The MYC transcription factor (also known as c-Myc) regulates the expression of genes controlling the growth and proliferation of cells. MYC was discovered almost 40 years ago as the cellular homolog to v-myc,
a viral oncogene from an avian myelocytomatosis virus that caused leukemia and sarcoma in chicken (Figure 1
]. Noticeably, v-myc
was the first retroviral oncogene to be found in the cell nucleus [3
], which hinted at its potentially direct role in gene regulation. Two additional human paralogs were eventually identified: MYCN (N-Myc) initially observed in neuroblastoma, and MYCL (L-Myc) identified in lung cancer samples [6
]. Both were later found to be expressed in many additional tissues and tumor types, and the nuclear localization was confirmed for the all Myc family protein members (MYC, MYCL, and MYCN, from now on Myc). MYCN and MYCL display mostly overlapping functions with MYC although with a more limited tissue-specific expression pattern. All Myc proteins are frequently deregulated in human cancers, where their expression level generally correlate with tumor aggressiveness [8
Initial analysis of the MYC sequence hinted, based on the homology with other transcription factors, at the possibility that it would bind to specific DNA sequences; however, when tested, MYC alone displayed only surprisingly weak DNA binding [10
]. It was the discovery of MYC’s obligate partner MAX (MYC-associated factor X) [11
] that enabled progress towards a better understanding of MYC biology (Figure 1
). Indeed, Myc is part of a network of transcription factors, the Proximal MYC Network (PMN). The PMN acts as a central hub in the nucleus, integrating signals from diverse upstream signaling pathways to coordinate and regulate the expression of thousands of target genes necessary for cell cycle progression, arrest/differentiation, and metabolism, among others [7
]. The members of the PMN, of which MAX is the central node, dimerize and bind DNA through a conserved bHLHLZ domain. The interaction of the heterodimers with the Enhancer box (E-box) elements in the promoters of target genes allows them to recruit multiple interacting proteins, leading to transcriptional regulation and active chromatin remodeling [12
]. Myc is generally considered a transcriptional activator, recruiting coactivator partners through its TAD domain, although it can also repress the transcription of some target genes [7
]. MAX proteins can form homodimers but are devoid of additional functional domain, and thus generate transcriptionally inactive complexes when binding to MYC-target promoters [12
]. The heterodimers formed by MAX with the MAX dimerization proteins X (MXD1, MXD3, MXD4), MAX-binding protein MNT and MAX gene-associated protein (MGA), constitute functional antagonists of Myc, shutting down the transcription of Myc-activated target by recruiting corepressor complexes (e.g., in the case of MXD1, 3, and 4, through their SID-mSin3 interacting domain) [12
In most normal cells, MAX is constitutively expressed [13
]. In contrast, quiescent cells express low or undetectable Myc levels, which are normally upregulated in response to mitogenic and development signals [7
]. Ectopic expression of Myc is sufficient to drive cell growth and proliferation, and it is the relative expression of Myc and MXD that determines the proliferation or differentiation fate of normal cells [12
]. Myc displays a short half-life, and its sub-cellular distribution, stability and degradation are finely tuned through multiple post translational modifications (PTMs) [14
] and the coordinated interaction with a vast number of cofactors [15
]. Unlike many other oncoproteins that promote cellular transformation following activating mutations (e.g., EGFR, Ras or B-Raf), Myc-driven cancers are virtually always due to its overexpression (e.g., following gene amplification) or deregulation (e.g., via tonic signaling from upstream growth pathways, or impaired degradation). Therefore, there is no real opportunity to target any cancer-specific mutant of Myc. Intriguingly, many, and perhaps all tumors appear to become “addicted” to its activity, and even short-term shutdown of its function leads to apoptosis and/or rapid tumor regression [16
Despite the huge body of literature collected since its discovery, our understanding of the molecular determinants underlying Myc function remains surprisingly limited, in part due to the challenges inherent to the study of intrinsically disordered proteins (IDPs). Nonetheless, the demonstration of the relevance of Myc as therapeutic target in cancer [17
] has provided significant drive to overcome the technical hurdles to identify potent and specific inhibitors [20
]. In this review, we summarize the structural and biophysical data that have unveiled distinctive features of Myc biology and some hints they provide to target it more efficiently.
