Incorporation of an Isohexide Subunit Improves the Drug-like Properties of Bioactive Compounds

Abstract

An enhanced ability to pre-engineer favorable drug-likeness qualities into bioactive molecules would focus and streamline the drug development process. We find that phenols, carboxylic acids, and a purine react with isosorbide (“GRAS” designated) under Mitsunobu coupling conditions to deliver the isoidide conjugates selectively and efficiently. Such conjugates show improved solubility and permeability properties compared with the bare scaffold compounds themselves, and the purine adduct may have applications as a 2′-deoxyadenosine isostere. We anticipate additional benefits, implied by their structures, in metabolic stability and reduced toxicity of the isoidide conjugates as well.

Effective new methods to streamline drug development and address issues of drug-likeness earlier in the process would lead to savings in time and expense.¹⁻¹⁰ We have found that incorporation of an isohexide substructure into bioactive molecules can dramatically improve their solubility, permeability, bioavailability, and other attributes.

Isosorbide (1, Chart 1) is a bowl shaped bis(ether) diol, MW 146, mp ∼64 °C, with high water solubility. It is inexpensive, cellulose sourced,^11,12 thermally and acid/base stable, and nontoxic (FDA designated GRAS: “generally recognized as safe”). As a component of the prodrug isosorbide mononitrate, an FDA approved treatment for angina,¹³1 confers useful drug-like properties, including excellent solubility and permeability, and favorable metabolic stability. The adoption of 1 and the two other isohexide stereoisomers, namely, isoidide (2) and isomannide (3), as “green” components of polymers, plasticizers, liquid crystals, and other materials is an active area of research.¹⁴⁻¹⁸ There are now a variety of examples of the incorporation of isohexides into bioactive scaffolds.¹⁹⁻³¹

Chart 1. Structures of the Isohexide Stereoisomers.

The isohexide unit embodies a special combination of desirable features that make it a compelling candidate for wider and deeper exploitation: (1) multiple uncharged hydrogen bond accepting and donating sites that improve water solubility and that can, potentially, interact with protein receptors as well; (2) rigidity (few rotatable bonds, an important consideration for oral availability³²), chirality, and three dimensionality³³ that improve permeability; (3) resistance, as a scaffold or byproduct, to first pass metabolism;³⁴ and (4) versatility in its transformations and coupling reactions. In principle, significant improvements in drug-likeness could be achieved by a simple structural change to known bioactive scaffolds that features the attachment or incorporation of an isohexide subunit. The modifications described herein are outside the normal range of many medicinal chemistry SAR investigations, which might feature incremental and/or site-specific structural changes. Instead, we employ the efficient and selective Mitsunobu coupling reaction of 1 that allows the one-step incorporation of the isoidide unit into bioactive phenols and carboxylic acids.

The Isosorbide Mitsunobu Reaction

Isosorbide (1) features two secondary hydroxyls, one that is exo with respect to the bicyclic ether scaffold and less-hindered (O-6, convex face), and the other that is endo and can intramolecularly H-bond (O-3, concave face) (Figure 1). They are often quite similar in reactivity.³⁵ Only a select few reactions allow site selective transformation at O-6 (e. g., enzymatic O-3 deacetylation^36,37 or NaH promoted benzylation³⁸) or O-3 (e. g., thionyl chloride promoted chlorination,³⁹ PbO promoted acetylation,⁴⁰ or organocatalyzed esterification³⁵). The monoprotected derivatives of 1 can be further transformed at the free hydroxyl in predictable ways, but this adds steps to the synthetic routes. We have discovered a straightforward Mitsunobu procedure⁴¹⁻⁴⁶ that leads to direct replacement of the endo hydroxyl of 1 by phenolic or carboxylic oxygen nucleophiles with excellent yield and site selectivity.

Figure 1.

4-Hydroxyquinolone-derived isoidide conjugates prepared by Mitsunobu coupling of isosorbide (1).

