The Effect of N-Alkyl Azole Difluorination on Molecular Properties Relevant for Compound Optimization: A Comparative Study

Abstract

Despite recent interest in N-trifluoromethyl azoles, N-α,α-difluoroalkyl azoles [(azole)N–CF₂R] remain understudied and underutilized in medicinal chemistry. To address this deficiency, we have conducted a comparative study of medicinally-relevant properties for a series of (azole)N–CF₂R and their nonfluorinated matched molecular pairs (MMPs) that revealed fluorine-induced reductions in azole pK_a, hydrophilicity, experimental polar surface area, and metabolic oxidation of a labile vicinal position. Additionally, computational analysis supports the fluorine-induced suppression of metabolic aliphatic oxidation but suggests a limited impact of fluorination on conformational preferences within MMPs. Along with a newly-provided synthetic method to install such a substructure, this information will facilitate rational incorporation of (azole)N–CF₂R groups in drug optimization campaigns.

Graphical Abstract

Authors are required to submit a graphic entry for the Table of Contents (TOC) that, in conjunction with the manuscript title, should give the reader a representative idea of one of the following: A key structure, reaction, equation, concept, or theorem, etc., that is discussed in the manuscript. Consult the journal’s Instructions for Authors for TOC graphic specifications.

INTRODUCTION

Fluorine atoms and fluorinated motifs are frequently introduced to medicinal agents to modulate their pharmacokinetic (PK) and pharmacodynamic (PD) profiles. These effects are often attributed to the influence of the highly electronegative fluorine atom on the electronic and conformational characteristics of proximal heteroatom-containing functional groups, which can significantly perturb a molecule’s physicochemical properties.^1,2 For example, N-trifluoromethyl azoles [(azole)N–CF₃] possess decreased pK_a, increased lipophilicity, and greater stability towards metabolic N-dealkylation relative to their non-fluorinated counterparts (Figure 1A).^3,4 Specifically, the three electronegative fluorine atoms withdraw electron density from the azole nitrogen atoms, causing the observed modulation of pK_a and logD.^1,5,6 Additionally, the energetic barrier to metabolic oxidation at the N-α carbon atom is increased for the N-trifluoromethyl relative to N-methyl azoles, which likely explains the improved stability to metabolic N-dealkylation.⁷

Figure 1.

Extending fluorinated azole property profiling to (azole)N–CF₂R.

A number of medicinally relevant azoles contain structurally diverse and activity-driving N-substituents, which include systemic (i.e., fluconazole, itraconazole, and voriconazole) and topical (i.e., econazole, ketoconazole, and miconazole) azole antifungals (Figure 1A).⁸ Analogous to N-methyl azole [(azole)N-CH₃], α,α-difluorination of N-alkyl azoles [1, (azole)N–CH₂R] could improve drug-like properties while retaining the structural diversity of the N-alkyl group [1→2, Figure 1B]. However, currently no (azole)N–CF₂R-containing clinically approved drugs (DrugBank) exist and less than 3,000 (azole)N–CF₂R compounds tested in biological studies (SciFinder). Given the prevalence of N-alkyl azoles in biologically active molecules and the potential advantages of N-α,α-difluoro substitution, evaluation of the benefits and deficiencies of (azole)N–CF₂R may promote integration of this motif during compound optimization.

Herein, we disclose a comprehensive comparison of a series of (azole)N–CH₂R and (azole)N–CF₂R matched molecular pairs (MMPs, Table 1) that assesses differences in molecular properties relevant to absorption and distribution (pK_a, logD, kinetic solubility) and metabolism (in vitro liver microsomal stability and metabolite profiling). The difluoro substitution’s impact on the molecule’s conformation was also explored, to further enable rational utilization in drug optimization campaigns.

Table 1.

In General, α,α-Difluorination Increases Measured LogD, Lowers EPSA, and Decreases Microsomal Stability of N-Alkyl Azoles.

^aLM CL_int = In vitro intrinsic clearance from human (H), rat (R), and mouse (M) liver microsomes [mL/min/kg].

^bTurnover observed in the absence of NADPH in human liver microsomes (CL_int HLM-N 4b = 9.27, 9b = 13, 10a = 7.73, 10b = 58.9).