3. Lessons Learned from the Direct Myc Inhibitor Screens
Classical approaches for drug discovery have focused on optimizing the interactions between small molecules (SM) and defined ligand binding or enzymatic pockets on globular proteins [100
]. However, this approach does not apply to MYC, which lacks binding pockets. Hence, MYC inhibition must rely on the disruption of protein–protein interactions (PPIs). Typically, PPIs encompass relatively large surface areas associated with high binding free energies [101
]. However, MYC also lacks significant secondary and tertiary structure when not complexed with one of its biological partners, and therefore does not display well-defined features to be specifically targeted by SM inhibitors. In fact, signaling proteins are often rich in intrinsically disordered regions (IDRs) and these can share a surprisingly high degree of similarity, which renders the identification of selective inhibitors particularly challenging. As a corollary, the screening efforts for such SM should be complemented with additional screenings against homologous or even non-related targets displaying ID regions in order to minimize the risk of off-target toxicity.
Because the MYC/MAX heterodimerization and DNA binding are essential for MYC-driven oncogenesis, the disruption of either interface constitutes an attractive approach, which has been sought by many groups [20
]. High-throughput screens have yielded many SM candidates able to prevent the dimerization of MYC with MAX. Such molecules were first described by Berg et al. [103
], who developed a fluorescence resonance energy transfer (FRET) assay to screen a library of 7000 peptidomimetic compounds, from which they selected five candidates. Four of those were further evaluated using EMSA and ELISA-based assays. The best ones, IIA6B17 and IIA4B20, showed a 2-fold lower IC50 for MYC compared to the homologous transcription factor Jun. Both compounds performed surprisingly well in cell-based assays, demonstrating that despite the limitations of such screening methods (i.e., irreversible formation of conjugated complex, slow maturation of the fluorescent signal and FRET interference from the auto-fluorescent compounds [104
]), they could still prove useful when using the appropriate orthogonal controls. Shi et al. eventually developed 120 analogues from IIA6B17 and IIA4B20, and screened them in colony-formation assay, from which they identified mycmycin-1 and mycmycin-2.
The same FRET approach was used to screen a library of 285 planar, hydrophobic SM, also termed “credit card” inhibitors. This yielded 40 shortlisted compounds, which were then tested by EMSA to confirm their ability to disrupt the MYC/MAX/DNA complex formation. Four candidates were further studied in a far-UV CD-based assay monitoring the structural changes accompanying their binding to synthetic MYC and MAX LZ-derived peptides tethered by a thioester ligation [105
]. Surprisingly, none of those compounds reduced the helicity of the heterodimeric LZ construct; instead, all maintained or increased the dimer helicity. In a biological assay, the most promising candidate, termed NY2267, appeared more selective towards MYC-transformed cells than Src, Jun, or PI3K-transformed cells. However, it was unable to discriminate between MYC and Jun luciferase reporters.
Yin et al. based their screening on yeast two-hybrid system (including 32 pairs of bHLHLZ, HLH, HLHLZ, or bLZ interactions), which inherently takes into account the cell-penetration capacity of the screened molecules. Out of 10,000 compounds tested, they identified a number of candidates with acceptable affinity for MYC/MAX, including 10074-G5 and 10058-F4 [106
]. The same group then used virtual screening to identify more potent 10058-F4 analogues. Binding to MYC was confirmed by monitoring changes in the intrinsic fluorescence of the compounds upon binding to a synthetic MYC bHLHLZ monomer (residues 353–439) [107
]. An initial selection of 48 candidates was eventually narrowed down to four SM showing higher potency (4.6–18 µM). In an optimization effort, the group combined the best attributes of these molecules to generate 17 new analogues. However, this effort did not result in any significant improvement, suggesting that these non-additive properties could very well simply represent the expected behavior of IDP regions capable of binding multiple structures with relatively weak affinities. In fact, two binding sites on the MYC bHLHLZ were identified for 10058-F4 and 10074-G5 through a series of point mutations of a C-terminally labelled MYC bHLHLZ construct and FP analysis of binding to p21MAX and to an E-box probe [108
]. Site I for 10058-F4 encompassed residues 402–409 and Site II for 10074-G5 included residues 366–375 (Figure 5
). However, competition assays taking advantage of the intrinsic fluorescence of 10058-F4 and 10074-G5 revealed that many non-fluorescent and non-structurally-related inhibitors could also efficiently bind the same sites on MYC [109
]. Accordingly, Heller et al. found that the binding of 10058-F4 to the monomeric MYC peptide 402–412 was highly diffuse and mainly driven by entropic contribution [110
]. Indeed, the heats of dilution measured by ITC at 25 °C and at 15 °C indicated a low enthalpic contribution to the binding. Titration experiments based on the intrinsic fluorescence of one Tyr residue within the MYC peptide sequence revealed an affinity of 14 µM. The van’t Hoff analysis indicated that the binding free energy (−27.6 ± −8.5 kJ/mol at 25 °C) is dominated by the entropic contribution (−20.7 ± −4.2 kJ/mol), although an enthalpic contribution is also present (−7.0 ± −4.3 kJ/mol). Consistent with previous studies [111
], molecular dynamics based on backbone NMR data recorded for the same complex [112
] indicated that the binding is diffuse, as MYC (402–412) remains disordered and 10058-F4 is delocalized across several residues.