Several examples (4a–d, Scheme 1) illustrate the power of this transformation with phenolic nucleophiles. The new aryloxy group is introduced at isoidide C-3 in a single step without resorting to protecting group chemistry, and the exo hydroxyl and aryl substituents are left free for further transformation. The isoidide stereochemistry (both substituents exo) is confirmed by the vicinal coupling constants between the bridgehead H’s and the adjacent CHX (∼1 Hz for endo H; ∼ 5 Hz for exo H).³⁹ Successful Mitsunobu coupling reactions of 1 begin by preforming the triphenylphosphine–diisopropyl azodicarboxylate adduct at 0–5 °C in dry, peroxide-free THF. The nucleophile is then added, followed by 1. The reaction is complete with 36–48 h at 23–60 °C. A silica gel column chromatography is required in most instances, as triphenylphosphine oxide invariably contaminates the crude coupled product. Nevertheless, a column-free gram scale experimental procedure for the synthesis of 4d is given in the Supporting Information. We also illustrate the usefulness of nitro as a precursor group for subsequent amidation or other coupling steps. To that end, the nitro group in 4d was selectively reduced (zinc dust and ammonium chloride in aqueous methanol) to provide aniline 5 (Scheme 1).

Scheme 1. endo-Selective Mitsunobu Reaction of Isosorbide with Phenolic Nucleophiles.

The mechanism of the Mitsunobu reaction⁴⁶ features a tetracoordinate oxyphosphonium intermediate (e.g., 6, Scheme 2) wherein phosphorus is bound to the carbinol oxygen atom. S_N2 displacement, with inversion of configuration at the carbinol carbon, by an anionic nucleophile releases triphenylphosphine oxide and provides the product of substitution. For selective formation of the exo substitution products 8, the precursor oxyphosphonium intermediate (6) must have formed at C-3. When isosorbide was subjected to the Mitsunobu conditions in the absence of an added nucleophile (Scheme 2), the product of intramolecular displacement, trianhydromannitol 9,⁴⁷ was isolated as the major product. Cyclization to give 9 requires that the alternative oxyphosphonium ion intermediate (7) must have formed from 1 at C-6. While the site of initial oxyphosphonium formation is not revealed, there is no requirement that the site selectivity of the Mitsunobu reaction of 1 is governed by selective kinetic formation of (the more hindered) isomer, 6. A more likely rationale, given the generally greater reactivity of the isosorbide (endo) 3-O-sulfonates with nucleophiles,³⁹ is that the Mitsunobu site selectivity is governed by the better stereoelectronic S_N2 alignment toward intermolecular displacement of the endo oxyphosphonium intermediate 6 relative to the alternative, 7.

Scheme 2. Two Isosorbide-Derived Oxyphosphonium Intermediates.

The 4-hydroxy-3-carboxy-quinolin-2-one scaffold families appear in a wide variety of bioactive compounds intended for drug development targeted to, for example, Alzheimer’s disease,⁴⁸ multiple sclerosis,⁴⁹ tuberculosis,⁵⁰ bacterial infection,⁵¹ inflammation,⁵² and cancer.⁵³ However, many representatives exhibit undesirable features such as poor water solubility and metabolic vulnerabilities. Incorporation of the isoidide unit at C-4 can enhance polarity, three dimensionality, and resistance to first pass metabolism. Three N-benzyl-4-hydroxyquinolones were coupled with isosorbide under Mitsunobu conditions, resulting in the derivatives 10–12 (Scheme 3). In one case, prolonged exposure to excess diisopropyl azodicarboxylate and triphenylphosphine gave the product of elimination (13) rather than a second substitution. This potentially useful transformation allows conversion of a phenolic scaffold to less polar isohexide conjugates as compared with 10–12. An attempt to convert 11 to the corresponding secondary fluoride derivative with DAST⁵⁴ led instead to the alkene, 14. These two eliminations reflect the tendency of exo-situated isohexide leaving groups to eliminate in preference to the substitution alternative when treated with bases that are weak nucleophiles.^39,55−57 However, strong nucleophiles, such as azide and phthalimide anions, can effect disubstitutions.⁵⁸⁻⁶¹

The computed structure⁶² of isosorbide (1) reveals a dihedral angle of 160° between the C–O bond to exo–OH (at C-6; see Chart 1) and the trans, vicinal endo C–H at C-5 (H shown in green in Chart 1), which is close to the optimal anti transition state for an E2 reaction.⁶³ In contrast, the endo OH at C-3 has trans, vicinal CHs at 87.1° (exo, H-2) and 77.9° (exo, H-3a), neither of which is well-aligned for elimination. Thus, there is a stereoelectronic preference for E2 elimination of an isosorbide exo leaving group at C-6 vs an endo leaving group at C-3.

