarepta Therapeutics) developed AVI-4126 (Resten-RG), a phosphorodiamidate morpholino antisense oligomer (PMO) that inhibits Myc expression by preventing ribosomal assembly, thereby preventing mRNA translation (Arora et al., 2000). PMO is a modification shown to improve in vivo stability and bioavailabilty of the compound. A Phase I trial of Resten-NG carried out by AVI BioPharma enabled the determination of its bioavailability in solid tumor patients and established the feasibility of using PMOs in human cancer (Devi et al., 2005). Moreover, local delivery was feasible and safe in a related target disease, cardiovascular restenosis that involves neointimal hyperplasia (Kipshidze et al., 2003, 2004) and a Phase II study reported positive results following local delivery (Kipshidze et al., 2007; Philipp et al., 2012).

However, none of these drugs was further developed to reach the market and in fact very few antisense oligonucleotides have done so (Moreno and Pego, 2014). It is not clear though why the promising myc antisense approaches were not followed up, particularly in cancer studies and with newer nanocarrier delivery systems.

Another approach to directly inhibit Myc expression has been used successfully in vitro by lentiviral delivery of shRNA (Wang et al., 2008, Figure 2A) and also attempted in clinical trials using a lipid nanoparticle formulation to deliver myc RNAi (DCR-MYC). While preliminary trials provided evidence of destruction of myc RNA in patients and a clinical response (Tolcher et al., 2015), later trial data did not meet the company's expectations for level of knockdown or efficacy, and Dicerna has halted its development.

As a relatively poor pharmacokinetic profile seems to be a limiting development factor, many attempts are being made to overcome the rapid degradation of siRNA by its incorporation into nanoparticles. For example, gold particles modified with branched polyethyleneimine have been used as efficient and non-toxic intracellular delivery agents for c-myc siRNA both in vitro (Shaat et al., 2016) and in vivo, where they reduced lung tumor growth after intratracheal instillation (Conde et al., 2013). Similarly, folate nanoliposomes carrying siRNA targeting N-myc induce tumor cell apoptosis in a neuroblastoma model in vivo (Zhu et al., 2013), and lipid/calcium/phosphate nanoparticles combining c-myc siRNA and gemcitabine into a single nanovesicle were shown to inhibit lung tumor growth with little toxicity after systemic administration in xenograft models (Zhang et al., 2013).

A related approach making use of oncolytic viruses has also been used to successfully deliver N-myc siRNA in vivo to inhibit xenograft neuroblastoma tumor growth (Li et al., 2013). Oncolytic viruses are starting to show great promise in cancer treatment (Patel and Kratzke, 2013; Andtbacka et al., 2015).

The peptide mimetic compound IIA6B17 was the first identified small molecule inhibitor of Myc/Max dimerization (Berg et al., 2002). Unfortunately, the activity of IIA6B17 also extended to c-Jun (Berg et al., 2002), as did the inhibitory effect of NY2267 (Xu et al., 2006), likely due to their similar structural features in their leucine zipper.

Another compound identified early on to affect Myc/Max interaction was 10058-F4 (Yin et al., 2003). Chemical modifications of 10058-F4 resulted in improvements in efficacy in vitro, generally correlating with their ability to disrupt Myc/Max association and DNA binding. 10058-F4 and its active analogs bind specifically to monomeric Myc, interacting with the H2/leucine zipper domain with a K_D of 42 μM (Wang et al., 2007). Another small molecule arising from the same screen, 10074-G5, has a K_D of 20 μM in vitro and binds to a distinct region of Myc, the basic region/H1 domain (Yin et al., 2003). Binding of both these drugs to intrinsically disordered Myc limits its ability to adopt a more defined conformation and prevents interaction with Max (Follis et al., 2008). However, both 10058-F4 and 10074-G5 are rapidly metabolized and showed poor tumor distribution, limiting their applicability in vivo (Guo et al., 2009; Clausen et al., 2010; Fletcher and Prochownik, 2015).

In fact, the therapeutic utility of potent small molecule inhibitors of Myc/Max dimerization has so far been limited by poor bioavailability, rapid metabolism, and inadequate target site penetration. Nevertheless, development of these small molecules continues in the hope of improving their in vivo characteristics (for a summary, see Fletcher and Prochownik, 2015). For example, structure-activity relationship studies of 10074-G5 led to the generation of an analog, JY-3-094, showing higher ability to disrupt the association between recombinant Myc and Max protein (Wang et al., 2013; Yap et al., 2013), but did not solve the issue of poor cell penetration. Esterification of a critical para-carboxylic acid enhanced cellular uptake, although unfortunately it impaired the ability to disrupt Myc/Max association in vitro (Wang et al., 2013).