A third binding site (Site III, residues 375–385) was identified for the non-fluorescent compound 10074-A4 using a CD-based binding assay and shorter MYC bHLHLZ variant sequences [113
]. Interestingly, there is very little similarity among the many “low micromolar affinity-displaying hits” identified by all those screening methods. Moreover, Hammoudeh et al. showed that the three sites can be bound simultaneously [113
]. Panova et al. used CDT-NMR to build a perturbation/binding map for SM inhibitors, which describes the stabilizing tertiary interaction between residues 360–380 and 400–410 of MYC and 10058-F4, with the hope of improving the selectivity of this scaffold [55
]. So far though, the SM deriving from the 10058-F4 appear to display very weak structure–activity relationship [107
], with most compounds binding to MYC with similar micromolar range affinities. Of note, these sites on MYC are poorly conserved in MYCL and MAX, which in a way provides some basis for specificity (Figure 5
Kiessling et al. established a FP-based assay to quantify the MYC/MAX bHLHLZ dimer formation and binding to a fluorophore-tagged E-box oligonucleotide, and assessed specificity by comparing it with Jun and C/EBPα homodimers binding to their respective DNA probe [115
]. Screening a 17,298-member library identified Mycro1, and a following analysis of structure–activity relationships (SARs) with commercially available derivatives eventually yielded the analog Mycro2. Mycro1 and Mycro2 inhibited DNA binding with IC50s of 30 μM and 23 μM, respectively, while for Jun and C/EBPα complexes the IC50 was >100 μM. The IC50s for MAX homodimeric binding to the same oligonucleotide were only 2–3-fold higher. In a follow-up study, the same group reported the results of a screen of a ~1700-member library of pyrazolo[1,5-α]pyrimidines based on the structures of Mycro1 and Mycro2, from which 5 candidates were selected that also displayed a selectivity factor of 2 for MYC/MAX heterodimers compared to MAX homodimers, revealing a weak specificity of the approach [116
]. Hart et al. used the same FP assay to screen a Kröhnke pyridine library, which yielded the KJ-Pyr-9 compound with a Kd for MYC in vitro of 6.5 nM. The KJ-Pyr-9 compound was shown to inhibit the proliferation of various cancer cells with IC50 values from 5 to 10 µM [117
]. To our knowledge, no additional study has been reported for this molecule.
As suggested by the NMR studies from Sammak et al. [27
], the plastic nature of MYC-H1 in the apo MYC/MAX dimer may constitute an attractive feature to exploit for targeting MYC with SM, which could trap it in a form unable to bind DNA. In fact, this strategy was the focus of a library of synthetic α-helix mimetics targeting the helical conformation of residues 363–381 of MYC, at the junction with the arginine-rich BR [118
]. The initial screen relied on EMSA, then multidimensional NMR spectroscopy of the MYC/15
N-MAX bHLHLZ was used to characterize more directly the specific binding of the selected candidates. This study concluded that the SM 4da specifically binds to the heterodimer and not to either monomer, and that this binding modifies the heterodimer conformation. Indeed, the structure stabilized by 4da is a heterodimeric form incapable of binding to DNA. SPR was also employed to verify the DNA binding impairment of the heterodimeric MYC/MAX complex following addition of 4da. Cell-based assays indicated an IC50 in the 10–20 μM range in MYC-dependent cancer cells. Of note, the early designs of helical peptidomimetics derived from the H1 of MYC and aimed at disrupting the MYC/MAX heterodimer were also found to directly bind to MYC bHLHLZ, to prevent MYC binding to DNA in EMSAs and to display with potencies of 1–10 µM in cell-based assays [21
Using similar approaches, Chen et al. [119
] initially screened 273 compounds by CD, monitoring the spectral change of MYC (370–409) upon addition of the compounds. From this initial screen, seven candidates were selected and further characterized by SPR to confirm their ability to prevent heterodimerization with MAX. The SPR competitive assay consisted of injecting serial concentration of a mixture of the best candidate with the MYC bHLHLZ onto a CM5 chip functionalized with a GST-MAX construct. NMR spectroscopy of unlabeled MYC bHLHLZ indicated that the addition of the SM PKUMDL-YC-1205 to the protein sample caused disappearance of the TOCSY crosspeaks corresponding to Arg372 Hβ-Hγ-Hδ and Ser373 Hα-Hβ. Molecular dynamic simulations also indicated that this SM binds to multiple conformations of the MYC bHLHLZ.
