Truncated Itraconazole Analogues Exhibiting Potent Anti-Hedgehog Activity and Improved Drug-like Properties

Abstract

We conducted a structure–activity relationship study to explore simplified analogues of the itraconazole (ITZ) scaffold for their ability to inhibit the hedgehog (Hh) signaling pathway. These analogues were based on exploring the effects of chemical modifications to the linker and triazolone/side chain region of ITZ. Analogue 11 was identified as the most potent compound in our first generation, with an IC₅₀ value of 81 nM in a murine Hh-dependent basal cell carcinoma. Metabolic identification studies led us to identify truncated piperazine (26) as the major metabolite in human liver microsomes (HLMs) and an improved Hh pathway inhibitor (IC₅₀ = 22 nM). This work verifies that continued truncation of the ITZ scaffold is a practical method to maintain potent anti-Hh activity while also reducing the molecular weight for the ITZ scaffold and achieving improved pharmacokinetic properties.

The Hedgehog (Hh) pathway is a signaling cascade critical for cell development and tissue patterning.¹ Inappropriate modulation of Hh signaling has been implicated in developmental disorders and cancers including basal cell carcinoma (BCC), medulloblastoma (MB), and leukemia.²⁻⁵ The approval of vismodegib and sonidegib as treatments for advanced BCC confirmed the clinical efficacy of small molecule pathway inhibitors for the treatment of Hh-dependent cancers.⁶⁻⁹ Of continued concern in the development of Hh pathway inhibitors is the identification of multiple mutations in Smoothened (Smo), the molecular target of vismodegib and sonidegib, which reduce binding affinity for the drugs and decrease overall efficacy.¹⁰⁻¹²

Recent years have seen the identification and exploration of itraconazole (1, ITZ, Figure 1), an FDA approved antifungal agent, as an Hh pathway inhibitor.¹³⁻¹⁵ Strong evidence exists to suggest that ITZ binds and directly inhibits Smo, a key component of the Hh pathway;¹³ however, ITZ retains potent Hh inhibitory activity in vitro and in vivo in the presence of vismodegib-resistant Smo mutants.^16,17 With this in mind, the ITZ scaffold represents a promising anticancer strategy that has the possibility to address acquired resistance after treatment associated with the clinically approved Smo antagonists.

Figure 1.

Structure and regions of ITZ, lead analogue 2, and analogues made in current work.

As shown in Figure 1, the structure of ITZ can be separated into three general regions, the dioxolane region, the central diphenylpiperazine core, and the triazolone/side chain region. Our research group has been exploring structure–activity relationships (SARs) for the ITZ scaffold with respect to its anti-Hh activity.^14,15 In these previous studies, we demonstrated that the stereochemistry around the dioxolane is critical for potent Hh inhibition. Analogues with the 4R orientation are significantly more active than the comparable 4S analogues.¹⁴ We also demonstrated that the triazole ring is not required for ITZ-mediated Hh inhibition.¹⁴ Our most recent SAR suggests that modification in the triazolone/side chain area is well-tolerated. Potent compounds have been discovered through side chain group modification,^14,18 (hetero)cycle replacement,¹⁵ and even direct truncation of the triazolone ring.¹⁴ In the current study, the simplified triazolone analogue 2, which retained the potent anti-Hh activity and a decreased molecular weight compared to ITZ, was used as our lead structure. We sought to begin probing SAR for the tether region adjacent to the triazolone by directly appending the aromatic functional group to the central piperazine through an amide bond. These modifications would also result in analogues with reduced molecular weight.

Following our previously reported procedures, 1-(4-hydroxyphenyl)piperazine 21 and the stereochemically defined dioxolane regions 22 (2S,4R-cis) and its enantiomer (2R,4R-trans) were synthesized.¹⁴ As shown in Scheme 1, intermediate 21 was coupled with an appropriate carboxylic acid in the presence of the condensing agent 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide (EDCI) and 4-dimethylaminopyridine (DMAP) at room temperature. Selective addition to the piperazine secondary amine was observed for the majority of our analogues. Amide intermediates were then coupled with dioxolane 22 (2S,4R-cis) under alkaline conditions to afford analogues 5–8, 15, and 16. The analogue bearing a para-nitro (8) was reduced to the corresponding aniline 9 by palladium on charcoal. For analogues bearing an allyl group at the terminal phenol, free phenols were protected as allyloxy before coupling with 21. By treating with sodium borohydride and iodine, the allyl group was removed under a translocation mechanism to give phenols 3, 4, 10–14, and 17.¹⁹ Annotated structures for the final analogues are listed in Table 1.

Scheme 1. General Synthesis of Amides and Hydrazides.

