2.1. Assays

Descriptions of the assays follow Table 1.

2.2. Probe Chemical Characterization

The chemical structure of the probe was verified by analysis of its 400 MHz ¹H NMR spectra (Figure 1) obtained on a Brüker 400 MHz instrument. The chemical structure was also corroborated by its LC/MS molecular ion (calc for M+1: 416.2, found 416.2, using an Agilent 1200/6140 multimode quadrupole rapid resolve system) as well as its expected fragmentation patterns. Purity was measured at >95% (LC/MS analysis, confirmed by analytical HLPC analysis; HPLC purity data is shown in Figure 2). HPLC data was obtained using an Agilent 1200 analytical HPLC with an Agilent Eclipse XDB-C18 column, 4.6×150mm. The HPLC solvents used were acetonitrile and water with 0.1% formic acid added to each mobile phase as the pH modifier.

Figure 1

NMR Spectra ML311. Full spectrum (top) and aryl region expansion (bottom).

Figure 2

Analytical RP HPLC Spectra for ML311. At 254 nM, 280nM, and 320 nM, purity >95%.

The solubility of the probe in PBS at pH 7.4 was determined to be >2.7 μM. Solubility was higher (~20 μM) in conditions more reflective of the cell-based assay environment (DMEM, 10% FBS). Importantly, its solubility is fully adequate to provide the high potency seem in multiple cell-based assays (0.4–5 μM) and is also adequate for broad use as a biological probe to be used in a variety of aqueous-based media.

The probe has a half-life of 10 hours in PBS at room temperature when tested at 10 μM. Disappearance of the LC peak for the ML311 is not altered by addition of excess glutathione, indicating that ML311 is not thiol-reactive under physiologically-relevant conditions.

ML311 is stable in DMSO solution at room temperature (no erosion of peak intensity after 7 days) and is also stable as a free base dry powder.

NMR and HLPC spectra are shown in Figures 1 and 2.

The compounds listed in Table 2 have been submitted.

Table 2

Probe and submitted analogs.

2.3. Probe Preparation

General Synthesis Scheme. The probe compound and all analogs were prepared efficiently in a one step/one pot procedure from commercially available starting materials, as shown in the general scheme in Figure 3. In this procedure, termed the Betti reaction¹⁸ of 8-hydroxyquinolines¹⁹, an aldehyde A and amine B condense to form an iminium ion intermediate C, which then reacts with the 8-hydroxyquinoline D at the 7-position to give the product E. The probe analogs and SAR compounds were prepared by this procedure using available structural variants of reagents A, B, and D.

Figure 3

Betti reaction of 8-hydroxyquinolines (X = N): general synthesis scheme for preparing the probe ML311 and all analogs.

Detailed experimental procedures for the preparation of 7-((4-Ethylpiperazin-1-yl)(4-(trifluoromethyl)phenyl)methyl)quinolin-8-ol, or ML311. (See Figure 4.) The general synthesis scheme from Figure 3 was followed.

Figure 4

Detailed experimental procedures for the preparation of ML311.

A mixture of quinolin-8-ol (470 mg, 3.24 mmol), 4-(trifluoromethyl)benzaldehyde (564 mg, 3.24 mmol) and 1-ethylpiperazine (0.41 mL, 3.24 mmol) was heated overnight at 110 ˚C in an oil bath. The solution was cooled to room temperature and the mixture was precipitated in ethanol. The solid was collected by filtration to yield the title compound (471 mg, 35% yield) in high purity (>95%) as judged by ¹H NMR and LCMS analysis. The ¹H NMR spectrum and analytical HLPC trace were shown in section 2.2. Supporting data: MS (M+1) cal’d = 416.2, found 416.2. ¹H NMR (400MHz, DMSO-d₆) δ 10.17 (br s, 1H), 8.82 (dd, J = 1.6, 4.4 Hz, 1H), 8.25 (dd, J = 1.6, 8.4 Hz, 1H), 7.71–7.63 (m, 5H), 7.51 (dd, J = 4.2, 8.2 Hz, 1H), 7.39 (d, J = 8.8 Hz, 1H), 5.05 (s, 1H), 2.49–2.29 (vbr s, 8H), 2.29 (q, J = 7.2 Hz, 2H), 0.95 (t, J = 7.2 Hz, 3H).