A cell-based protein-fragment complementation assay-high throughput screening platform was implemented by Choi and colleagues. The compound sAJM589 produced a dose-dependent inhibition of proliferation with IC50 values of ~1 μM in multiple cancer cell lines. Binding to MYC was confirmed by Biolayer interferometry using a biotin-tagged MYC bHLHLZ [120
]. The precise binding site of this molecule has not been described yet.
Han et al. reported a combination of in silico screening for a 16 million compounds library, from which they identified 61 hits then tested in EMSA to confirm their ability to disrupt the MYC/MAX/DNA complex [109
]. After evaluation of their ability to reduce cell viability and MYC-driven transcription of a reporter gene, one SM, called Min9 was selected for further testing. Evaluation of eight additional analogues of Min9 led eventually to the identification of 5 candidates with suitable properties. A cell line stably expressing a reporter luciferase plasmid was engineered and inoculated in a tumor xenograft mouse model. Three compounds caused a significant reduction in the MYC-driven transcriptional signal in vivo, with compound 361 being the most potent. To establish target engagement by the compound, a cellular thermal shift assay (CETSA) was developed. This assay monitors ligand-induced changes in protein thermal stability from cells treated with the test item, and confirmed that treatment with 361 destabilized MYC. A competition pull-down assay using Biotin-MYC revealed that 361 binds to the same site as 10074-G5 (residues 366–378). Perhaps unsurprisingly, the competitive FP assay indicated an affinity in the micromolar range (3.2 µM). The authors also showed that 361 treatment destabilizes MYC by selectively promoting Thr58 (but not Ser62) phosphorylation, and that this residue on MYC is essential to the mode of action of the compound. Indeed, mutations of Thr58 or Ser62 to Ala both abolished the anti-MYC activity of 361. Protein ligation assay (PLA) confirmed that 361 prevented the MYC/MAX dimer formation in cells even with the Thr58A MYC mutant. Hence, the disruption of the heterodimer appears to directly promote the phosphorylation on Thr58 and consequently leads to enhanced MYC degradation. Hart et al. also identified another compound with similar activity termed 975. Unbiased MS analysis revealed that both compounds are bound to approximately 135 proteins, of which ~38% belong to the MYC interactome.
Recently, our group demonstrated that the Omomyc mini-protein, a MYC dominant negative encompassing the MYC bHLHLZ with four mutations in its LZ [17
], constitutes a clinically viable direct Myc inhibitor [121
]. The structural characterization of Omomyc confirmed that, as previously published for its transgenic counterpart [17
], Omomyc can block Myc function by a three-pronged mode of action: Omomyc can form homodimers and heterodimers with MAX able to bind the E-boxes and displace Myc from the promoters of its target genes, while also forming heterodimers with Myc that are unable to bind DNA. The 1
N-HSQC and CD spectra and thermal denaturation curves of the bHLHLZ constructs evidence that the Omomyc homodimers and the Omomyc/MAX heterodimers display similar affinities in the absence of DNA, with Kd(37 °C) values of ~300 nM, indicating that both species coexist. Upon biding to DNA, the cooperativity of the transition and the melting temperatures of both the Omomyc homodimer and the Omomyc/MAX heterodimer mixtures increase dramatically, and show a transition half-point at ~70 °C. The affinity of those complexes for DNA, although not calculated, appears significantly higher than that of the MAX homodimer for DNA (hence lower than nanomolar Kd value). In contrast, while we detect evidences for the Omomyc/MYC heterodimer formation by NMR (through chemical shifts displacements) and by CD (via the increased helical signal intensity compared to the arithmetic sum of each sample), neither the CD spectra nor an FP assay allowed us to detect any significant binding to DNA of this heterodimeric form. Recently, Demma et al. reported that the binding of MYC, MAX, and of Omomyc to MAX occurs co-translationally, both for homo- and heterodimeric complexes [122
]. Moreover, the association of MYC and MAX proteins with the translating ribosomes was efficiently blocked upon treatment with Omomyc. In fact, in Omomyc treated cells, the biotinylated Omomyc and MAX both bound MAX RNA, while the binding of MYC to MAX RNA was ablated.