Reagents and conditions: (a) appropriate carboxylic acid, EDCI, DMAP, TEA, DMF, RT, overnight, 31–89%; (b) Cs₂CO₃, DMSO, 60 °C, 4 h, 51–93%; (c) Pd/C, hydrazine hydrate, EtOH, reflux, 4 h, 91%; (d) NaBH₄, I₂, THF, RT, 2 h, 31–62%; (e) phenyl chloroformate, K₂CO₃, DCM, 0 °C to RT, 4 h, 62%; (f) hydrazine hydrate, 1,4-diox, reflux, 12 h, 89–91%; (g) appropriate benzaldehyde, EtOH, reflux, 4 h, 62–92%.

Table 1. Anti-Hh Activity for ITZ and Its Analogues^a.

^aIC₅₀ values represent the mean ± SEM of at least two separate experiments performed in triplicate.

For the analogues bearing a hydrazide, intermediate 21 was first coupled with phenyl chloroformate under alkaline conditions. The resulting amide 23 was coupled with dioxolane 22, followed by refluxing with hydrazine hydrate in 1,4-dioxane to afford hydrazide 24. The hydrazide terminal was later coupled with appropriate benzaldehydes to give final phenols 18 and 19. When 23 coupled with dioxolane in 2R,4R-trans configuration, followed by the same hydrazinolysis and condensation procedures, hydrazide 20 was synthesized (structure in Table 1, complete syntheses are shown in the Supporting Information).

As noted earlier, our previous SAR studies for the ITZ scaffold demonstrated that the triazolone/side chain region of ITZ is amenable to modification. To continue our SAR exploration of this region of the scaffold, we sought to synthesize and evaluate analogues that continue the scaffold truncation that we previously probed by removing the phenyl ring on the “right side” of the scaffold and directly appending the 3-phenolic carboxylic acid of 2 to the piperazine amine to provide analogue 3. We evaluated this analogue for its ability to decrease Gli1 mRNA expression in the Hh-dependent murine BCC cell line ASZ001.²⁰⁻²² Analogue 3 demonstrated comparable anti-Hh activity compared to 2 (IC₅₀ values = 0.26 μM and 0.16 μM, respectively), verifying that this series of compounds retains potent Hh inhibition and encouraging us to synthesize additional analogues that incorporate the 4-phenylpiperazinyl amide.

To further probe SAR for functionality on the aromatic moiety, aromatic carboxylic acids with a range of substitutions 4–9 were synthesized and evaluated. Analogue 4 (IC₅₀ = 0.58 μM), with the phenol in the para-position, was less active than 3. Replacing the phenol with pyridines resulted in compounds 5–7 with reduced anti-Hh activity, with the 4-pyridine (7, IC₅₀ = 0.57 μM) representing the best in this small collection. Replacing the phenol with a nitro group (8, IC₅₀ = 0.16 μM) demonstrated improved activity compared to 3 and equivalent activity to the lead 2. Reduction of the nitro to the 4-aniline (9, IC₅₀ = 0.28 μM) resulted in a modest decrease in activity.

Next, we sought to increase the distance between the amide bond and the aromatic moiety to understand the optimal distance between these functional groups for potent Hh inhibition (10–20). The addition of a single methylene unit (10) resulted in a slight decrease in Hh inhibition compared to 2; however, the increase to two methylenes (11, IC₅₀ = 0.081 μM) significantly enhanced the anti-Hh activity of the scaffold. Constraining the linker as a trans-alkene 12 resulted in an approximately three-fold loss in Hh inhibitory activity, a trend that was also present when the phenol was located in the 4-position (13 and 14). Deletion of the phenol (15, IC₅₀ = 0.23 μM) or masking it as the methoxy (16, IC₅₀ = 0.46 μM) decreased activity, suggesting the free phenol is not essential for anti-Hh activity, but it is optimal in our scaffold. Because of the commercial availability of 3-hydroxybenzenebutanoic acid, analogue 17 with a propyl linker, but a para-phenol tail was synthesized and evaluated. A decrease in activity was observed with the extended linker in 17 (IC₅₀ = 0.47 μM).

Finally, we sought to vary the chemical composition of the linker by introducing heteroatoms; therefore, a small collection of hydrazide-containing analogues was synthesized and evaluated. Analogue 18 (IC₅₀ = 0.12 μM) demonstrated superior Hh inhibitory activity to 2 and ITZ, once again suggesting that this region of the scaffold is amenable to modification. Not surprisingly, analogue 19 (IC₅₀ = 0.96 μM), which incorporates the para-phenol, was less active compared to the analogous meta-phenol 18. Our previous truncated ITZ analogue series identified the cis-dioxolane as the ideal orientation for these compounds; however, we also synthesized and evaluated the 2R,4R-trans-dioxolane hydrazide to determine whether this trend is replicated for the analogues described herein. This analogue, 20 (IC₅₀ = 0.58 μM), was approximately five-fold less active, highlighting the importance of the cis-orientation around the dioxolane ring for this series of compounds.

