Codrug Approach for the Potential Treatment of EML4-ALK Positive Lung Cancer

ABSTRACT: We report on the synergistic effect of PI3K inhibition with ALK inhibition for the possible treatment of EML4-ALK positive lung cancer. We have brought together ceritinib (ALK inhibitor) and pictilisib (PI3K inhibitor) into a single bivalent molecule (a codrug) with the aim of designing a molecule for slow release drug delivery that targets EML4- ALK positive lung cancer. We have joined the two drugs through a new, pH-sensitive linker where the resulting codrugs are hydrolytically stable at lower pH (pH 6.4) but rapidly cleaved at higher pH (pH 7.4). Compound (19), which was designed for optimal lung retention, demonstrated clean liberation of the drug payloads in vitro and represents a novel approach to targeted lung delivery.

Of the nonsmall cell lung cancers (NSCLCs), 3−7% possess chromosomal rearrangements of anaplastic lymphoma kinase (ALK).1,2 ALK gene translocations such as EML4-ALK, resulting from inversions in the p-arm of chromosome 2, lead to the fusion of the echinoderm microtubule-associated protein-like 4 (EML4) and ALKgenes and produce constitutively active ALK fusion proteins.3,4 The ALK-EML4 fusion gene5 contains exons of ALK that code for the intracellular kinase domain of ALK and eight EML4 exons, all of which activate downstream signals of the Ras/Raf/ MEK/ERK1/2, JAK/STAT, and PI3K/AKT/mTOR path- ways,6 ultimately leading to uncontrolled cell growth and thus its oncogenic nature.As ALK is required for oncogenic activity, ALK inhibitors, such as crizotinib,7,8 have been developed as targetedantitumor therapies. Ceritinib 1,9,10 developed by Novartis,operate with different mechanisms of action, thus reducing the probability of mutations. As the overall goal is to identify a bivalent inhibitor where the two drugs are conjugated with a pH-dependent-cleavable linker (Figure 1), the aim of this work is to identify a drug combination that exhibits synergy in a 1:1 molar ratio. Therefore, the drug combinations were tested in a 1:1 molar ratio.

The combination of ceritinib (1) and the pan PI3K inhibitor pictilisib (2)19 was investigated. We initially examined in vitro activity adopting 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte- trazolium bromide (MTT) assays against an isogenic pair of human lung carcinoma cell lines, A549 and A549 EML4-ALK, referred to as ALK− and ALK+ respectively. The A549 cell lines used and their respective known mutations are listed in Table 1.In vitro evaluation of pictilisib revealed GI values of 1217has been approved by the FDA (April 2014) for patients thathave relapsed on crizotinib.11 Similarly to crizotinib, ceritinib initially shows promising results. However, patients relapse as resistance rapidly emerges. The situation is similar to alectinib12 and lorlatinib,13 where patients initially respond; however, relapse is inevitable. Current treatments involve taking these ALK inhibitors in succession as resistance evolves.14,15Another strategy for overcoming resistance to ALK inhibitors is to use upfront combination therapies,16 targeting one or more signaling nodes that suppress the survival and emergence of resistance.17,18 However, it is unclear which effector is most critical for EML4-ALK driven cell survival. Drug combinations have been shown to slow down the evolution of resistance, as drugs in combination simultaneouslylines, respectively. Ceritinib showed greater inhibitory activity against ALK+ cells than ALK− cells (GI50 values 565 ± 102 and 845 ± 87 nM, respectively; Figure 2).Evaluating synergy using the Webb method20 (Figure 3) showed no statistical significance between the observed and expected effects across all drug concentrations in ALK− cells.

