De novo Synthetic Approach to 2,4-diamino-2,4,6-trideoxyhexoses (DATDH): Bacterial and Rare Deoxy-amino Sugars

Abstract

A synthetic route to 2,4-diamino-2,4,6-trideoxysugar stereoisomers in 6–7 steps and 22–33% overall yield is described. A key step in this pathway is the carbonyl coupling of d- and l-threoninol or d- and l-allothreoninol to a phthalimido-allene mediated by chiral iridium-H₈-BINAP, which allows for installation of two new chiral centers in one, highly diastereoselective (> 20:1 dr) step. This approach provides a more concise, diastereoselective, and versatile method to access these deoxy-amino sugars than is currently available.

Graphical abstract

With antibiotic resistance on the rise for many ESKAPE pathogens, there is a need for alternative treatments and preventions of bacterial infection.¹ It is known that the lipopolysaccharide (LPS) and capsular polysaccharide (CPS) of bacteria are composed of so-called “rare” sugar chains that differ from those found in eukaryotes. The LPS and CPS are found on the extracellular matrix of bacteria and are known to play important roles in cellular recognition.^2–4 As a result, these glycans make attractive targets for carbohydrate-based vaccine development, providing an alternative to traditional antibiotics.^5–8 Among these bacterial sugars are various 2,4-diamino-2,4,6-trideoxyhexoses (DATDH), which are found in repeating units of the O-Antigen of many Gram-negative pathogenic bacteria including A. baumannii, C. jejuni, P. alcaifaciens, N. gonorrheae, N. meningitidis, P. aeruginosa, etc.^4,8–12 (Figure 1). Because these trideoxy sugars cannot be easily isolated from their natural sources, they have become attractive synthetic targets. Additionally, different stereoisomers of DATDH sugars are present on different bacterial serotypes, so there is a need for a widely applicable route to synthesize libraries of these compounds.

Figure 1.

(a) C. jejuni N-linked glycan (b) A. baumannii K17 capsular polysaccharide

Various methods of synthesis have been reported in the literature for this class of bacterial sugars. Many of these methods rely on the stereochemical manipulations of sugar building blocks.^10,13–22 Others start with commercially available amino acids that undergo late-stage cyclization after installation of various chiral centers.²³ While these methods have proven effective, they rely on synthetic routes that are limited in scope and/or proceed with low diastereoselectivity in the construction of key intermediates. Herein, we report an efficient, diastereoselective, and versatile protocol for the synthesis of four unique 2,4-diamino-2,4,6-trideoxy sugars. Retrosynthetically, we envisioned that all four 2,4-diamino-2,4,6-trideoxy pyranosides could arise from threoninols via a Krische allylation to set the required stereocenters. In the forward direction, oxidative cleavage of the resulting terminal alkenes followed by TBS protection would afford the protected deoxy-amino sugars 9, 10, 11, and 12 (Scheme 1).

Scheme 1.

Retrosynthetic analysis of deoxy-amino sugar synthesis starting from derivatives of l-threoninol, d-allothreoninol, l-allothreoninol, and d-threoninol

Accessing these deoxy-amino sugars first involved synthesis of the threoninol compounds. To this end, the d-threonine 13 was Boc protected at the amine position under standard conditions, giving rise to compound 14. Treating 14 with methyl iodide and K₂CO₃ resulted in the formation of methyl ester 15. This compound was then reduced under standard conditions with sodium borohydride to yield d-threoninol 16 in 80% yield over three steps (Scheme 2). With 16 in hand, we turned our attention to the key allylation reaction (Table 1). We recognized that optimization of this reaction would be necessary because, to our knowledge, the Krische allylation had never been performed on chiral α-amino β-hydroxy alcohols.²⁴ Under the originally reported conditions, only trace amounts of impure allylation product were observed (Entry 1). Increasing the catalyst loading from 0.05 to 0.1 equivalents led to an increase in conversion (Entry 2). The starting material conversion could be further increased by increasing the reaction time (Entry 3), with the bulk of starting material consumed within the first 96 hours. We then further increased the catalyst loading to 0.2 equivalents (Entry 5) and temperature to 110°C in a sealed tube (Entry 7). Pleasingly, these conditions provided the desired allylation product as a > 20:1 mixture of diastereomers and showed 58% conversion. Attempts to run the same reactions using the (S)-catalyst resulted in lower conversion, demonstrating a clear case of stereochemical match and mismatch between substrate and catalyst (Entries 4 and 6).²⁵ Additional optimization attempts with alternative chiral catalysts failed to provide an improvement (See Table S2.1) Alternative solvents also failed to provide any improvement (See Table S.2.2).

