Introduction

Iminosugars are analogues of carbohydrates, chemically named as polyhydroxylated secondary and tertiary amines and found to be widespread in plants and microorganisms. Thanks to their structural similarity to sugar molecules and excellent metabolic stability, iminosugars are endowed with a high pharmacological potential for a wide range of diseases such as viral infections, tumor metastasis, AIDS, diabetes and lysosomal storage disorders1,2,3,4,5,6,7,8,9,10,11.

Iminosugars are generally classified into five structural classes: pyrrolidines, piperidines, indolizidines, pyrrolizidines and nortropanes12. Hydroxypiperidines are structurally six-membered iminosugars. Some of the hydroxypiperidines such as 1,5-dideoxy-1,5-iminohexitol derivatives have now been commercialized as drugs to treat type II diabetes mellitus, type I Gaucher disease, Niemann-Pick disease type C (NP-C) and Fabry disease13,14,15,16,17,18. Other Hydroxypiperidines like 1,5-Dideoxy-1,5-imino-ribitol and 1,5-Dideoxy-1,5-imino-arabinitol derivatives have also attracted considerable attention as enzyme inhibitors that mimic glycoside and nucleoside substrates. For example, 1,5-dideoxy-1,5-imino-ribitol derivatives was found to be a potent inhibitor of bovine β-galactosidases and almond β-glucosidase19, while 1,5-Dideoxy-1,5-imino-arabinitol N-carboxypentyl derivatives permitted the isolation of pure α-l-fucosidase from bovine kidney homogenate20.

There are a diversity of synthetic methodologies have been developed to access iminosugars in hydroxypyrrolidines series21,22,23,24,25, while the reports for hydroxypiperidine iminosugars synthesis, especial for hydroxypiperidine of pentitols, are limited26. Therefore, there is a need to develop a simple method for the preparation of 1,5-Dideoxy-1,5-imino-ribitol and 1,5-Dideoxy-1,5-imino-arabinitol (Fig. 1). In this report, we like to present an efficient and environmentally friendly route for 1,5-dideoxy-1,5-imino-ribitol and 1,5-dideoxy-1,5-imino-l-arabinitol synthesis (Scheme 1) from d-ribose. The synthetic strategy employs many principles of Green Chemistry such as avoiding toxic or noxious chemical, and recycling and reusing the reagents27.

Figure 1
figure 1

Structures of some hydroxypyrrolidine and hydroxypiperidine iminosugars.

Full size image
Scheme 1
scheme 1

Synthesis route from d-ribose to 1,5-dideoxy-1,5-imino-d-ribitol 4 and 1,5-dideoxy-1,5-imino-l-arabinitol 6.

Full size image

The oxidation of d-ribose to d-ribonolactone is the critical and also the most difficult reaction in our proposed route. Pd–Bi/C heterogeneous catalyst28 with molecular oxygen has been successfully applied to directly convert d-ribose to d-ribonolactone. However, Bi–gluconate complexes formed due to the interaction between the leached Bi (from Pd-Bi/C catalyst) and glucose substrate would contaminate the gluconic acid products. As a result, further purification of products are needed, introducing large amounts of metal pollution29. Additionally, to prevent the catalyst poisoning due to the adsorption of the product during reaction, the reaction medium has to be maintained at pH 9.0 by continual addition of alkaline solution. Thus, the alkaline systems induce plenty of wastes (e.g., inorganic salts). Hence, the alkaline-free direct oxidation of d-ribose over heterogeneous catalysts is an ideal green process to pursue. Recently, Song Guo et al.30 reported a simple incipient wetness protocol to prepare ultra-small gold clusters on TiO2 through using anthranilic acid as a stabilizing agent. The resultant Au/TiO2 catalyst exhibits excellent catalytic activity in the alkaline-free oxidation of glucose. Because of the excellent material physical properties of γ-Al2O3 such as high thermal stability, large surface area, and superior mechanistic strength, highly mesoporous γ-Al2O3 has been widely used in the chemical industry as catalyst support. Moreover, there are abundant of cation vacancies on γ-Al2O3 surface31. This unique surface structure of γ-Al2O3 helps to stabilize Au clusters and prevent clusters agglomeration on its surface32. Thus, we decided to explore ultra-small gold clusters on γ-Al2O3 (Au/Al2O3) as catalyst for alkaline-free oxidation of d-ribose. To achieve small gold particles (2–3 nm) deposited on Al2O3 support, a simple solid grinding method was used in catalyst preparation33. The applicability of the Scheme 1 to the synthesis of 1,5-dideoxy-1,5-imino-d-arabinitol 10 starting from d-lyxose was also attempted (Scheme 2).