Conversely, against the ALK+ cell line, there was good statistical significance of synergy at concentrations >500 nM.range of drug combinations (Table 2). The CIs at a fractional effect (Fa) of 0.5 (effectively at the IC50) were found to be0.72 and 0.37 for the ALK− and ALK+ cell lines, respectively. At higher concentration, at a fractional effect of 0.35, the CIs were 0.24 against both the ALK− and ALK+ cell lines. The CI plots for both cell lines are shown in Figure 4. The synergy observed against the ALK+ cell line is greater and observedacross a broad range of drug concentrations. Isobolograms at IC50, IC75, and IC90 clearly illustrate synergistic growth inhibition, with the strongest synergy at doses achieving higher growth inhibitory effects (Figure 4A,B).23,24 Clearly, synergy between pictilisib and ceritinib is observed against the ALK+ cell line.After observing substantial synergy specifically in ALK+ cells, Western blot analyses were adopted to explore signal transduction activation following treatment of cells with pictilisib (500 nM) and ceritinib (500 nM). Figure 5 demonstrates the presence of the EML4-ALK fusion protein in lysates prepared from ALK+ A549 cells. The presence of STAT3 protein is evident in both A549 populations irrespective of EML4-ALK expression (Figure 6A). In contrast, downstream of ALK, activation of the JAK-STAT pathway (P- STAT3) is detected exclusively in ALK+ cells. Also evident is inhibition of STAT3 signaling following treatment of cells with ceritinib alone and almost complete abolition of P-STAT3after treatment of cells with the drug combination (Figure 6B).A further issue associated with ceritinib treatment is adverse toxicity, where >50% patients in clinical trials experienced adverse events that necessitated a reduction in dose. There are also serious side effects observed with pictilisib (diarrhea, nausea, taste alteration, rash, fatigue, itchiness, vomiting, and decreased appetite), where a suggested daily oral dose is 340 mg.

With these adverse effects, either dose reduction or localized, topical application is desirable. One strategy to achieve such outcomes would be delivery via controlledThe Chou and Talalay method for drug combination analyses, based on the median-effect equation,21,22 is discussed in detail in the SI. The resulting combination index (CI) theory offers quantitative definition of additive (CI = 1), antagonistic (CI > 1), or synergistic (CI < 1) effects across arelease, i.e., in the form of a prodrug. Inhalation is an increasingly important delivery approach for respiratory therapeutics, and yet, there still remains an unmet clinical, commercial, and practical need for a “once-a-day” treatment. There has been much debate about the various strategies for the rationalization of agents that exhibit a sustained duration of action when applied topically to the lung.26 Examples of such approaches are compounds that display a reduction in solubility and permeability, where slow dissolution into the airway’s smooth muscle affords the potential for extended lung retention.27 In addition, increasing lipophilicity has been shown to be an important parameter in delivering compounds with an extended duration of action.28,29 Along with a modulation in lipophilicity, it has been shown that the incorporation of a dibasic pharmacophore within the compounds leads to an increase in the duration of action of inhaled compounds.30With these concepts in mind, we have brought together ceritinib and pictilisib into a single bivalent molecule (a codrug)31,32 with the aim of designing a molecule for slow release drug delivery that could target EML4-ALK positive lung cancer.

The codrug is designed as a bivalent ligand, sporting a pH-dependent-cleavable linker, which is more hydrolytically stable at lower pH (pH 6.5 vs 7.4). This approach is interesting in that the pH of the lung environment is slightly acidic (pH ≈ 6.5 to 6.8), enabling a potential lung tissue retention for the biologically inactive codrug.The piperidine ring on ceritinib and the indazole of pictilisib were identified as appropriate synthetic handles for explora- tion. In a codrug model system (5), 4-phenyl piperidine (3) and 5-bromo indazole (4) were used as drug mimetics. With the indazole as a superior leaving group as compared to the piperidine, this was used to form an α-amino amide. The α- amino amide was chosen with the hypothesis that the amino group will aid hydrolysis through anchimeric assistance via hydrogen bonding of the nonprotected amine to water (pKa ≈ 7.08).33 This hypothesis forms the basis of the pH-dependent linker (Figure 7).34 An aspartic acid analogue was chosen toform part of the linker due to the well-known cyclizationmechanism of amino acid cleavage, which would present itself following hydrolysis of the α-amino amide, releasing pictilisib. The concept was investigated through the synthesis of a codrug model 5. A kinetic study was performed using a time- course NMR method where the time-course experiment was run at pH 6.4, 6.8, and 7.4, and the corresponding half-livesand rate constants calculated (Table 3 and SI).Immediately, it could be seen that as the pH increases, the model codrug cleaves more rapidly (Table 3). The fact that the rate of reaction increases with the percentage un-ionized is evidence in favor of the hypothesis that the α-amino group delivers water to the amide carbonyl through anchimeric assistance.