The site-selective one-step Mitsunobu coupling of 1 with bioactive carboxylic acids can be exploited to prepare isoidide ester derivatives with potential as prodrugs. Three Mitsunobu coupling products (15–17) are shown in Figure 2 along with the yields obtained after reaction at 23 °C for 48 h. The carboxylic acid precursors to 15 and 16 have been reported as potent α4β1/α4β7 receptor antagonists.^64,65 The precursor to 17, monomethyl fumarate (MMF), has been FDA-approved in prodrug form for multiple sclerosis treatment.⁶⁶ We find that isoidiode ester 17 releases MMF in rats at about half the rate shown by dimethyl fumarate (Figure 3), suggesting that the methyl ester and the isoidide ester linkages are similarly labile. A related diester, bis(methyl fumaryl)isosorbide, has been recently evaluated as a treatment for psoriasis.³¹

Figure 2.

Ester isoidide conjugates prepared by Mitsunobu coupling of isosorbide (1) with carboxylic acids (reaction at 23 °C, 48 h).

Figure 3.

Measured blood level of monomethyl fumarate released from isoidide fumarate ester 17 (red trace) compared with dimethyl fumarate (LC-MS/MS analysis, oral dosing, Sprague–Dawley rats, n = 6).

Mitsunobu coupling of 1 with 6-chloropurine⁶⁷ (65 °C, 48 h, 55% yield) led to the pseudonucleoside, 18 (Figure 4). An overlapped computed structure (MOE)⁶⁸ of 18 with computed 2′-deoxyadenosine (19) is depicted as 20. The nucleoside is shown in the anti-south conformation.⁶⁹ The spatial near-overlap of the respective 3 and 5′ hydroxyls is indicated by the red boxes. Compounds derived from 18 may prove useful as nucleoside isosteres that feature stability toward depurination (because there is no anomeric center) as well as reduced polar surface area.⁷⁰⁻⁷³

Figure 4.

Nucleoside isostere (18) prepared by Mitsunobu coupling of isosorbide (1) with 6-chloropurine, and structural comparison of 18 (dark gray carbons, green chlorine) with 2′-deoxyadenosine (19, light gray carbons, blue nitrogens, red oxygens).

Improvements in PK Properties

What level of improvement in the PK properties of a drug candidate might result upon incorporation of an isohexide unit? A 4-anilino-quinazoline example is used to illustrate (Table 1). The isoidide conjugate 22 was designed to improve upon the EGFR inhibitor 21, a standard comparison compound for the design of antitumor compounds.⁷⁴ Assembly of the analogue 22,⁷⁵ in which the fluoro substituent of 21 is an isoidide unit instead (Table 1, red boxes), was achieved through reaction of 4-chloro-6,7-dimethoxyquinazoline with the reduced Mitsunobu product 5 (isopropanol, reflux, 4 h, 53% yield). In vitro EGFR inhibitory potency is in fact somewhat improved (5.3 nM for 22 vs 9 nM for 21). Additionally, substantial increases are seen in measured solubility and permeability (PAMPA method), as well as relative AUC and C_max values in mice. While the solubility and permeability improvements might have been anticipated, the increase in oral exposure in this case is a testament to the potential of the isosorbide unit to improve PK aspects without adversely affecting target binding. The increase in molecular weight of the isoidide conjugate (21 MW 334.7; 22 MW 459.9) is not an important limiting factor here.

Table 1. PK Comparison of the EGRF Inhibitor, 21, with an Isoidide Substructure Analogue (22).

	compound
	21	22
human EGFR inhibition, IC50	9 nM	5.3 nM
pKa measured	4.51	5
solubility PBS at pH 7.4	1.3 μg/mL	19.82 μg/mL
solubility PBS at pH 9.0	1.53 μg/mL	45.41 μg/mL
solubility simulated gastric fluid	49.42 μg/mL	87.86 μg/mL
solubility simulated gastric fluid 0.5% SDS	23.61 μg/mL	56.01 μg/mL
solubility simulated intestinal fluid	51.72 μg/mL	65.40 μg/mL
solubility simulated intestinal fluid 0.5% SDS	63.88 μg/mL	80.79 μg/mL
PAMPA 10–6 cm/s	precipitated	7.45
[PAMPA, propanolol, + control, 3.2]
terminal half-life, male C57BL/6 mice, oral	1.90 h	1.19 h
AUC(0-t) oral 4–10 h	25480 ng·h/mL	44496 ng·h/mL
AUC(0-inf) oral	26281 ng·h/mL	44530 ng·h/mL
Cmax oral	7441 ng/mL	30085 ng/mL