Scheme 2
scheme 2

Synthesis route from d-ribose to 1,5-dideoxy-1,5-imino-d-arabinitol 10.

Full size image

Results and discussion

Catalyst characterization

The Au/Al2O3 catalysts were prepared through the solid grinding method33. The Al2O3-supported gold clusters were calcined in static air at 300 °C, and the sample is denoted Au/Al2O3-Air. Prior to use, the catalysts were reduced under H2 flow for 2 h at 150 °C and denoted Au/Al2O3-H2. The nitrogen adsorption–desorption isotherms (Fig. 2) of catalyst samples with type IV shape designated the presence of mesopores with uniform pore size distribution34. Table 1 presents the composition, BET surface area, pore volume, average pore diameter of γ-Al2O3 support and Au/Al2O3 catalysts. The BET surface area, pore volume and average pore diameter were found to be decrease slightly with the loading of Au content, showing that Au was deposited into the pores of γ-Al2O3 support.

Figure 2
figure 2

Adsorption/desorption isotherms of γ-Al2O3 and Au/Al2O3 samples.

Full size image
Table 1 Textural properties of γ-Al2O3 and γ-Al2O3 supported nano Au catalysts.
Full size table

XRD and TEM were then used on analysing these catalysts to determine the differences in the size and distribution of gold particles on γ-Al2O3 support. As shown in Fig. 3, no reflections associated with Au nanoparticles were found on XRD patterns of the Au/Al2O3. Actually, the diffraction pattern Au/Al2O3 is well in accordance with γ-Al2O3 supports (JCPDS 29-63). The XRD results indicate that the gold particles are uniformly dispersed on the oxide surface.

Figure 3
figure 3

XRD spectra of γ-Al2O3 and Au/Al2O3 samples.

Full size image

As shown in Fig. 4, the gold nano particles are uniformly distributed on Al2O3-H2 support. The average particle size of an Au/Al2O3-H2 catalyst is 2–3 nm, which was similar to particle size reported in the literature33. The small clusters and uniform dispersion results are in agreement with XRD analysis.

Figure 4
figure 4

STEM image and size distribution of Au/Al2O3-H2 sample.

Full size image

To understand the surface state of Au particles on γ-Al2O3 surface, the XPS Au 4f spectra were carefully examined for the catalysts before and after H2 reduction. Based on some XPS spectra obtained from various Au catalysts35,36 previously, Au 4f5/2 87.7 eV and Au 4f7/2 84.0 eV could be referred as neutral Au species. As shown in Fig. 5, compared with Au/Al2O3 without H2 reduction, H2 reduction reduce positive gold species and lead Au 4f peaks shift upward in binding energies (BEs) by 0.3–0.4 eV to neutral Au species. The XPS results reveal that the H2 reduction could help to reduce positive gold species and produce neutral Au species on the γ-Al2O3 surface.

Figure 5
figure 5

XPS spectra of γ-Al2O3 and Au/Al2O3 catalysts for Au 4f.

Full size image

Synthesis of 1,5-dideoxy-1,5-imino-d-ribitol 4 and 1,5-dideoxy-1,5-imino-l-arabinitol 6 from d-ribose