After proof of concept with the model system, cleavable codrugs (19−21) were synthesized starting from ceritinib and pictilisib (Scheme 1).The synthesis of codrugs (19−21) started with the acylation of ceritinib (1) with 1-chloroethyl chloroformate (ACE-Cl) in the presence of N-methyl morpholine to give (6). Coupling of the amino acid partners (7−9) was achieved using silver oxide containing a catalytic amount of tert-butyl ammonium bromide in toluene at 60 °C. Following deprotection, the corresponding carboxylic acids (13−15) were reacted with 2 to give the Boc-assisted, depending on the pH, as the amino group will have a different percentage ionization at differing pH. Following the delivery of water and release of pictilisib, the codrug has been designed to degrade via a well-known cyclization mechanism, yielding the anhydride (which is subsequently hydrolyzed to the amino acid). This is followed by collapse, eliminating acetaldehyde (which disproportionates to acetic acid and ethanol—as seen by NMR), carbon dioxide, and ultimately ceritinib. It is known that release of acetaldehyde could have inflammatory effects in the lung tissue; however, this could be alleviated in future work through modification of the acetal linker group.protected codrugs (16−19). Once required, the protecting group was removed with TFA to give the codrugs (19−21). The protected codrugs were stable in air at room temperature>1 year after synthesisThe hydrolytic stability of (19−21) was investigated in buffered solutions supplemented with 4% bovine serum albumin (BSA) to aid solubility (Table 4).

In addition, rat plasma stability was assessed (Table 5). The stability results for(19) at both pH 6.5 and 7.4 (Figure 8) show clean conversion to pictilisib and ceritinib.The compounds were assayed as a mixture of two diastereomers. However, 19 was purified by HPLC, and it was observed that there was no difference in the rates of release of the single enantiomers.From Table 4, it can be seen that the half-lives of the codrugs at pH 6.5 are longer than the half-lives at pH 7.4.In rat plasma, the degradation of 20 is different from that of 19, which showed a zero-order degradation (a linear plot of [codrug] vs time). Compound 20 shows a first-order degradation profile.The biological evaluation of 19 was studied by MTT assay and compared to the ceritinib/pictilisib combination (Figure 9).The concentration−response curve in the ALK+ cell line appears to mirror the concentration−response of the combination but is right shifted. This suggests that the codrughad not fully cleaved to provide the drugs in the desired concentration to have the required biological effect. The most probable reason for this effect is that the codrug is highly protein bound. As accumulating evidence suggested that FBS negatively impacted biological activity, 19 was incubated in RPMI medium containing 10% FBS, and the drugs’ release wasfollowed by HPLC. Even after 72 h, the codrug had not released its active payloads, and this is highly likely a consequence of very high protein binding (Figure 10).

To conclude, we have demonstrated the synergistic biological activity between pictilisib and ceritinib and have used this finding to design, synthesize, and test a novel codrug system where the two drugs are joined by a new pH-cleavable linker. While the resulting codrug did not display the anticipated enhanced biological response, we have suggested that this is likely because of the inherent high albumin binding exhibited by the codrug due to its high lipophilicity (cLogP 6.9).33 However, we postulate that this could be a possible benefit for ALK inhibitor inhaled sustained delivery applications. Further work is ongoing to explore the albumin binding dependence and cleavage of the codrugs and will be reported in due course.