Phosphonoxins III: Synthesis of α-Aminophosphonate Analogs of Antifungal Polyoxins with Anti-Giardia activity

Abstract

A synthesis of α-aminophosphonate analogs of polyoxins, termed phosphonoxin C1, C2 and C3, has been achieved. The key step was the addition of lithium dimethyl phosphite to the aldehyde of a protected threose derivative. α-Hydroxyphosphonate analogs C4 and C5 were also obtained by taking advantage of an unprecedented conversion of an azide to hydroxyl during treatment with hydrogen on palladium on carbon. The resulting phosphonoxin C5 inhibited the growth of an intestinal protozoan, Giardia lamblia, at low micromolar concentration.

The natural products polyoxins isolated from the Streptomyces species, are a closely related class of peptidylnucleoside antibiotics.¹ They have shown to be potent inhibitors of many types of fungi and parasites due to their ability to disrupt cell envelope biosynthesis.² For example, one such parasite is Giardia Lamblia,³ a protozoan parasite that colonizes and replicates in the intestinal tract. It is one of the most common causes of diarrhea in the developed world and is not only restricted to humans, but affects livestock and companion animals, where there are an estimated 100 million cases per year worldwide.⁴ Therefore, the need for effective treatment continues to this date. Polyoxins⁵ most likely act by mimicking the stucture of, and thereby competing with, the natural substrate for cell envelope biosynthesis, uridine diphosphoryl-N-acetylglucosamine (UDP-GlcNAc, Figure 1).⁶ They have exhibited activity against many types of fungi⁷ and inhibit cell wall formation in Entamoeba.⁸ However, utility of these natural products as drugs is compromised by their poor bioavailability and metabolic instability resulting in low efficacy evidenced by high inhibitory concentrations against fungal pathogens.⁹ Over the years many groups have synthesized polyoxin analogs with the goal of obtaining compounds that would retain the anti-fungal and anti-parasitic activity of polyoxins but could be more suitable as drugs.^5,10

Figure 1.

UDP-GlcNAc, Polyoxins J and L and Phosphonoxins

We recently published the synthesis of a new class of polyoxin analogs, termed phosphonoxins, that replaced the peptide linkage to the nucleoside with a phosphonate linkage.^11,12 Phosphonates are chemically and enzymatically stable and many such derivatives are able to penetrate cells.¹³ One of the phosphonoxins we synthesized (here termed phosphonoxin A, Figure 1) is a potent inhibitor of Giardia trophozoite growth and cyst formation in vitro.¹¹ We also synthesized phosphonoxins B1 and B2, which more closely resemble the structure of the natural polyoxins.¹²

We report here the synthesis of phosphonoxins C1 (1) and C2 (2), which differ from the phosphonoxins B only in that they do not possess the methylene spacer between the amino phosphorus groups; they are α-instead of β-aminophosphonates. During the course of this synthesis we also obtained α-hydroxyphosphonate analogs, phosphonoxins C4 (4) and C5 (5), by the unexpected conversion of an azide into a OH group under catalytic hydrogenation conditions. Our approach to the synthesis of the phosphonoxins C involved a similar approach to that taken to synthesize phosphonoxins B.¹²

The α-amino stereocenter would be generated by reaction of a chiral aldehyde with the anion of dimethyl phosphite¹⁴ instead of sulfinimine with dimethyl methylphosphonate.¹⁵ We also would use a p-bromo benzoyl instead of a benzyl protecting group to aid in the crystallization of intermediates for the determination of stereochemistry by X-ray crystallography. We began with the reaction of commercially available diol 6 with one equivalent of 4-BrBz-Cl in pyridine to give predominantly monobenzoylated product 7 (Scheme 1). Oxidation of alchohol 7 with IBX¹⁶ generated aldehyde 8. This aldehyde was then treated with the lithium anion of dimethyl phosphite¹⁷ to produce α-hydroxyphosphonate diastereomers 9a and 9b in a 3:2 ratio in 65% combined yield. These isomers were readily separated by column chromatography. Both isomers, 9a and 9b, were then independently converted to the corresponding azides¹⁸, 10a and 10b, via Mitsunobu substitution¹⁹ or triflation of the hydroxyl group followed by nucleophilic displacement with azide.

Scheme 1.

Synthesis of α-diazophosphonates

The stereochemistry of 10b was confirmed by condensation of aldehyde 8 with (R)-p-toluenesulfinamide^20,21 to produce (R)-sulfinimine 11 in low yield due to cleavage of the benzoate moiety (Scheme 2). This sulfinimine was then reacted with the lithium anion of dimethylphosphite to give phosphonate 12a with high selectivity (90% de) over its diastereomer 12b. The absolute configuration of 12a was confirmed by X-ray crystallography.

Scheme 2.