An Isohexide-Based Prodrug Example

The improvement in PK properties upon conversion of a drug candidate to an isosorbide-linked prodrug⁷⁶ is illustrated by the case of the published Roche compound, 23, a potent integrin α4β1/α4β7 antagonist.⁶⁵ Alpha4 integrins play a major role in a number of inflammatory diseases, including asthma, rheumatoid arthritis, multiple sclerosis, inflammatory bowel disease, and atherosclerosis.⁷⁷ PK data for 23 and its isoidide ester, the prodrug 16, are given in Table 2. Ester 16 retains the excellent solubility properties of carboxylic acid 23. Both compounds are highly active in vitro; the ester is rapidly cleaved to 23 in plasma. In fact, the resulting 23 (from cleavage of 16) shows improved exposure numbers in mice. Furthermore, 23 delivered in this way locates effectively in the mouse colon, the site of its biological activity and the possible target for anti-IBD treatment (see the Supporting Information). The ethyl ester and several other (more polar) esters of 23 proved inferior to 16.⁶⁵ Again, the molecular weight of prodrug 16 (598.1) is not an important factor.

Table 2. PK Comparison of an Integrin α4β1/α4β7 Antagonist (23) with Its Isoidide Ester (16).

	compound
	23	16 (prodrug)
logD (measured, pH 7.4)	–1.1	1.2
pKa (measured)	3.4	7.9
solubility in PBS
at pH 7.4	95.4 μg/mL	107.3 μg/mL
solubility, simulated gastric fluid	94.7 μg/mL	125.5 μg/mL
solubility, simulated intestinal fluid	91.9 μg/mL	99.6 μg/mL
PAMPA 10–6 cm/sec	0	1.68
MDR-MDCK permeability	1.3 × 10–6 cm/s	2.5 × 10–6 cm/s
hydrolysis to 23 in plasma		90%, 2 h
plasma levels of 23 (PO) in mice	at 50 mg/kg	at 63.5 mg/kg
AUC(0–6 h), 23	616 ng·h/mL	2100 ng·h/mL
Cmax, 23	425 ng/mL	3193 ng/mL
tmax, 23	1 h	0.25 h
t1/2, 23	0.67 h	1.63 h
oral bioavailability	<1%	12%
[23] in mice, from 16		at 100 mg/kg oral
in plasma, 3 h	500 ng/mL
in plasma, 6 h	200 ng/mL
in colon, 3 h	200,000 ng/g
in colon, 6 h	100,000 ng/g

Conclusion

The one-step incorporation of an isoidide substructure into bioactive phenols, carboxylic acids, and a purine has been accomplished by a site-selective Mitsunobu coupling reaction with isosorbide (1). Improved drug-like properties of the resulting ethers and esters have been demonstrated in several applications, including an EGFR inhibitor (22) and an integrin α4β1/α4β7 antagonist (16). The purine adduct (18) may serve as a nucleoside isostere.

Glossary

Abbreviations

GRAS

generally recognized as safe

SAR

structure–activity relationship

DAST

(diethylamino)sulfur trifluoride

MMF

monomethyl fumarate

MOE

molecular operating environment

PK

pharmacokinetic

EGFR

epidermal growth factor receptor

PAMPA

parallel artificial membrane permeability assay

AUC

area-under-the-curve

C_max

maximum concentration

IBD

inflammatory bowel disease

CRO

contract research organization

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00476.

• Full experimental procedures, including spectroscopic characterization of new compounds; copies of H-1 and C-13 spectra (PDF)

• Details of the physicochemical and PK characterization of 16, 21, 22, and 23 (PDF)

Author Contributions

A.S. and S.K. designed the research and wrote the manuscript. A.S. and M.J.D. carried out the research. All authors have given approval to the final version of the manuscript.

We acknowledge shared support of this work by Aunova Medchem and the Rutgers HealthAdvance program, partially supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number U01HL150852.

The authors declare no competing financial interest.