Identification of Small‐Molecule Inhibitors of the Antiapoptotic Protein Myeloid Cell Leukaemia‐1 (Mcl‐1)

Abstract Protein–protein interactions (PPIs) control many cellular processes in cancer and tumour growth. Of significant interest is the role PPIs play in regulating apoptosis. The overexpression of the antiapoptosis regulating Bcl‐2 family of proteins is commonly observed in several cancers, leading to resistance towards both radiation and chemotherapies. From this family, myeloid cell leukemia‐1 (Mcl‐1) has proven the most difficult to target, and one of the leading causes of treatment resistance. Exploiting the selective PPI between the apoptosis‐regulating protein Noxa and Mcl‐1, utilising a fluorescence polarization assay, we have identified four small molecules with the ability to modulate Mcl‐1. The identified compounds were computationally modelled and docked against the Mcl‐1 binding interface to obtain structural information about their binding sites allowing for future analogue design. When examined for their activity towards pancreatic cell lines that overexpress Mcl‐1 (MiaPaCa‐2 and BxPC‐3), the identified compounds demonstrated growth inhibition, suggesting effective Mcl‐1 modulation.

Protein-protein interactions (PPIs) are involved in many cellular processes [1] and have therefore emerged as an attractive drug target in recent years. Specifically they have been shown to regulate several processes within cancera nd tumourg rowth, [2] and targeting PPIs is considered ap romising strategy towards next generation anticancer therapeutics. [3] However,P PIs pose ac onsiderable challenge to the medicinal chemistry community due to their large, flat, shallow interfaces, which possess ah igh degree of flexibility and thus are deemed problematic for drug design. [4] Despite this, small molecules have proved successful at modifying their actions. [5] In particular,m odulation of the p53-MDM2 interaction and Bcl-2 familyi nteractions has been achievedb ydrug candidates in clinical trials, overcomingthe perception that PPIs are "undruggable". [6] The PPIs of the Bcl-2 family play an important role in apoptosis as key regulators, ap rocess that is highly conserved and controlled. [7] The family consists of both pro-and anti-apoptotic proteins, and there is ac arefulb alance within ac ell that controlsi ts fate. [2a] It is believed that upon receipt of cellular stress, the proapoptoticp roteins BAX and BAK are activated by the BH3-only proteins,w here they migrate to the surfaceo f the mitochondria. Here, they form oligomers andi nsert themselves into the outer mitochondrial membrane forming pores This process is known as mitochondrial outer membrane permeabilisation (MOMP). This in turn leads to the rapid and irreversible release of cytochrome cf rom the mitochondria into the cytosol, which activates downstream caspases resulting in apoptosis. High levels of the antiapoptoticp roteins (e.g.,B cl-2, Bcl-xL and Mcl-1) are often observedi nc ancer, andt hey not only contribute to the development of the tumourb ut also conferr esistance to current therapies including chemotherapy and radiationt reatment. [2a] In particular,o verexpression of myeloid cell leukemia-1 (Mcl-1) is one of the most common forms of genetic abnormalitiesi nc ancer, [8] with av ariety of human cancers, including pancreatic cancer, exhibiting high levels of the protein. [9] The antiapoptotic Bcl-2 family of proteins are well-validated anticancer targets. The most successful small-molecule inhibitors to date, ABT-737 and its orally availablea nalogue ABT-263 (Navitoclax), inhibitB cl-2 and Bcl-xL with sub-nanomolar affinity. [10] Although ABT-263 has entered clinicalt rials, like most small-molecule Bcl-2 inhibitors, it does not inhibit Mcl-1 and lacks efficacy in tumours with high levels of Mcl-1r endering it ineffective as as ingle agent. [11] Furthermore, Mcl-1 overexpression has been linked to resistance observed against paclitaxel and vincristine, [12] as well as the first-line treatment for pancreatic cancer, gemcitabine. [9] Therefore, compounds that specifically target Mcl-1 have the potentialt oo vercome this resistance. The first selectiveMcl-1 inhibitor was identified as recently as 2010; [13] however,t here are currently no compounds undergoing clinicaltrials that target Mcl-1.