3.1. Dose Response Curves for Probe

ML311 has robust on-target activity in biochemical assays (see Figure 5 for target and anti-target curves). The primary criteria of the probe report have been achieved: ML311 displays >100× selectivity in a biochemical assay (fluorescence polarization) for Mcl-1 over Bcl-xL. In addition, as shown in Figure 6, ML311 induces cell death in a highly active Mcl-1 cell line (Mcl-1/1780, EC₅₀ = 0.3 μM). The compound does not induce appreciable cell death in the anti-target cell based assay (DHL-10) even at the highest dose tested (5 μM), and in other runs for a previous batch (SID 131465452) this lack of anti-target effect was verified at higher concentrations (to 25 μM). There is thus at least 8-fold selectivity for the DHL-6 cell line versus the DHL-10 cell line, meeting the criteria established for probe in the CPDP of greater than 3-fold selectivity. ML311 is also approximately 4-fold more active in Mcl-1-1780 cells than vs. the highly Bcl-2 active cell line Bcl-2-1863 (EC₅₀ =1.1 μM), meeting another probe development objective. Batch-to-batch consistency in activity is seen in all assay formats, as indicated in Figures 5 and 6.

Figure 5

Primary target, biochemical assay (Mcl-1). Anti-target, biochemical counterscreen (Bcl-xL)

Figure 6

Cell-based assays. Targets and DHL-10 anti-target

3.2. Cellular Activity

The probe potently halts viability of several types of Mcl-1 primed cells, including MCL-1-1780 (EC₅₀ = 0.31 μM), DHL-6 (EC₅₀ = 3.3 μM), and NCI-H929 (EC₅₀ = 1.6 μM), with generally high maximal effect (>80%). All of the cell-based curves were shown previously in Section 3.1 (Figure 6).

ML311 also displayed activity in a leukemia-derived cell line particularly reliant upon Bcl-2 function (Bcl2-1863, EC₅₀ = 1.1 μM). Targeting of Bcl-2 is considered highly relevant for the treatment of a number of hematological malignancies, as evidenced by the successes observed with ABT-263 (Navitoclax) and related compounds. However, for the purposes of this probe development effort, some selectivity over Bcl-2 was desired, with a selectivity goal of 3X for Mcl-1 1780 versus Bcl-2 1863 cell lines. The exclusion of all Bcl-2 activity was not sought, merely the objective as set forth in the CPDP was to obtain compounds with Mcl-1 dependant effects and with a bias for Mcl-1 over Bcl-1, a profile unknown for a small molecule at the time when this effort was initiated. This goal was achieved with the development of ML311.

The finding that the DHL-10 anti-target is largely unaffected is highly significant (EC₅₀ >25 μM, Figure 6). DHL-10 is Bax/Bak deficient and is properly unaffected, since it is incapable of the Mcl-1/Bim interaction. This result shows the probe’s promise for useful on-target specificity and selective cytotoxicity in the mode of action.

To assess cell-based antitumor activity more broadly, ML311 has been submitted for analysis in the the NCI’s CTD2 (Cancer Target Discovery and Development program) through the Broad Institute and in the NCI60 panel. The CTD2 program involves screening 1,000 well-characterized cancer cell lines with results made publicly available.

The NCI60 screening results were completed and are summarized in Figure 7. ML311 has strong growth inhibitory effects in many cell lines, with GI50 <900 nM for nine cell types, and <2 μM for 14 additional types (not shown). Mcl-1 is known to be highly expressed in many forms of lung, prostate, breast, ovarian, renal cancers, and gliomas by mRNA analysis.³¹ Correlation of NCI60 activity shown in Figure 7 with intrinsic Mcl-1 mRNA levels is unclear, though mRNA levels are also not necessarily reflective of protein levels as determined by other means, such as Western blot analysis. Also certain tumor types with only modest Mcl-1 levels may be strongly affected by dampening the activity of Mcl-1 present.

Figure 7

NCI60 growth inhibition.