In addition to determining whether significant truncation to the triazolone/side chain region results in compounds that retain potent Hh inhibition, a key reason for the preparation of these analogues was to evaluate whether truncation also provides compounds with improved solubility or stability. Several representative compounds (3, 10, 11, and 15) were chosen to evaluate these two parameters for this series of analogues. Unfortunately, all tested analogues failed to improve the overall solubility as each analogue was less soluble than ITZ at both pH 4.0 and pH 7.4 (Table 2). Overall, our analogues were less stable than ITZ in human liver microsomes (HLMs) and demonstrated predicted intrinsic clearance levels significantly higher than the parent compound. On the basis of the reduced stability previously determined for amide-containing analogue 2, it was not too surprising that these compounds were also less stable.

Table 2. In Vitro Pharmacokinetic Data^a.

	solubility, PBS (μM)		microsomal stability
cmpd	pH 4.0	pH 7.4	T1/2 (min)	Clint (μL/min/mg)
ITZb		0.80	27.0 ± 6.2	25.2 ± 3.1
2b		0.02	19.9 ± 0.6	34.7 ± 1.0
3	0.31	0.17	15.9 ± 0.2	43 ± 0.6
10	0.25	0.13	20.4 ± 0.4	33.9 ± 0.6
11	0.13	0.03	3.2 ± 0.1	218 ± 9.7
19	0.05	0.02	11.4 ± 0.2	60.0 ± 1.1

^aMore information about the assay protocols is shown in the Supporting Information.

^bValues for ITZ and 2 are taken from ref (15).

Disappointingly, our most active analogue, 11, was also the least stable (T_1/2 = 3.2 min). The only structural difference between 11 and 10, which was approximately 6.5-fold more stable, was the addition of a single methylene between the amide and the phenol-substituted benzene ring. To more fully determine how the additional methylene influences stability, we performed follow-up metabolic identification (MetID) studies in HLMs for both 10 and 11.

Incubation of 11 with HLMs resulted in the identification of two primary metabolites (Figure 2 and Figures S1–S4). Peak #1 indicated the metabolite (26, 30.09%) had undergone hydrolysis of the amide bond, an unexpected metabolite considering amides are generally deemed as a reliable tether and exist in many marketed drugs. The second major metabolite (Peak #2, 15.94%) resulted from single oxidation to the 3-hydroxyphenethyl region. While the exact location of the hydroxyl/phenol in metabolite #2 could not be conclusively determined, CYP-mediated oxidation at the para-phenyl position is well-documented.^23,24 On the basis of the measured m/z+ peaks, we presumed that 3,4-bis-phenol 28 is the most likely structure for metabolite #2.

Figure 2.

Metabolite identification for 11 in the presence of HLMs. Percentages indicate the relative peak area under experimental conditions.

For analogue 10, our MetID studies suggested this analogue was involved in more complicated metabolic pathways including hydrolysis, hydrogenation, deoxidation, and then oxidation (Figures S5–S10). Interestingly, no direct oxidation to the phenol region was observed in the metabolites of 10. The difference in metabolic pathways for 10 and 11 may explain the difference in HLM stability. Of note, 26 was the primary metabolite for both 10 and 11, indicating that the N-piperazine amide bond was a clear metabolic soft spot on our amide ITZ analogues.

On the basis of the results of the metabolism studies, our follow-up experiments were focused in two directions. First, we synthesized and evaluated the metabolites (26 and 28) of our most active analogue 11 to determine if they retained potent anti-Hh activity. In parallel, we sought to enhance the stability of 11 by eliminating the common metabolic soft spot, which led to the design of saturated analogue 31. The syntheses are described in Scheme 2. For the analogue truncated at the piperazine, selective protection of 21 as the secondary Boc amine afforded phenol 25. After coupling with dioxolane 22, Boc removal was performed by stirring with TMSOTf under alkaline conditions to give the free piperazine 26. Starting with 3,4-bis-allyloxybenzenepropanoic acid and using the same procedures as in Scheme 1, bis-phenol 28 was synthesized. For the des-amide analogue, intermediate 29 was synthesized after reduction and tosylation. Selective addition to the piperazine amine of 21 was observed by heating in the presence of 29 to give phenol 30. Dioxolane coupling and final deprotection were performed under the same conditions as described earlier to give amine 31.

Scheme 2. Synthesis of Metabolites 26 and 28 and Amine 31.