Oxidation of d-ribose

In the first step as outlined in Scheme 1, the aerobic d-ribose oxidation was carried out in pure water using O2 gas (1 MPa) as an oxidant agent. We first examined the catalytic performance of conventional gold nanoparticles (Au/C of 3–5 nm Au particle size) and commercial Pd/C catalyst as well as the synthesized Pd–Bi/C catalyst28. It is found that the Au/C catalyst gave a 72% d-ribose conversion with a relatively low ribonolactone selectivity of 82% (Table 2, entry 2). By-products of glutarate and oxalate were formed due to overoxidation and degradation of d-ribose, respectively. The Pd/C and Pd–Bi/C catalysts showed much lower catalytic performance (45% and 36% for d-ribose conversion (Table 2, entries 3 and 4) due to catalyst poison28. Intriguingly, their selectivities to ribonolactone was low (≤ 20%)due to the over-oxidized targeted products. For Au/Al2O3-Air catalyst, 24% low yield with > 95% selectivity for ribonolactone was obtained. Interesting, simply pretreated the Au/Al2O3-Air under H2 flow for 2 h at 150 °C could significantly improve its catalytic activity, and 93% conversion with > 95% selectivity to ribonolactone (Table 2, entry 6) was achieved. Of note, Au/Al2O3 catalyst showed low activity under relatively low reaction temperature conditions (Table 2, entries 7).

Table 2 d-Ribose oxidation under different conditions.
Full size table

Song Guo et al.30 proposed that the glucose and dioxygen molecules are adsorbed and activated on the Au0 sites and support, respectively. The occupancy of Auδ+ species at the particles surface by water solvents and oxygen species would block the nearby active Au0 sites for glucose and O2 absorption, which would result in a low catalytic performance. From the XPS results (Fig. 5), pretreating the Au/Al2O3-Air catalyst under H2 could almost fully reduce surface Auδ+ species to Au0 species, which helps to reduce/remove the blocking effect from Auδ+ species and improve the catalytic activity.

The recyclability of Au/Al2O3 was evaluated under the same reaction conditions using recycled catalysts. Five cycles were carried out for d-ribose oxidation at 100 °C for 2 h under 1 MPa O2 using a recycled Au/Al2O3 catalyst and fresh reactants. After each reaction, the catalyst was simply recovered by filtration and washed alternately with water and ethanol followed by drying before being used again. The used catalyst was tested in 5 batch reactions without loss of activity and selectivity (Fig. 6). Hence, the results showed that Au/Al2O3 is a reusable and selective catalyst for d-ribose oxidation.

Figure 6
figure 6

Conversion and selectivity of d-ribose oxidation with recycled Au/Al2O3 catalyst.

Full size image

Acetalization of d-ribonolactone to 2,3-O-isopropylidene-d-ribonolactone

It was found that anhydrous CuSO4 served as critical Lewis catalyst28 for the conversion of ribonolactone to 2,3-O-isopropylidene-d-ribonolactone. Although relatively high yield (72%) was achieved, the CuSO4 is hard to be fully recovered due to its excellent solubilty in water. Marek Marczewski37 reported a SO42−/Al2O3 system, where the acid strength of SO42−/Al2O3 catalyst could be tuned based on the concentration of H2SO4 solution through incipient wetness method. Therefore, we next optimized the condition of SO42−/Al2O3 catalysts for the acetalization of ribonolactone with acetone to produce 2,3-O-isopropylidene-d-ribonolactone.

As shown in Table 3, no 2,3-acetonide 3a was formed when there was no catalyst in the reaction system. The addition of anhydrous CuSO4 improved the yield to 68% when the reaction mixture was refluxed at 60 °C for 2 h (Table 3, entry 2). Interestingly, the surface sulfates of γ-Al2O3 created through H2SO4 solution treatment could increase 2,3-acetonide 3a yield significantly: The yield for commercial γ-Al2O3 is only 37%, while the reactions using 1 wt.% SO42−/Al2O3 and 3 wt.% SO42−/Al2O3 produced 71% and 85% 2,3-acetonide 3a, respectively. However, the use of 9 wt.% SO42−/Al2O3 resulted in charring and only 11% yield of 3a was obtained (Table 3, entry 7). In addition, extending the refluxing time to 4 h with 3 wt.% SO42−/Al2O3 catalyst led to the formation of some unknown products (Table 3, entry 6).