Scheme 2.

Original synthetic pathway toward threoninol compounds for the d-threonine model system

Table 1.

Key optimizations of allylation chemistry using d-threoninol as the model substrate

entry	catalyst chirality	catalyst loading (mol eq.)	time (days)	temperature °C)	percent conversiona	product ratio 8:17
1	(R)	0.05	2	100	trace	1:0
2	(R)	0.1	2	100	27%	1:0
3	(R)	0.1	7	100	41%	1:0
4	(S)	0.1	7	100	27%	0:1
5	(R)	0.2	4	100	51%	1:0
6	(S)	0.2	4	100	37%	0:1
7	(R)	0.2	4	110	58%	1:0

^a> 20:1 diastereomeric ratio determined.

To determine the stereochemistry of the C-4 hydroxyl on compounds 8 and 17, we carried out a Rychnovsky acetonide analysis.²⁶ Per this analysis, an acetonide protected syn-1,3-diol adopts a chair conformation, which is expected to possess acetonide methyl-group ¹³C chemical shifts with a > 9 ppm difference (Figure 2a). In contrast, the acetonide protected anti-1,3-diol adopts a twist-boat conformation to minimize 1,3-diaxial interactions. This would be identified by diagnostic acetonide methyl-group ¹³C chemical shifts with a < 9 ppm difference, respectively (Figure 2b). Based off this analysis, we were able to establish that compound 18 possessed a syn-1,3-diol configuration (~19 and 30 ppm ¹³C shifts) while compound 19 possessed an anti-1,3-diol configuration (~24 and 25 ppm ¹³C shifts) (S2.21 and S2.22).

Figure 2.

Stereochemical determination of 1,3-diols using Rychnovsky’s analysis (a) The acetonide protected syn-1,3-diol 18 with distinct acetonide methyl-groups highlighted in red (b) The acetonide protected anti-1,3-diol 19 with acetonide methyl-group highlighted in red

It was our initial hope that we could carry the crude allylation product 8 on to the subsequent oxidative cleavage. However, all attempts at ozonolysis led to decomposition. Noting the coordinating ability of diols, we reasoned that metal contaminants from the allylation may be causing problems in subsequent steps. To avoid this, we decided to protect the secondary alcohol prior to allylation. To this end, we examined different silyl protecting groups on the threoninol O-3 due to their inability to form a chelate with metal centers.^27,28 We also synthesized an oxazolidine protected threoninol to examine the effect of conformationally constraining substrates on the reaction (Scheme S2.1).²⁹

While changing protecting groups, we also sought to streamline the synthesis of the threoninol compounds. Synthesis of a TBS protected substrate 4 commenced with a two-step Boc protection and esterification of d-threonine 13 yielded methyl ester 15.³⁰ Subsequent treatment with freshly distilled TBSOTf and 2,6-lutidine provided the silyl protected methyl ester 20. Finally, reduction with LiBH₄ resulted in the desired d-threoninol 4 in 81% overall yield in two days of total reaction time (Scheme 3). These conditions, once realized, were used for the other substrates in this study.

Scheme 3.

Revised synthetic pathway toward threoninol compounds for the d-threonine model system employing a TBS protecting group

Attempts to synthesis more reactive silyl ethers (TES) were unsuccessful. The oxazolidine did not successfully undergo allylation under our optimized conditions, which rendered it unusable (Scheme S2.1). Ultimately, the TBS protected threoninol 4 was deemed the optimal substrate, providing the desired allylation product 21 in 43% isolated yield as a > 20:1 mixture of diastereomers (Scheme 4). Due to the presence of the Boc protecting group, rotamers in both ¹H and ¹³C NMR was not surprising and was confirmed using 1-D NOE experimentation according to the procedure of Ley and co-workers.³¹ Broadening of the carbon peaks on ¹³C NMR was also observed, as is expected for carbamate rotamers.³² The stereochemistry of the product was confirmed using the Rychnovsky analysis (S2.21).²⁶ Additionally, allylation product 8 (Table 1) and TBS deprotected product 8 (Scheme 4) were found to be the same compound, indicating that the protecting group did not alter the stereochemical outcome of the reaction in this case, but did allow for a more isolatable compound.