3.3. Profiling Assays

The cell-based assays used in screening and in support of our lead optimization efforts define a clear mode of action for ML311: disruption of the interaction of Mcl-1 with Bim in the BH3 binding region of Mcl-1. To support the selectivity for this mode of action hypothesis, we studied the properties of the probe and its precursors.

Analysis of PubChem assay results for the initial HTS hit, CID 2966040, suggests that this compound (and perhaps the scaffold itself) is not assay-promiscuous, with activity in 20 of 508 Pubchem assays (3.9%), when Mcl-1-related assays are excluded from the analysis. The lack of broad activity for this HTS hit was one factor leading to the selection of this scaffold for probe development.

The lack of activity of CID 2966040 and many of its analogs in the Bcl-xL counterscreen and the DHL-10 cell-based antitarget counterscreen also provides further evidence of target-specific activity. Analysis of the structure using online cheminformatics tools to assess shape similarity to known ligands (molinspiration bioactivity analysis) ³¹ suggests a low probability that ML311 is a significantly potent GPCR ligand, ion channel modulator, kinase inhibitor, nuclear receptor ligand, protease inhibitor, or other enzyme inhibitor.

To gauge selectivity in a broader sense, ML311 was submitted for analysis in a lead profiling screen with Ricerca, a screen profiling binding to 67 protein targets of therapeutic and/or toxicological interest (Figure 8). The probe is inactive vs. most targets. Moderate activity (50–80% at 10 μM) was noted for certain GPCRs and ion channels, as indicated. For a few targets, notably hERG, significant effects are seen. This potential for GPCR and ion-channel mediated effects should be considered when using ML311 in animal studies. Reduction of these off-target effects is desirable in follow-up compounds.

Figure 8

Lead profiling screen results summary.

4.1. Comparison to Existing Art and How the New Probe is an Improvement

At the onset of the uHTS campaign no published prior art small molecules were known to selectively disrupt the interaction of Mcl-1 and Bim. The relevant prior art consisted of three non-selective compounds (Table 3).

Table 3

Comparison of EU-517, obatoclax, gossypol, and ML311.

The aryl sulfone termed “EU-517” and analogs were discovered at Harvard in a screen for inhibitors of the interaction of BH3 with Bcl-xL³³. This compound has modest potency vs. both Bcl-xL and Mcl-1-primed cell lines (11.2, 5.3 μM, respectively). The assay provider (Eutropics) has evaluated the activity of EU-517 against Mcl-1 in in vitro protein binding and cell-based assays. This compound is effective in killing tumor cells in vitro with EC₅₀ values in the micromolar range³³. In unpublished work, the assay provider has found that EU-517 aids survival of mice bearing lymphatic tumors and tumors of the spleen. The specific mechanism of action of this series of compounds, including EU-517, was validated by NMR in the lab of Dr. Gerhard Wagner³⁴, who demonstrated binding to the same hydrophobic groove as the BH3 peptides. Further, NMR chemical shift perturbation studies were used to map the interaction interfaces and to define the conformational space for molecular modeling.

The second relevant prior art compound to consider is obatoclax, which has progressed into phase II trials. It is reported to cause neurological symptoms (euphoria) in CLL patients and neural toxicity in mice^35–37, in vivo toxicities inconsistent with its action as a BH3 mimetic. The conjugated pyrrole structure is also related to ATP-competitive kinase inhibitors, which could explain some off-target activity, including added cytotoxicity. The absence of thrombocytopenia, demonstrated to be an on-target effect of Bcl-xL inhibition, in two independent studies is also telling. We obtained a sample of obatoclax and found that it has ~5-fold lower activity in the Mcl-1 FP assay than does ML311 (1.5 μM for obatoclax, 0.31 μM for ML311).

A third prior art compound to consider is gossypol (AT-101), which like obatoclax exhibits moderate pan-Bcl-2 inhibitory activity, blocking Mcl-1 action along with that of its structural homologs Bcl-2 and Bcl-xL³⁸. Recent characterization of the mechanism of action for gossypol and obatoclax, as well as findings of significant yet poorly understood off-target cytotoxicity, raise concerns with respect to their potential for clinical success. Despite their pan-Bcl-2 activity, these compounds (unlike ML311) kill cells in a Bax-Bak independent fashion and do not induce caspase-9 dependent cell death³⁹. Instead, they likely function through a non-apoptotic pathway as mitochondrial toxins, raising a concern of causing non-mechanism-based toxicities in vivo. Indeed, gossypol is used as a male contraceptive in China.