Pancreatic ductal adenocarcinoma has the lowest survival rates of any cancer. [14] According to Cancer Research UK, less than 4% of patients diagnosed with the diseasewill survive for at least five years, and this drops to less than 3% over at enyear period. [15] More worrying though is the factt hat these figures have not changed over the last 40 years despite the research efforts of many groups.R ecent studies have shownt hat downregulating Mcl-1 enhances the sensitivity of humanp ancreatic cancerc ells to gemcitabine and radiation,r esulting in increased levels of apoptosis. [9,16] Furthermore, knockdown of Mcl-1 in pancreatic cancer cellst reatedw ith ABT-737 triggers apoptosis, indicating Mcl-1 as an important and significant therapeutic target in this type of cancer. [17] The Bcl-2 family has proteins that regulate the activity of Bcl-2, Bcl-xL and Mcl-1, including Bim, Bid, Puma, Bad, and Noxa. Of the apoptosis regulator proteins, Noxa displays the greatest selectivity towards Mcl-1,b inding exclusively to Mcl-1 and Bfl-1/A1. [18] Interestingly,t ot he best of our knowledge, only one other group has exploited this selectivity to explore small-molecule Mcl-1 binding. [19] Herein, we reportt he exploration of the Mcl-1 binding pocket and the identification of novel leads for Mcl-1 inhibition utilising the binding domain of the selective apoptosis regulating protein Noxa.
To identify potential inhibitors of Mcl-1, we employed acompetitivef luorescence polarization( FP) assay similar to the one we reported for identifying inhibitors of the p53-Mdm2i nteraction. [20] The assay utilises the 19-residue alpha helix binding domain of NoxaB (AAQLRRIGDKVNLRQKLLN)t agged on the Nterminus with fluoresceinisothiocyanate (FITC) and measures the competitive bindingb yt he displacement of the tagged peptidefrom Mcl-1, resulting in an increaseinf luorescencepolarisation. The NoxaB peptide AAQLRRIGDKVNLRQKLLN was synthesised on Rink amide resin to generate the amide at the Cterminus. Fluorenylmethyloxycarbonyl (Fmoc)-aminohexanoic acid was subsequently coupled to the Nterminus followed by coupling with FITC to generate the fluorescently tagged NoxaB peptide( FITC-NoxaB). Ac himeric mouse/human Mcl-1 protein, previously reported by Colman [21] and used in an FP assay reported by Yu and Wang, [22] was employed by us in the polarization assay. The chimeric Mcl-1 protein has good solubility in water and maintains the biological function of human Mcl-1, with the BH3 binding groove consisting entirely of the human Mcl-1 sequence. An acetylated NoxaB peptide withoutthe aminohexanoic acid or FITC tag was used as ap ositivec ontrol and exhibiteda nI C 50 value of 0.65 mm and a K i value of 0.22 mm.T o confirm the reproducibility of our data, aZ -primet est was undertakenp roducing ar esulto f0 .78 indicating the assay is suitable for high-throughput screening.
Following optimisation of the FP assay,w es creened the US National Cancer Institute (NCI) diversity set IV for potential Mcl-1 inhibitors. The NCI diversity set is ac ollection of 1600 natural and synthetic compounds with ad iverses tructural landscape that have been evaluated as potential anticancer agents.C ompounds were screened initially at ac oncentration of 100 mm,a nd seven compounds were identified as potential hits (0.44 %h it rate). Af ull dose-response assay revealed that four of these compounds (Figure 1) displayed an IC 50 value of less than 20 mm in subsequent dose-response assays ( Table 1).
The structureo fp urpurogallin (1)f irst appears in the literature in 1882 and is obtained by the oxidation of pyrogallol. [23] The potential of purpurogallin as aB cl-2 familyb inder has been recognised since 2003, [24] and in 2011, it wasp art of ap atent that covers small molecules that modulate Mcl-1. [25] Compound 1 displayed sub-micromolar bindinga ffinity (K i = 0.80 mm)t owards Mcl-1.O ur findings support those already reported in the literature and demonstrate the ability of purpurogallin to bind competitively to Mcl-1. The structure of purpurogallin derivative 2 has, to the best of our knowledge, only appearedi nt he literature twice previously, [26] and here, we have identified it as aM cl-1 inhibitor with micromolar affinity (K i = 6.99 mm). The decreased binding affinity,w hen compared with compound 1,s uggests that the triphenol moiety is of importance for binding.