Also related to 10074-G5 is the small molecule 3jc48-3, a congener that shows increased potency and stability in cell-based assays (Chauhan et al., 2014). More exciting still is Mycro3, for which daily treatment by oral gavage increased survival in a mouse model of pancreatic ductal carcinoma and orthotopic xenografts of human pancreatic cancer cells (Stellas et al., 2014). Finally, KJ-Pyr-9 is a new inhibitor found in a Kröhnke pyridine library, with a notably lower K_D of 6.5 nM (Hart et al., 2014). In vivo, KJ-Pyr-9 effectively blocks the growth of Myc-amplified human cancer cell xenografts (Hart et al., 2014). Furthermore, it penetrates the blood-brain barrier and is not acutely toxic at doses as high as 10 mg/kg (Hart et al., 2014).

New small molecules that target Myc are also being generated thanks to novel computational techniques able to virtually screen binding to different intrinsically disordered protein conformations, maximizing the chances of structure-based drug discovery. Four compounds (with the prefix PKUMDL) show micromolar affinity for Myc and activity in cell-based assays (Yu et al., 2016).

An additional “reverse” approach involves stabilizing instead the Max/Max homodimers, thus preventing Myc/Max transactivating dimers from forming. A virtual ligand screen identified a potent stabilizer of the homodimer (NSC13728) that interferes with Myc-induced transformation and transcriptional activation (Jiang et al., 2009).

In addition, attempts are being made to incorporate the small molecules into nanoparticles for increased stability and targeted delivery. For example, an Sn2 lipase-labile pro-drug inhibitor (MI1-PD) conjugated to integrin-targeted nanoparticles extended survival in a mouse model of multiple myeloma (Soodgupta et al., 2015).

A more extensive review of these small molecule inhibitors is provided elsewhere (Chen et al., 2014). Such inhibitors have been an intense focus in Myc inhibition research for many years, and we hope that further preclinical development of promising leads is ongoing.

Other small molecule inhibitors such as MYRA-A and NSC308848 have achieved high selectivity in targeting the DNA-binding domain of Myc/Max, and preventing specific interaction with DNA (Mo and Henriksson, 2006; Mo et al., 2006, Figure 3).

Some naturally-occurring molecules have also been shown to directly interact with Myc/Max heterodimers. Celastrol and celastrol-inspired triterpenoids bind to and alter the quaternary structure of the pre-formed dimer and abrogate its DNA binding (Wang et al., 2015b).

KSI-3716 also blocks Myc/Max binding to DNA, and inhibited orthotopic tumor formation after local instillation to the bladder (Jeong et al., 2014)—the typical treatment route for this cancer type—even in gemcitabine-resistant tumors (Seo et al., 2014).

As mentioned above, though, small molecules, despite good affinity for their target in vitro, often display lack of selectivity in cells or in vivo. In order to try to overcome this issue, synthetic α-Helix “mimetics” based on biphenyl were developed (Jung et al., 2015). These have an increased interaction surface and recognize Myc/Max dimers (not free Myc) and disrupt DNA binding. However, their K_D was not increased (13 μM) and specificity was not enhanced: similar activity was observed in non-cancer cells lacking Myc overexpression (Jung et al., 2015).

Miniproteins or protein domains comprise a group of structurally-related molecules based on domains from Myc family members (Figure 3). The best characterized so far and especially notable for its in vivo use, is Omomyc. Omomyc has been well validated as a gene therapy, providing the proof of concept for the feasibility of systemic Myc inhibition. It comprises the bHLH-Zip domain of Myc carrying four aminoacidic substitutions that alter its dimerization specificity, such that in addition to binding Myc's natural partner Max, it can also heterodimerize with Myc as well as homodimerize (Soucek et al., 1998, 2002; Savino et al., 2011). As a result, Omomyc acts as a dominant negative of Myc transcriptional function, being able to disrupt Myc/Max interaction, sequester Myc away from DNA and occupy the E-box with transcriptionally inactive dimers (Omomyc/Omomyc and/or Omomyc/Max). Notably, in doing so, it antagonizes all Myc family members (Soucek et al., 2008; Savino et al., 2011; Fiorentino et al., 2016). Multiple studies in mouse models of cancer demonstrated Omomyc's therapeutic impact in different types of cancer, independently of their driving mutation or tissue of origin, pointing to the key role of Myc in tumorigenesis downstream of the diverse oncogenic lesions (Soucek et al., 2004, 2008, 2013; Sodir et al., 2011; Annibali et al., 2014; Galardi et al., 2016). Importantly, in each model Omomyc showed only minimal side effects, suggesting its safety and potential applicability in patients. Work to translate its use from gene therapy to pharmacological application is currently ongoing (Peptomyc S.L.).