RESULTS

Fragment-sized match pairs were designed to explore diazoles (imidazoles 3 and 8; pyrazoles 4, 5, 9 and 10) and benzannulated derivatives (1,2,3-benzotriazoles 6, indazoles 7) across a logD range of 0.2–3.3. For all MMPs, logD was consistently higher for the difluoro analogs. Interestingly, while the measured change in logD was small for some analogs (0.2–0.3; 10a vs. 10b, 5a vs. 5b), in other examples, the logD difference was more significant (0.5–1.4; 9a vs. 9b, 3a vs. 3b, 4a vs. 4b, 7a vs. 7b, 8a vs. 8b, 6a vs. 6b). The overall increase in logD for the difluoro substituted analogs is consistent with the change in experimental polar surface area (EPSA) observed across analogs. On average, difluorination decreases EPSA by 12 A² and can be rationalized through a combination of polarity shielding and electronic effects on the azole ring’s heteroatoms. All MMPs included in this analysis had moderate to high solubility and permeability (Table S1), which suggests that fluorination across these MMPs did not negatively impact the solubility or permeability of these compounds.

Metabolic stability was evaluated in human, rat and mouse liver microsomes in the presence of nicotinamide adenine dinucleotide phosphate (NADPH). Accompanying the increase in logD for the difluorination, (azole)N–CF₂R tended to display decreased stability relative to their analogous (azole)N–CH₂R across all three species (compare LM CL_int). Interestingly, turnover was observed even in the absence of NADPH in human liver microsomes (HLM-N) for four analogs (4b, 9b, 10a, 10b). For these analogues, the cause of the observed turnover in HLM-N is presently unknown. However, chemical instability is unlikely a contributing factor, as (1) 4a, 4b, 9a, 9b, 10a, and 10b demonstrated minimal decomposition in human plasma (via recovery from a plasma protein binding assay, Table S1), though 4a and 4b showed presumed amide hydrolysis in rodent plasma, and (2) 4b and 10b did not degrade in pH 7.4 phosphate buffered saline over 70–77 hours (via ¹⁹F NMR, see SI, S6). Regardless, instability in HLM-N was not observed for any of the other tested compounds, thus suggesting that the N–CF₂R motif itself is inherently stable.

To further understand how difluorination impacts both the rate and the site(s) of metabolism, we evaluated two of the matched pairs in metabolite identification (metID) studies in human microsomes (Scheme 1A). While limited metabolism was observed in all cases and the parent compound remained the major species detected, two notable trends emerged from this metID study: fluorination both shifted the site of metabolism of the analogs and increased the rate of metabolism, consistent with observations from the in vitro liver microsomal clearance assay. For benzotriazole 6a, oxidation was observed at both benzylic positions, with the major metabolite arising from N-dealkylation. In addition to blocking oxidation at the N-α carbon atom, difluorination largely ablated metabolism on the remaining unsubstituted benzyl position (<1%) but increased the degree of N-demethylation of the distal pyrazole. In the second matched pair (7a,b), a similar pattern of oxidation and N-dealkylation was observed along with a second oxidation site on the indazole. Again, the difluoro substitution minimized oxidation of the benzylic positions and shifted the oxidation to the indazole ring. Similarly, an increase in distal N-dealkylation was observed for the fluorinated member of the matched pair. Notably, as there are numerous avenues for mitigating N-dealkylation, in a drug discovery campaign, this fluorine-induced change may be combined with other strategies to improve the overall stability of the compounds.

Scheme 1. α,α-Difluorination Blocks Vicinal Benzylic C–H Oxidation and Shifts Metabolic Processes to Alternative Sites.

(A) Fluorinated analogs (6b, 7b) displayed minimal liver microsome-mediated benzylic (β-fluoro) oxidation relative to their nonfluorinated analogs (6a, 7a). %Metabolite in the pie charts is not quantitative but reflects the ratio of extracted ion count (XIC) for the metabolite relative to the XIC of the parent compound at time = 0. Only metabolites representing >1% are noted. No degradation was observed in the absence of NADPH. (B) Fluorinated analogs have stronger calculated (see SI, S16) benzylic (β-fluoro) C–H bonds than their nonfluorinated analogs, which might contribute to the observed decrease in benzylic oxidation for 6b and 7b.