Scheme 4.

Synthesis of protected DATDH from d-threoninol in 33% overall yield, > 20:1 dr

Following allylation, attempts to subject 21 directly to oxidative cleavage led to complex mixtures resulting from partial loss of the TBS group. To circumvent this, the TBS group was removed prior to ozonolysis. Pleasingly, diol 8 underwent clean oxidative cleavage to afford 22a/b, which was directly protected at the anomeric position using TBS chloride to afford 12 as a single diastereomer (Scheme 4). Product 12 was isolated as a mixture of rotamers.^31,32

Having established a route to 12, we next turned our attention to the construction of other isomers. To this end, the allylation chemistry was applied to 1, 2, and 3 under match-pair conditions. Yields for each allylation product ranged from 40–56%, with a > 20:1 diastereomeric ratio in each case. The presence of rotamers in both ¹H and ¹³C NMR was again observed in each case.^31,32 This confirmation also simultaneously reaffirmed the purity and diastereoselectivity of each allylation product. The stereochemical relationship of every 1,3-diol was again confirmed using Rychnovsky’s analysis (S2.23-S2.25). ²⁶

With clean allylation intermediates 23, 25, and 27 in hand, facile TBS deprotection under standard conditions yielded the free 1,3-diols 5, 6, and 7 in quantitative yield.^31,32 Subsequently, ozonolysis was performed on all substrates. Similar to 8 the l-threoninol allylation product 5 successfully underwent oxidative cleavage and spontaneous in situ cyclization via ozonolysis.³³ Interestingly, the d-allo and l-allothreoninol allylation products 6 and 7 underwent decomposition upon exposure to ozone. As an alternative, 6 and 7 were subject to oxidative cleavage via catalytic OsO₄ paired with NaIO₄.^34,35 Subsequently, the hemiacetals of interest 24, 26, and 28 were TBS protected at the anomeric position to facilitate the stereochemical assignment at the C-2 position (Scheme 5–7). In all cases the products 9, 10, 11, and 12 were obtained exclusively as the β-anomer as evidenced by J-values > 8.0 Hz at the anomeric position. This analysis also allowed us to establish that the C-2 amine substituent was in an equatorial configuration. Notably, an anti-configuration between the C-3 and C-4 stereocenters was simultaneously established for the d- and l- threoninol products 21 and 23, while a syn-configuration was established about these positions for the l-allo and d-allothreoninol products 25 and 27. For DATDH 10 and 11, high temperature V-T NMR experimentation was conducted due to the presence of rotamers. Overall, the procedure resulted in a highly diastereoselective synthesis of two enantiomeric pairs of four, protected DATDH products ranging from 22–33% overall yield from their corresponding threoninol starting material.

Scheme 5.

Synthesis of protected DATDH from l-threoninol in 33% overall yield, > 20:1 dr

Scheme 7.

Synthesis of protected DATDH from l-allothreoninol in 22% overall yield, > 20:1 dr

In summary, we have demonstrated a facile route to unusual 2,4-diamino-2,4,6-trideoxyhexoses (DATDH) that is both flexible and highly diastereoselective. This improved synthesis relies on what is, to our knowledge, the first example of using Krische allylation chemistry on encumbered α-amino alcohols to set two chiral centers in one coupling step. The efficiency of this approach permits the construction of DATDH in seven or fewer steps and good yield. Furthermore, it is flexible enough to provide access to multiple diastereomers, some of which are not easily available from commercial sugars. The application of this technology to the construction of other microbial glycans of therapeutic interest is currently under investigation.

Scheme 6.

Synthesis of protected DATDH from d-allothreoninol in 28% overall yield, > 20:1 dr

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.