Another modified Myc peptide, this time a 14 amino acid sequence from the helix 1 (H1) carboxylic region of Myc harboring 2 changes, was shown to be active against breast cancer cells in culture when fused to a fragment from Antennapedia to enable cellular uptake (Giorello et al., 1998). This has not performed well in vivo, at least partly because it does not efficiently cross the nuclear envelope. However, recently a staggered, “dual-strike” strategy was employed, whereby a first treatment with docetaxel arrested the cells in G2/M, prolonging the period of nuclear envelope disassembly, followed by a second treatment, this time with the H1 peptide (Li et al., 2016). In vivo, this procedure reduced the growth of subcutaneously-inoculated HeLa cells and prolonged animal survival (Li et al., 2016). The peptide was delivered in the macromolecular carrier HPMA.

The H1 peptide was also used for an in vivo study by fusing it to both a cell-penetrating peptide sequence and an elastin-like polypeptide (ELP) that is thermally stable and allows targeted delivery to particular tissues by local hyperthermia (Bidwell et al., 2013). This multi-functional peptide reduced tumor growth in rodent orthotopic models of glioma and breast cancer (Bidwell et al., 2012, 2013).

Another protein domain that could be used to inhibit Myc is the bHLH-Zip of Max (Montagne et al., 2012). This truncated protein spontaneously transduces into cells and inhibits Myc transcription. The idea behind this strategy would be to provide excess homodimeric Max to out-compete Myc/Max heterodimers binding to DNA.

The BET bromodomain and extra-terminal domain inhibitors are a significant area of focus at the moment and were originally described to target Myc expression. A selective small molecule inhibitor (JQ1) of BET bromodomains was unexpectedly found to downregulate Myc (Delmore et al., 2011). JQ1 displaces the bromodomain chromatin regulators from the large super-enhancers of genes connected with multiple myeloma, notably myc (Delmore et al., 2011; Loven et al., 2013, Figure 1). JQ1 preferentially impacts on myc transcription, causing cell-cycle arrest, and cellular senescence, as well as resulting in significant anti-tumor activity in mouse models of multiple myeloma, and xenograft models of Burkitt's lymphoma and acute myeloid leukemia, diseases in which the myc gene is amplified (Delmore et al., 2011; Mertz et al., 2011).

After the initial excitement about their potential for Myc inhibition, it is now becoming clear that JQ1—and other BET inhibitors—may function by inhibiting additional oncogenic factors besides Myc (Andrieu et al., 2016), whose levels remain unaffected in some cellular contexts (Ambrosini et al., 2015; Yao et al., 2015; Bid et al., 2016; Garcia et al., 2016; Hogg et al., 2016; Donato et al., 2017). Moreover, compensatory mutations have been identified that can prevent Myc downregulation by BET inhibitors and/or their therapeutic efficacy (Shimamura et al., 2013; Fong et al., 2015; Rathert et al., 2015; Shi et al., 2016). Nevertheless, several BET inhibitors are currently in early-phase clinical trials in various hematologic malignancies (Abedin et al., 2016). Some patients in dose-escalation Phase I studies have shown complete or partial remission of acute leukemia after treatment with OTX015 (Berthon et al., 2016), and data is currently available for 3 patients treated with TEN-010 in a study of NUT-midline carcinoma showing clinical responses in both patients receiving the higher dose (Shapiro, 2015). It remains to be seen how much of the effect in patients is actually due to Myc inhibition.

An unrelated approach to blocking myc transcription targets CDK7, a catalytic subunit involved in the transcriptional initiation complex that phosphorylates serine-5 of RNA pol II (Nilson et al., 2015). A covalent inhibitor (THZ1) was developed that showed selectivity for CDK7, and specifically downregulated Myc in neuroblastoma (Chipumuro et al., 2014; Kwiatkowski et al., 2014). While THZ1 selectivity for Myc is unclear and it is possible that its therapeutic impact is due to other targets as well, it is effective at treating lung cancer (Christensen et al., 2014) and triple-negative breast cancer cell lines (Wang et al., 2015c) as well as N-Myc driven neuroblastoma in mice (Chipumuro et al., 2014). Furthermore, the analog THZ2 was developed that has improved pharmacokinetics (Wang et al., 2015c).

Another approach to interfere with Myc protein expression is to block its translation (Figure 2). myc mRNA can be translated both by 5′ cap- and internal ribosome entry site (IRES)- dependent mechanisms (Nanbru et al., 1997). The central role of mTOR in mediating the translation of mRNAs such as myc via these mechanisms suggests that targeting mTOR or upstream controllers of its activity (PI3K/PTEN/AKT and Ras/Raf/MEK/ERK) could be a fruitful strategy, and there are multiple inhibitors of these pathways.