4. Structure–Function Relationship of an Intrinsically Disordered Transcription Factor
Regions of intrinsically disordered structure are highly abundant and occur in approximately one-third of eukaryotic proteins [123
] and in up to 60–80% of those involved in signal transduction. Under physiological conditions, IDPs exist as dynamic ensembles with minimal defined structure. IDRs (intrinsically disordered regions) constitute functional units acting in a disordered state, in which the polypeptide chain undergoes continuous conformational fluctuations. Their discovery in the 1990s abolished the dogma that, in order to accomplish their biological function, proteins required a folded structure [125
]. IDRs are often characterized by a relatively high net charge and low hydrophobicity, owing to their low content in hydrophobic side chains (which often drive a favorable entropic component for proteins to adopt a folded structure), and high content in hydrophilic and charged residues [126
]. The conformational status of IDPs ranges from a complete lack of secondary structure to a combination of residual or even significant small segments of secondary structure [127
]. The extended nature of ID regions and their surface-exposed area makes them suitable to coupled folding-binding reactions with diverse targets through the formation of large interfaces [129
]. Instead, other ID segments can constitute flexible linkers participating in the assembly of macromolecular arrays. Their intrinsic flexibility enables them to form complexes of various conformations with a multitude of partners, also termed fuzzy complexes [130
]. The high prevalence of hydrophilic residues in the fuzzy complexes is known to enable for a rapid and direct modulation of the interaction pattern through posttranslational modifications [131
These multiple weak and transient interactions by IDPs/IDRs also contribute to the formation of liquid–liquid phase separation (LLPS) [132
], a fundamental mechanism used by cells to isolate internal material and compartmentalize the intracellular space. LLPS occurs when a supersaturated solution spontaneously separates into two phases, a dense one and a more dilute phase, stably coexisting. The resulting condensates, also called coacervates when referring to oppositely-charged macromolecular species, can form membraneless organelles within cells (for instance Cajal bodies, P-bodies nuclear bodies, and granules). The absence of membrane allows the spontaneous exchange of components in response to alteration in the environment. LLPS plays a crucial role in many important processes, for instance by forming functional centers for biochemical reactions within the cytoplasm and the nucleus. In the nuclei of eukaryotes, LLPS can produce nuclear bodies that maintain, store and modify transcription regulators. Such condensates provide the ideal context for PTMs (e.g., acetylation, sumoylation) that will determine their function [134
]. Examples include nuclear speckles, polyleukemia bodies, nucleolus, histone locus and others [137
]. The relevance of LLPS in regulation of gene transcription has been evidenced multiple times [138
]. Among others, the dynamic association and dissociation events of nuclear condensates was found to regulate many processes associated with gene expression [139
] including chromatin structure organization [140
], RNA processing [141
] and ribosome biogenesis [142
Another singular aspect of IDPs is linked to their ability to bind multiple partners in dynamic interactions of modest affinities. Because of it, IDPs are especially prone to functional modulation by their concentration [143
]. In fact, at high concentrations, the mass-action drive can easily overcome the specificity of their interactions (the difference in energy between the desired and undesired binding interactions), leading to superfluous binding and to consequent toxicity. In line with this, the concentrations of IDPs are typically tightly regulated by transcript clearance, translation rate, and protein degradation.