The difluoro substitution potentially drove the observed metabolic shift of CYP-mediated oxidation away from the benzylic position through destabilization of a benzylic radical intermediate,⁹ as evidenced by higher calculated¹⁰ benzylic C–H bond dissociation enthalpies (BDE) for the fluorinated analogs (2.1–2.6 kcal/mol, Scheme 1B). Ultimately, while difluorination did increase the overall degree of metabolism of the analogs, it mitigated oxidation of benzylic carbon atoms, which are notoriously metabolically susceptible positions on a molecule. In an additional perturbation of medicinally relevant properties, the introduction of fluorine atoms decreased azole Brønsted basicity relative to the nonfluorinated analogs with the magnitude change correlating with the number of σ-bonds between fluorine atoms and the basic nitrogen atom (Table 2). Specifically, a ¹H NMR-monitored titration¹¹ was employed to determine both the site of protonation and a pK_a value for the basic residues. For N-alkyl imidazoles 3 and 8, fluorinated analogs were 2.7–3.8 pK_a units less basic than their nonfluorinated counterparts. This magnitude of change is similar to previously reported values for (imidazole)N–CF₃ (ΔpK_a = 2.9 to ≥3.3).³ Moreover, with pK_a values of 3.5–3.8, these (imidazole)N–CF₂R would remain unionized in systemic circulation. Similarly, (pyrazole)N–CF₂R 5b was >1.3 pK_a units less basic than its nonfluorinated analog. However, a precise magnitude of change could not be obtained, as the basicity of 5b was too low for detection by this titration. Difluorination also reduced basicity of the distal pyrazole, though to a lesser extent (3, 6, and 7; ΔpK_a = 0.2–0.9). This magnitude of pK_a change is consistent with an empirical method for aliphatic amine pK_a prediction (predicted ΔpK_a = 0.8),⁵ whereas the method underpredicted the impact of difluorination for the aforementioned geminal imidazoles (3 and 8; predicted ΔpK_a = 2.0). Similar to the observed reduction in Brønsted basicity, α,α-difluorination should reduce the Lewis basicity of azoles, which might minimize off-target interactions with metalloenzymes.^8,12,13

Table 2.

α,α-Difluorination Reduces Basicity of Azoles and Remote Heterocycles.^a

^apK_a values and sites of protonation were determined via a ¹H NMR-monitored titration assay (see SI, S6–15).¹¹

Density functional theory (DFT) calculations¹⁰ suggest that fluorination minimally impacts the conformational preferences of (azole)N–CX₂R (Table 3), which contrasts the effect of fluorination on many functional groups for which fluoro/proteo MMP comparison exists.^1,14 Specifically, for N-CF₂R-imidazole, -pyrazoles, -benzotriazoles and -indazoles, computations suggest a negligible energetic penalty (0.0–0.2 kcal/mol), i.e. strain energy, to force the fluorinated compounds into the minimum energy conformations of their nonfluorinated MMPs (e.g., 3a → 3b, 6a → 6b, 7a → 7b, 9a → 9b) and vice-versa. Moreover, the absolute conformational minima of each MMP contain a similar gauche orientation of (hetero)aryl substituents about the central CH₂–CX₂ bond (see SI, S29–30). Combined, the small computed energetic penalties and similar geometries of the minima imply that the fluorinated derivatives should be able to bind a target in a similar conformation to their non-fluorinated counterparts. This trend was corroborated at alternate conformational minima (see SI, S27–30), which further supports the viability of direct H→F replacement.

Table 3.

α,α-Difluorination Has a Minimal Impact on Conformational Preferences.^a

^aSee the supporting information document (S27–28) for strain energy calculation details.

Importantly, for (azole)N–CF₂R 3b and 6b (see SI, S52), the torsional profiles and estimated rotational energy barriers of the bonds on either side of the fluorination are similar to their nonfluorinated MMPs and accessible at physiological temperatures. These results suggest that fluorination should not prevent sampling of the conformational landscape of these bonds.

SYNTHESIS

N-α,α-difluoroalkyl azoles were synthesized in under 4 steps via routes that feature a photocatalytic hydroazolation of gem-difluoroalkenes (Scheme 2).¹⁵ Specifically, commercially available aldehydes were converted to gem-difluorostyrenes,^16–18 which were coupled with the corresponding azoles in moderate to high yields using an acridinium photocatalyst and diselenide co-catalyst^15,19 to deliver (azole)N–CF₂R (3b, 4b, 6b–8b, 10b; Scheme 2A). However, consistent with prior work,¹⁵ gem-difluorostyrenes bearing strongly electron-withdrawing groups failed to react, so access to aryl methanesulfonate 5b and aryl nitrile 9b required modifications to this route. Specifically, hydroazolation reactions of gem-difluorostyrenes bearing methyl sulfide (11) and bromide (12) groups generated arylthioether and aryl bromide moieties that underwent further oxidation and cyanation reactions to afford 5b and 9b, respectively (Scheme 2B).

Scheme 2. Synthesis of Difluoroalkyl Azoles.