Reagents and conditions: (a) Di-tert-butyl dicarbonate, TEA, DCM, RT, 2 h, 93%; (b) 22, Cs₂CO₃, DMSO, 60 °C, 4 h, 39–98%; (c) TMSOTf, 2,6-lutidine, DCM, 0 °C to RT, overnight, 72%; (d) 3,4-bis-allyloxybenzenepropanoic acid, EDCI, DMAP, TEA, DMF, RT, overnight, 81%; (e) NaBH₄, I₂, THF, RT, 2 h, 41–71%; (f) 29, K₂CO₃, DMF, 40 °C, 4 h, 41%.

We were excited to determine that both metabolites retain potent inhibition of Hh signaling (Table 3). Analogue 28 (IC₅₀ = 0.089 μM), which incorporated the bis-phenol, showed equivalent activity to the parent 11, whereas the truncated piperazine 26 (IC₅₀ = 0.022 μM) was more potent than 11. Amine 31 was significantly less active than 11 or the other two metabolites (IC₅₀ = 0.87 μM). In C3H10T1/2 cells, Hh-dependent mouse embryonic fibroblasts (MEFs), 11 and 26 were also more potent than ITZ when using endogenous oxysterols as the Hh pathway activator (Table S1). On the basis of the improved activity of the truncated analogue 26, which contains approximately half of the intact ITZ scaffold, we also evaluated this analogue in Sufu^–/–  MEFs to determine whether Smo is still the most likely target of the metabolite. The anti-Hh activity of 26 was completely abolished in this cell line, suggesting 26 (and ITZ) is functioning upstream of Sufu, most likely at the level of Smo (Figure S11).

Table 3. Anti-Hh Activity^a and Pharmacokinetic Data^b for Metabolite-Based ITZ Analogues.

^aIC₅₀ values represent the mean ± SEM of at least two separate experiments performed in triplicate.

^bMore information about the assay protocols is shown in the Supporting Information.

The activity for both metabolites prompted us to study their PK profile (Table 3). Not surprisingly, analogue 26 demonstrated improved stability (T_1/2 = 50.7 min) compared to parent compound 11. The stabilities of compound 28 and 31 (T_1/2 = 6.6 and 14.0 min, respectively) were minimally improved compared to 11. In addition, significant improvements in solubility at pH 7.4 and pH 4.0 compared to ITZ were found for analogue 26. Finally, we evaluated the general toxicity of compounds 11 and 26 by exploring their antiproliferative activity in HC-04 cells, a human hepatocyte-derived cell line that abundantly expresses drug metabolizing enzymes (Table S2).²⁵ Both compounds demonstrated moderate toxicity (26, GI₅₀ = 27.3 μM; 11, GI₅₀ = 13.18 μM) in the HC-04 cells, indicating their anti-Hh activity is not a result of broad toxicity and further highlighting their potential as selective anticancer therapeutics.

In conclusion, we have developed a series of ITZ analogues focused on continued truncation of the “right-side” of the scaffold. Analogue 11 represents the most potent analogue in the first-round SAR study, with an IC₅₀ value of 81 nM, approximately two-fold more potent than lead 2 and ITZ. Our initial SAR suggested that a proper length linker with certain flexibility is critical for potent anti-Hh activity. In addition, a stereochemically defined dioxolane cap (2S,4R-cis) and a meta-phenol tail are also important for the anti-Hh activity of 11. A metabolite of compound 11, analogue 26, demonstrated improved Hh pathway inhibition, increased stability, and enhanced water solubility. Moreover, the molecular weight of 26 (MW = 423.3) is significantly decreased compared to ITZ (MW = 705.6), establishing analogue 26 as a promising candidate for further in vitro and in vivo analysis. Overall, these studies provide clear rationale for continued modification to the central core of the scaffold as a practical design strategy to retain potent anti-Hh activity while improving the overall drug-like properties of the scaffold.

Glossary

Abbreviations

ITZ

itraconazole

Hh

hedgehog

BCC

basal cell carcinoma

MB

medulloblastoma

Gli

glioblastoma associated oncogene

Smo

Smoothened

SAR

structure–activity relationship

EDCI

1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide

DMAP

4-dimethylaminopyridine

Boc

tert-butyloxycarboryl

TMSOTf

trimethylsilyl trifluoromethanesulfonate

HLMs

human liver microsomes

MEFs

mouse embryonic fibroblasts

MetID

metabolic identification

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.9b00188.

• Detailed MetID results, general experimental methods, detailed synthetic procedures, compound characterization (PDF)

Author Contributions

J.W. synthesized and evaluated ITZ analogues and wrote the manuscript. D.C. synthesized ITZ analogues. S.R.M. evaluated ITZ analogues. All authors have given approval to the final version of the manuscript.

We gratefully acknowledge support of this work by the National Institutes of Health/National Cancer Institute (CA190617).

The authors declare no competing financial interest.