Redoxal (3)w as first reported in 1960 as ar edox indicator in alkalinem edia. [27] It is part of ap atentf rom Cadone and coworkers that details biphenazinec ompounds fort reatingh ematopoietic cancers. [28] Compound 3 possessed ab inding affinity in the low micromolar region (K i = 2.95 mm). This could suggest that the activity observedi nh ematopoietic cancers by Cardone could be ar esult of Mcl-1 modulation, as it has been widely reported that hematopoietic cancers cells survivefor extended periods due to Bcl-2 family overexpression. [29] Compound 4 was first prepared by Cain and Atwell in 1972 as apotential antitumour agent but was shown to be inactive in the lymphocytic leukaemia cell line L1210. [30] It has been identified as am odulator of protein-RNAi nteractions, specificallyt he Gag polyprotein and the viral RNA packaging signal. [31] Here, we have identified 4 as an ovel Mcl-1 binder with low micromolar affinity (K i = 1.09 mm). Of the identified compounds, only compound 1 approached the potency of the untagged NoxaB control (K i = 0.22 mm), but all compounds demonstrated sufficient activity to be worthy of consideration for further elaboration.  In order to rationalise the binding activity of 1-4 and to facilitate the structure-based design of analogues, computational docking experiments were performed.C omputations were based on the published structure of mouseM cl-1 bound to am odified Noxa BH3 peptide (PDB ID:2 NLA). [21] The modified Noxa and Mcl-1 make key interactions at the Mcl-1 amino acids Met 212, Lys215, Asn 223, Asp 256 and Arg263 (Figure 2a). [21] Compound 1 was predicted to bind in ag roove cre- www.chemmedchem.org ated by the BH1 and BH3 domains, displaying electrostatic interactions with Arg 222 and Val321 (Figure 2b). It sits in ah ydrophobic pocket that is also occupied by the mNoxa peptide. Compound 2 appears to bind in as imilar pocket to 1,s hifted slightly to form electrostatic interactions with Asn 223 (comparable to mNoxa) and His 224 (Figure 2c). The docking of compound 3 indicated that it might act as aB H3 mimetic, binding in the mNoxa bindingg roove created by the BH1, BH2 and BH3 domains. Electrostatic interactions with His 224 and Asn 260 were observed. The ligand is predicted to bind across the hydrophobic groove, and aromatic stackingi nteractions are potentially observedb etween 3 and Phe 319 (Figure 2d). Similarly,t he resultso ft he docking of 4 suggest that it too may be aB H3 mimetic, binding in as imilarm anner to compound 3.D ocking suggested electrostatic interactions to Arg 222 and the backbone of Gly219 (Figure 2e).
The diverse range of binding interactions predicted by molecular docking suggestst hat improvements could be made to all compounds. Ac ombination of the bindersc ould be explored,atechnique that has been shown to be highly successful at targeting protein-protein interactions. [11,32] Initial considerations would explore the incorporation of the triphenol moiety of 1,w hich appearst op lay an important role in binding, into compound 4.A dditionally,t he alteration of aromatic groups in 4,u tilising palladium cross coupling chemistry during synthesis, could be exploited to increase interactions with Phe 319, an interaction observed in the binding of 1 and 3.T he studies also suggest that potential ligandsc ould perform well as selectiveM cl-1 binderse ven if they do not mimic the BH3 domain, and that hydrophobic binding may prove to be more effective than electrostatic interactions.
Finally,w ee xamined compounds 1-4 fort heir ability to inhibit the growth and induce cell death in pancreatic cell lines MiaPaCa-2 and BxPC-3.B oth cell lines showa ni ncreased expression in Mcl-1 and Bcl-xL, but only MiaPaCa-2 shows an increase in Bcl-2. [33] As ummary of the cell growth inhibition is presented in Table 1. Compounds 2 and 3 displayed no significant activity towards either of the cell lines, with IC 50 values greater than 100 mm.C ompound 1 inhibited the growth of Mi-aPaCa-2 with an IC 50 value of 17.6 mm,b ut displayed inhibition greater than 100 mm towards BxPC-3. Compound 4 demonstrated inhibition in both MiaPaCa-2 and BxPC-3, with IC 50 values of 88.8 mm and 15.1 mm,r espectively.T hese results could indicatet hat 1 is capable of modulating Bcl-2 in pancreatic cancer,b ut is ineffective towards Mcl-1. Compound 4 shows micromolar inhibition in both cell lines, perhaps demonstratinga na bility to effectively modulate Bcl-2 family interactions. In fact, Ta kahashi and co-workers have recently reported that Bcl-xL and Mcl-1 might co-operatively play ar ole in the apoptotic cell deatho fp ancreatic cancer andt hat targeting both proteins may be av iable therapeutic strategy. [33] In summary,w eh ave demonstrated that the selective Mcl-1 ligand Noxa can be exploited to identifyr egulators of the antiapoptotic protein Mcl-1. The modified Noxa peptide was utilised in af luorescencep olarization assay to screen 1600 compounds, identifying four hits with binding affinities of less than 10 mm.T hese compounds were used to explore the bind-ing pocket of Mcl-1 computationally,a llowing for the binding of mNoxa to be compared with the identified hits (compounds 1-4), identifying potential synthetic enhancements for the novel binders 2 and 4.T he viability of cells treated with compounds 1-4 was examined, demonstratingt he ability of 1 and 4 to inhibit the growth of pancreatic cell lines that overexpress Mcl-1 (MiaPaCa-2 and BxPC-3). Examination of target selectivity, structure-activity relationship examination, and mode of cell growth inhibition of compounds 1-4 will be reported in due course.