The 40 years of intense research on Myc biology and characterization of thousands of its target genes converged into a common conclusion: Myc function is highly cell type and cell context dependent [68
]. Paradoxically to their well-defined sequence-specific DNA binding, in some conditions Myc transcription factors can bind to the promoters and intergenic regions of virtually all active genes in a given cell population, as well as to multiple enhancers [144
], even beyond E-box-containing regions [145
]. However, of all the active genes bound by Myc, only a small subset actually directly responds to the changes in Myc levels at its promoter [147
]. Indeed Myc proteins have also been described to enhance the overall rate of genome-wide transcription via Myc’s direct impact on RNA polymerase activity [148
], leading to a phenomenon termed global amplification. Nevertheless, tumors triggered by Myc and certain Myc-driven biological contexts present characteristic patterns of up-regulated and down-regulated gene subsets that suggest the existence of extra layers of specificities, which cannot simply be explained by the role of Myc as a global amplifier for genes transcribed by RNA polymerase II [147
]. Furthermore, many studies have indicated that the actual transcriptional output of specific promoters depends on the Myc levels, suggesting that promoters affinities account for specificity in Myc-dependent gene regulation, and the existence of productive and non-productive modes of DNA binding [146
]. By determining the nuclear MYC concentration and occupancy of every promoter at endogenous and exogenous MYC levels by ChIP sequencing, Lorenzin and co-authors [146
] estimated the relative affinity of MYC for all MYC bound promoters. The calculation of the concentration of MYC required for half-maximal occupancy of each promoter (EC50) was then used as a measure for the apparent binding affinity. Promoters with low EC50 values (high affinity) encompassed genes involved in RNA binding, translation, ribosomal and biosynthesis, and got saturated at values equivalent to low levels of endogenous nuclear Myc. Conversely, promoters linked to receptor activity, TGF-ß, or hypoxia presented significantly greater EC50 values, which translate into a substantial promoter occupation only when nuclear Myc levels are elevated [146
The following model can integrate to a certain extent the various structural and biophysical data gathered on MYC to date: Upon increased MYC expression (following growth signal induction), the spontaneous, co-translational and preferential heterodimer formation with MAX occurs at the ribosomes. The nascent MYC/MAX dimers readily translocate into the nuclei, and the superior affinity of the MYC/MAX dimer compared to MAX/MAX homodimer rapidly displaces MAX homodimers from E-boxes. The access to these genomic locations is facilitated by the relative accessibility maintained by the MAX homodimer through its IDP extremities. The increase in the DNA-bound MYC/MAX population contributes to grow the pre-existing condensates or coacervates through a local augmentation in the concentration of fuzzy complexes via the TAD of MYC [149
]. As the coregulators of transcription begin to distribute between the fuzzy complex units and the nucleoplasm, the chemical potential of the condensates changes. Early events consequent to this condensates increase would include, for instance, the stimulation of transcriptional elongation by promoting pause release of PolII via localization of P-TFEb (similar to the case observed for other transcription factors [138
]) and the phosphorylation of MYC Ser62 by CDKs. The newly synthesized RNAs, in turn, modify the local chemical potential and promote the formation of coacervates. Accordingly, MYC has been found to contribute to the formation of local loops that enhance the availability of polymerase-populated DNA loci through an ATP-dependent process. In this sense, in an energy-rich context, MYC has a dual role of active and passive factor of chromatin regulation [150
]. Additional phosphorylation of MYC continuously and spontaneously occurs, as the kinases continue to localize transiently among the fuzzy complexes and coacervates, according to their partition coefficient. Eventually, the prolyl-isomerase reaches the coacervate, leading to MYC phosphorylation on Thr58 by GSK3, which in turn promotes MYC ubiquitination [94
]. Concomitantly to these events, phosphorylation of MYC on its bHLHLZ (e.g., Thr400, Thr358, or Ser373) eventually occurs, destabilizing the DNA-bound complex and causing the detachment of MYC from MAX and from DNA (presumably, in this energy rich context, MAX also undergoes phosphorylation by CKII). At intermediate stages, this provides an additional kinetic drive to the MYC/MAX complex to move to other proximal genomic locations.