While inhibition of mTOR alone in a mouse model of colorectal tumorigenesis failed to inhibit translation of myc, a small molecule inhibitor of eIF4A, silvestrol, effectively reduced myc translation, inhibiting tumor growth (Wiegering et al., 2015). Multiple mTOR and mTORC1/2 kinase inhibitors are currently approved for clinical use, and there is a significant focus on targeting many members of these signaling pathways (Polivka and Janku, 2014; Roohi and Hojjat-Farsangi, 2016).

Again though, inhibitors of such targets will have effects not limited to myc alone.

Recent data also indicate that concomitant targeting of HDAC and PI3K would be beneficial to the treatment of Myc-driven tumors. In particular the use of CUDC-907, a small molecule inhibitor of both HDACs and class I PI3Ks, was shown to be effective in reducing the growth and survival of Myc-transformed cancer cells and demonstrated therapeutic impact in multiple mouse models of Myc-dependent tumors (Sun et al., 2016). CUDC-907 is currently in Phase II clinical trials to study and evaluate its efficacy and safety (alone or in combination with Rituximab) in patients with Relapsed/Refractory (RR) Myc-altered Diffuse Large B-Cell Lymphoma (DLBCL), including patients with Myc alterations.

Another recent translation inhibition approach made use of Src kinase blockade with the small molecule inhibitor saracatinib in preclinical studies in premalignant breast cells and tissue. Among other effects, saracatinib inhibited the ERK1/2-MNK1-eIF4E-mediated cap-dependent translation of myc (Jain et al., 2015).

The regulation of Myc stability is complex. Numerous studies propose the targeting of ubiquitinases or phosphatases for the degradation of Myc in cancer cells. Clearly, whether these can be specific for Myc and/or have sufficiently minimal side effects is still unclear.

Myc is ubiquitinated by a number of E3 ligases, such as SCF (FBW7) and SCF (Skp2). One approach is therefore to inhibit the deubiquitinases that help stabilize Myc, such as USP28, USP38, and USP36 (Sun et al., 2015). On the other hand, proteasomal degradation could be triggered, for example by oridonin, a natural plant diterpenoid that induces FBW7-mediated Myc breakdown (Huang et al., 2012). Oridonin derivatives are in clinical trials (e.g., HAO472 for leukemia treatment), although such anticancer drugs (and ubiquitin/deubiquitinases) have numerous other potential mechanisms of action and targets, not solely limited to Myc inhibition.

It has been shown that N-Myc complexes with the Aurora-A kinase and is thus protected from proteasomal degradation (Otto et al., 2009). Aurora-A inhibitors (MLN8054 and MLN8237) disrupt the complex and promote N-Myc degradation by FBW7 ubiquitin ligase (Brockmann et al., 2013). While clinical development of MLN8054 was terminated by Millenium in 2008 due to side effects (Macarulla et al., 2010), the more potent, selective, second generation inhibitor MLN8237 (Alisertib) is currently being evaluated in multiple Phase II and III studies (based on the obligate role of Aurora kinases in mitosis). Since N-Myc is a critical target of this class of Aurora-A inhibitors, perhaps it is possible to design allosteric inhibitors that more strongly distort Aurora-A and thereby more effectively disrupt the Aurora-A/N-Myc complex, while retaining the ability of Aurora-A to promote mitotic entry, thus finding a more specific inhibitor (Brockmann et al., 2013). Notably, the Aurora kinases have turned up in multiple screens for Myc synthetic lethal targets (see later section).

Another ubiquitin ligase—HUWE1 (HECTH9, ARF-BP1, MULE)—has also been targeted by small molecule inhibitors. These reduce Myc-dependent transactivation in colorectal cancer cells, but not in stem and normal colon epithelial cells, by influencing Myc and Miz1 degradation (Peter et al., 2014).

Alternatively, the tumor suppressor protein phosphatase 2A (PP2A) targets serine 62 of Myc and causes its inactivation and destabilization (Sears, 2004). Cellular inhibitors of PP2A (the SET oncoprotein and CIP2A, the cancerous inhibitor of PP2A) are increased in human cancers and lead to overexpression of Myc (Westermarck and Hahn, 2008). Thus, inhibitors of SET and CIP2A were used to reduce Myc expression and activity, decreasing the tumorigenic potential of cancer cells (Farrell et al., 2014; Janghorban et al., 2014). Inhibitors are in preclinical development; the effects of PP2A activation though would likely not be limited to Myc degradation alone, since other pathways would be targeted as well.