For all of these compounds the lack of a clear, single, and selective mechanism of action diminishes the feasibility of identifying patient populations likely to benefit from treatment.

There have been a number of commercial efforts to develop drugs that inhibit BH3 binding to anti-apoptotic proteins, without selectivity for Mcl-1. Most notable is a program at Abbott that has yielded a very active Bcl-2/Bcl-xL inhibitor ABT-263 now in clinical trials¹³. This compound however, has been shown to be ineffective in cancer cells that express abundant Mcl-1^40,41. This is a significant problem given the key role that Mcl-1 plays in blood cancer onset and in cellular chemoresistance^42–44. Elevated expression of Mcl-1 and elevated Mcl-1/Bax ratio correlates with chemoresistance in CLL and NHL patients, pointing to the importance of targeting Mcl-1 as a therapeutic strategy. There are serious obstacles to the successful development of ABT-263, including that it induces thrombocytopenia in patients due to inhibition of Bcl-xL, which is essential for platelet survival^45,46. This on-target toxicity requires precisely calibrated dosing regimens to gain a tolerable therapeutic window.

Given the high interest in targeting Bcl-2 family proteins, it is not surprising that additional reports have surfaced since the onset of our probe development efforts. In July 2010 a structure related to ML311 termed “compound 65” was disclosed⁴⁷. No biochemical data or single agent cellular data was reported for compound 65, rather only a positive synergy with the kinesin spindle protein inhibitor ARRY-520 was disclosed. We independently studied this compound (CID 3625042, SID 125001187, Figure 9) in the course of our probe development efforts and found it to be slightly less potent in all target assays than is ML311 (compound 65 has IC₅₀ = 1.1 μM for Mcl-1 in the FP assay, 3.4, 6.0, and 4.2 μM for Mcl-1-1780, DHL-6, and NCI-H929 cells, respectively; compare with values for ML311 of 0.31 μM in the FP assay and 0.3, 3.3, and 1.6 μM in the cell-based assays).

Figure 9

Recently-published Mcl-1 inhibitors. (Reported after our uHTS.)

In 2011, a group from the Dana-Farber Cancer Institute published a patent application claiming uses of twelve scaffolds with activity vs. Mcl-1, as based on single-point screening data. One of the twelve described scaffolds is the 8-hydoxyquinolines⁴⁸. The same group has also published data on non small molecule “stapled peptides” as Mcl-1 inhibitors⁴⁹. These compounds are not freely available for use, and no data is given to determine their usefulness as probes, such as IC₅₀/EC₅₀ values for biochemical and cell-based activity, cell-based selectivity, etc. A representative of the seven compounds in the 8-hydroxyquinoline class is also shown in Figure 9. The cursory and incomplete description of the activity of the compounds, as well as the lack of discussion as to whether any of them may be preferred over the numerous other compounds in the eleven other scaffolds that are exemplified in that patent application, fails to establish the compounds in the Dana-Farber disclosure as prior art that would seriously limit the usefulness of ML311. In addition, this data was published after the initiation of the current MLPCN effort and the identification of the hit scaffold.

Also, in 2011 an academic research team in China reported efforts to target Mcl-1 by modification of a scaffold previously shown to be a dual inhibitor of Bcl-2 and Mcl-1⁵⁰. The most potent compound in the series, compound 6h (Figure 9) showed low nanomolar competitive binding affinity with Mcl-1 in an ELISA assay, ~60-fold selectivity vs. Bcl-2, and impressive cytotoxicity vs. three tumor cell lines by an MTT assay (IC₅₀ = 12–18 nM). It is not clear, however, if the cytotoxicity is Mcl-1-dependent in the absence of more control studies, particularly since related compounds with lesser affinity for Mcl-1 were also cytotoxic to a similar extent. A serious issue is the poor drug-likeness and multiple structure alerts of this large, mostly planar, and highly hydrophobic structure, which is similar to other large heteroaromatic compounds that act as cytotoxins through DNA intercalation.