(i) azole, 5% (PhSe)₂, 3% 9-Mesityl-3,6-di-t-butyl-10-phenylacridinium tetrafluoroborate, 427 nm 40 W LEDs, PhMe, 30 °C; (ii) m-CPBA, DCM, rt; (iii) CuCN, DMF, 150 °C.

In contrast to the fluorinated analogues that were all accessed by a single route, access to (azole)N–CH₂R required distinct synthetic sequences. The analogous photocatalytic hydroazolation of nonfluorinated styrenes only delivered 4a, 6a, and 10a (Scheme 3A) – the reactions to generate the other target compounds either did not proceed or only generated trace yields of desired (azole)N–CH₂R product. Instead, to access (azole)N–CH₂R 3a and 7a, a Wittig olefination of 1-methyl-1H-pyrazole-4-carbaldehyde delivered bromostyrene intermediate 13, which was coupled with imidazole using a catalytic copper-based system²⁰ to generate vinyl imidazole 14 (Scheme 3B). Also from intermediate 13, a previously unreported photocatalytic C–Br functionalization of bromostyrene 13 using the aforementioned hydroazolation conditions afforded vinyl indazole 15. Finally, subjection of both vinyl azole intermediates (14 and 15) to Pd/C-catalyzed hydrogenation conditions delivered the corresponding products, 3a and 7a. Imidazole 8a was prepared by nucleophilic substitution of 3-(2-bromoethyl)-1H- indole with imidazole (Scheme 3C).²¹ Finally, the remaining targets 5a and 9a were synthesized through a two-step route involving a base-mediated hydroazolation of aryl alkynes²² to generate vinyl pyrazoles 16 and 17, and subsequent Pd/C-catalyzed hydrogenation provided 5a and 9a (Scheme 3D).

Scheme 3. Synthesis of Desfluoroalkyl Azoles.

(i) pyrazole, 5% (PhSe)₂, 5% {Ir[dF(CF₃)ppy]₂-(5,5’-dCF₃bpy)}PF₆, 427 nm 40 W LEDs, DCE, 30 °C; (ii) azole, 5% (PhSe)₂, 3% 9-mesityl-3,6-di-t-butyl-10-phenylacridinium tetrafluoroborate, 427 nm 40 W LEDs, DCE, 30 °C; (iii) Ph₃P⁺(CH₂Br)Br^–, t-BuOK, 0 °C → rt; (iv) imidazole, Cs₂CO₃, 10% CuI, DMF, 120 °C; (v) H₂, 5% Pd/C, EtOAc/MeOH (4:1) or MeOH, rt; (vi) imidazole, 1,4-dioxane, 80 °C; (vii) pyrazole, K₃PO₄, DMSO, 120 °C.

CONCLUSION

In summary, this comparison of (azole)N–CF₂R and (azole)N–CH₂R MMPs has revealed fluorine-induced perturbations of medicinally-relevant physicochemical properties, such as reductions in azole hydrophilicity, exposed polar surface area, and pK_a. Importantly, fluorination reduces CYP-mediated oxidation of a labile vicinal position. Computed BDE values point to a fluorine-induced strengthening of vicinal benzylic C–H bonds as a contributing factor to the metabolic shift. Additionally, computational analysis suggests a limited impact of fluorination on conformational preferences within MMPs, further supporting the suitability of (azole)N–CF₂R as a direct substitution for (azole)N–CH₂R in optimization campaigns. Along with a newly provided synthetic method to install (azole)N–CF₂R,¹⁵ this information will facilitate rational incorporation of this fluorinated substructure in drug optimization campaigns.

SAFETY

No unexpected or unusually high safety hazards were encountered.

Supplementary Material

The Supporting Information is available free of charge on the ACS Publications website.

Synthesis and characterization of compounds, physicochemical property assays, biochemical assays, and computational details (PDF)

ABBREVIATIONS

(azole)N–CF₂R

N-α,α-Difluoroalkyl Azoles

MMPs

Matched Molecular Pairs

PK

Pharmacokinetic

PD

Pharmacodynamic

(azole)N–CF₃

N-Trifluoromethyl Azoles

(azole)N–CH₂R

N-Alkyl Azoles

EPSA

Experimental Polar Surface Area

NADPH

Nicotinamide Adenine Dinucleotide Phosphate

LM CL_hep

In Vitro Hepatic Clearance from Liver Microsomes [mL/min/kg]

CYP

Cytochromes P450

metID

Metabolite Identification

BDE

Bond Dissociation Enthalpy

¹H NMR

Proton Nuclear Magnetic Resonance Spectroscopy

DFT

Density Functional Theory