Other large protein partners with more defined folds such as WDR5 help direct MYC/MAX to specific genomic locations. This could lead for instance to longer residency time at those sites, effectively stimulating the recruitment of the transcriptional machinery and activation of select gene sets. Both events (increase in ubiquitination and detachment from the DNA) are necessary to export MYC to the cytoplasmic proteasome. As an additional fail-safe measure to ensure rapid termination of MYC signaling, its degradation by the proteasome can also occur in the nucleoplasm, at higher MYC expression levels [92
]. This cycle of events leading to MYC degradation happens spontaneously from its very initial translation into a polypeptide, and provides an automatic auto-destruction route to ensure rapid onset and termination of its signaling and transcriptional output, and maintain it under the control of the external growth signals. In this perspective, the higher residual helical fold of MYC bHLHLZ compared to MAX might simply reflect a structural restraint to limit the conformational adaptability to controls, such as its phosphorylation, lowering the entropy of this specific module. In such chain of events, the formation/expansion of the liquid condensates, which depends on the rapid and enthalpy-driven binding of MYC/MAX to DNA, constitutes an important rate-limiting step. It provides an entropic drive and change in the local chemical potential that stimulates the partition of the coregulators of transcription and post-translational modifiers according to their sequence-specified solubility for each phase. This chain reaction offers a robust yet elegant way to coordinate the myriad of biochemical and enzymatic reactions involved in transcriptional activation. The very presence of additional, tissue-specific transcription factors would be expected to similarly modulate the formation, growth and sub-nuclear localization of the phase condensates, thereby fostering a tissue-specific expression program. Moreover, impaired MYC degradation in an energy-rich context would be expected to provoke significant (de-compacting) chromatin remodeling and, if sustained, drive uncontrolled global transcription.
5. Cornering a Slippery Target
The biological relevance and extreme conservation of several Myc structural domains throughout evolution became evident since their early discovery. However, Myc peculiar conformational behavior and elusive—albeit crucial—role in growth and proliferation of cells posed the hardly soluble enigma of how its structure leads to function. The growing body of evidence from molecular and biophysical studies increasingly points at its contribution to the archaic and robust sub-cellular phase separation process as a mean to coordinate the plethora of proteins Myc is found to interact with. This macromolecular and biochemical perspective can also provide relatively simple explanations for the implications of Myc deregulation in tumor progression.
Currently, the principal clinical approach towards targeted therapies mitigating high MYC expression in cancer has focused on BET bromodomain inhibition. The BET inhibitors (BETi) act by impairing BRD4 regulation of the myc promoter, thereby suppressing its expression. However, the efficacy of BETi is limited to some contexts, and does not apply to the majority of tumors where MYC deregulation arises instead from growth factor-independent tonic signaling through MYC or from impaired degradation. Moreover, not all tumors that display myc genetic amplification respond to therapy.
Another recent approach proposed to target Myc indirectly involves targeting Myc interactors that do display an enzymatic active site (e.g., AURKA). However, there are some problems with this approach too: the targeting of kinases, albeit clinically successful in some cases, remains challenging due to the chemical similarity of the ligand binding site across the enzyme family members, and the limited number of H-bonds (critical to define selectivity) that SM can form within the active site. Second, such enzymatic functions are often highly redundant, and cells have the potential to rapidly develop resistance to their inhibition. Instead, Myc itself provides for a much more attractive target, since its function appears to be not redundant.
However, direct clinical Myc inhibition has been so far unsuccessful. Unfortunately, the few Myc inhibitors that have reached clinical trial have typically displayed high toxicity. It should be noted that these inhibitors have mostly been SMs, which often bind to their target in a diffuse manner, with very limited selectivity and consequent off-target toxicity. Hence, the development of larger molecules, capable of establishing sufficient number of specific interactions to significantly stabilize Myc in an inactive form appears a promising way forward. As these larger molecules can be limited in their cellular diffusion, the use of cell-penetrating and/or nuclear targeting signals is likely required to efficiently reach the desired cellular compartment. In this sense, the optimization of their partition coefficient could help maximize the residency time in the desired compartment. In this sense, the coupling of the dominant-negative and direct inhibitor function of MYC by molecules like the cell-penetrating Omomyc mini-protein certainly constitutes a promising approach [121
]. Importantly, when tested in vivo, Myc inhibition by Omomyc showed a wide therapeutic window, and no toxicity has been observed in mice, even after prolonged treatment following systemic expression in all tissues [17
Besides this approach, the development of other molecular scaffolds to target Myc are very likely to require high molecular weight protein domains or peptidomimetics. Also, further screening efforts should take into account, whenever possible, a more detailed knowledge of the nuclear environment (e.g., pH and chemical potential of the sub-nuclear condensates) and test binding to Myc in those conditions when possible.
In our view, the detailed structural characterization of Myc coupled to a broader view bridging physics, chemistry and biology dynamic macromolecular ensembles will provide a richer perspective to grasp the full range of Myc biological activity and pave the way to its successful rational targeting for cancer treatment.