α-diazo stereocenter confirmation of 10b

Converting sulfinamide 12a, of known stereochemistry, to amine 13 by acid catalyzed cleavage of the sulfinyl group followed by diazo transfer with TfN₃ and catalytic CuSO₄²² gave an α-aminophosphonate identical in all aspects to 10b. As the diazo transfer is known to proceed with retention of configuration,²² the structure of 10b was determined as depicted. The azides 10a and 10b were next reacted independently with LiBr at 70 °C to cleave one of the phosphonate methyls²³ giving monomethyl esters 14a and 14b, respectively (Scheme 3). The resulting phosphonic acids were then coupled with 2′,3′-O-isopropylidineuridine under Mitsunobu conditions,²⁴ followed by removal of the remaining phosphonate methyl by again heating in the presence of LiBr to give uridine phosphonates 15a and 15b. The benzoate ester was cleaved by heating at 50 °C in methanolic ammonia, resulting in 16a and 16b. Carbamates 17a and 17b were produced by treating 16a and 16b with trichloroacetyl isocyanate followed by ammoniolysis of the resulting trichloroacetyl group.²⁵ Phosphonoxin C2 was finally obtained via removal of the acetonide group of compound 17b by heating in 80% AcOH (Scheme 4) followed by the reduction of the azide under Staudinger conditions²⁶ to give uridine amine 2 in 34% yield.

Scheme 3.

Uridine Phosphonoxin intermediates

Scheme 4.

Phosphonoxin C2

To avoid the low yields of the Staudinger reduction, the azide compounds 16a and 17a were instead reduced by hydrogenation with catalytic 5% Pd on BaSO₄ (Scheme 5) in methanol to give amines 19 and 21 as the major products in 72% and 48% yields, respectively. Removal of the acetonide protecting group of each by heating in 80% acetic acid in water gave phosphonoxin C3 (3), lacking the carbamate moiety, and phosphonoxin C1 (1). To our surprise, the palladium mediated azide reduction reaction also gave significant amounts (16% and 34% yields) of methyl α-hydroxyphosphonates 20 and 22 from 16a and 17a, respectively (Scheme 5). We investigated this process further and found that the reaction requires hydrogen gas to be present, as no reaction occurs with 16a and 17a when only the palladium catalyst is present. The identities of 20 and 22 were confirmed by cleaving the methyl ester with LiBr to obtain 23 and 24, each with 94% diastereomeric purity, then synthesizing 23 by an alternate route from 9a (Scheme 6).

Scheme 5.

Phosphonoxins C1, C3, C4 and C5

Scheme 6.

α-hydroxy stereocenter confirmation of 23

Protection of the free OH group of 9a as an acetate group, coupling with 2′,3′-isopropylidineuridine followed by ammonia deprotection gave 23. Compound 23 obtained via this alternate route was identical to major isomer 23 obtained from 20. Removal of the acetonide groups of 23 and 24 gave α-hydroxyphosphonate analogs, phosphonoxin C4 (4) and C5 (5) (Scheme 5). To our knowledge this is the first example of conversion of an azide to an alcohol by catalytic hydrogenation.

In the treatment of azide 16a with H₂ and Pd on BaSO₄ we propose the methyl phosphonate ester product 20 is consistent with a reactive intermediate involving the phosphonate moiety (Scheme 7). Imine intermediates such as structure 27 have been proposed for catalytic hydrogenation of azides²⁷ and oxaphosphiranes similar to proposed intermediate 28 have been described in other reaction pathways.²⁸ Attack of methanol on such an intermediate should give heminaminal 29, that after collapse to an α-ketophosphonate and stereoselective reduction would give the observed product 20.²⁹

Scheme 7.

Proposed mechanism of Azide to Alcohol conversion

The phosphonoxins were examined for their inhibition of Giardia lamblia growth in culture (Table 1).³⁰ To our surprise, phosphonoxin C5 (5) obtained from the unique conversion of azide to alcohol was by far the most active, with an IC₅₀ of 2.3 μM.

Table 1.

Inhibition of Giardia lamblia trophozoite growth.

compound	IC50 (μM)
Phosphonoxin C1 (1)	>20
Phosphonoxin C2 (2)	>20
Phosphonoxin C3 (3)	>20
Phosphonoxin C4 (4)	>20
Phosphonoxin C5 (5)	2.3

In conclusion we have synthesized three novel α-aminophosphonate analogs of polyoxins, which we termed phosphonoxins C1, C2 and C3. In addition, we synthesized two α-hydroxyphosphonate analogs, C4 and C5, from the co-products of the azide reduction, the latter of which inhibited Giardia trophozite growth. The scope and mechanism of the rearrangement that results in the co-products is still under investigation. Further chemical and biological studies of the phosphonoxins and the α-hydroxyphosphonates will be described in due course.