Table 3 Acetalization of ribonolactone with acetone.
figure a
Full size table

To further understand the catalytic performance of the Al2O3 samples, catalysts were examined with NH3-TPD and Pyridine-IR. As shown in Fig. 7, all the catalysts show two NH3 desorption peaks at around 104 and 524 °C, which are attributed to the weak and strong acid sites39,40,41, respectively. In addition, the third peak at around 800 °C was observed on SO42−/Al2O3 samples. The NH3 desorption peak at 800 °C could be ascribed to the super strong acid sites, demonstrating the presence of super-acidic centers on SO42−/Al2O3 catalysts. The peak at 800 °C was increased with the introduction of more SO42− species. Figure 8 shows the Pyridine-IR of the catalysts. Note that both Brønsted and Lewis acid are presence on SO42−/Al2O3 catalyst, whereas the γ-Al2O3 has a very little peak at 1540 cm−1, indicating that there is only a small amount of Brønsted acid sites on γ-Al2O3 surface. Thus, the good catalytic activity of the 3 wt.% SO42−/Al2O3 should be from the synergetic effect of Brønsted and Lewis acid sites as well as its appropriate amount of surface super-acid sites.

Figure 7
figure 7

NH3-TPD curves of Al2O3 catalysts.

Full size image
Figure 8
figure 8

Pyridine-IR spectra of Al2O3 catalyst.

Full size image

It was found that the catalytic acivities of recycled SO42−/Al2O3 catalysts dropped significantly (Table 3, entry 9) even after re-calcination at 500 °C for 24 h under dry air. The total super acid sites of recycled 3 wt.% SO42−/Al2O3 sample was reduced to 0.448 mmol/g comparing to 0.623 mmol/g of fresh 3 wt.% SO42−/Al2O3, indicating the release of substantial sulfate species from Al2O3 surface in reaction (Figure S1). To recover the performance of SO42−/Al2O3 catalyst, re-deposition of SO42− on spent catalyst (upto 3 wt.%) by incipient wetness method is required (Table 3, entry 10).

Treated 2,3-acetonide 3a with methanesulfonyl chloride (MsCl) in pyridine to afford mesylate 3 as yellow crystals (91% yield)38,42. Mesylate 3 then was treated with aq ammonia and worked up44, the 2,3-O-isopropylidene ribonolactam could be isolated crystalline in 84% yield. The carbonyl group on the anomeric carbon of the resulting lactam was reduced with NaBH4 in methanol to give colorless syrup (95% yield)28. Finally, the resulting syrup was hydrolyzed using acidic Amberlite IR-120H resin to afford 1,5-dideoxy-1,5-imino-d-ribitol 4 in 96% yield. On the other pathway, treated mesylate 3 with potassium hydroxide to inverse the reacting carbon center and then followed by reacting with MsCl in pyridine to afford mesylate 5 in 80% yield43. Following the same experiment procedures (treated with aq ammonia, reduction with NaBH4 and hydrolysis using acidic Amberlite IR-120H) described above to obtain 1,5-dideoxy-1,5-imino-l-arabinitol 6. Hence, through above novel and environmental friendly synthetic route, 1,5-dideoxy-1,5-imino-d-ribitol 4 and 1,5-dideoxy-1,5-imino-l-arabinitol 6 was synthesized from d-ribose with an overall yield of 54% and 46%, respectively. The yield for 4 is better than those synthesized from d-ribonolactone (38%) and allylic alcohol (30%)43. The yield for 6 is also significantly higher than those from using d-ribose (5–27%)19. and l-lyxose (8%)44 as starting materials. In addition, this synthetic strategy shows more environmental benefit and utilizes more principles of green chemistry than those conventional routes19,43,44.

The generality of this new synthetic protocol was tested for the synthesis of 1,5-dideoxy-1,5-imino-d-arabinitol 10 from d-lyxose (Scheme 2). d-lyxonolactone could be obtained by oxygenation of the far cheaper d-galactose by the Humphlett oxygenation45. However, the yield is low (68%) and potassium d-lyxonate instead of d-lyxonolactone was obtained, which requries extra work to convert potassium d-lyxonate to d-lyxonolactone. Different from Humphlett oxygenation, the aerobic oxidation of d-lyxose 7 with 1 wt.% Au/Al2O3 gave higher yield of desirable lactone product and > 99% conversion was achieved after 3 h. The selectivity to d-lyonolactone was > 95%.

Table 4 shows the transformation of d-lyxonolactone 8 to 2,3-O-isopropylidene-d-lyxonolactone 9a under different reaction conditions. The yield for 9a was < 5% with the use of CuSO4 or γ-Al2O3 in reaction (Table 4, entries 1 and 2). When the reaction mixture were refluxed over 1 and 3 wt.% SO42−/Al2O3, the isomer 3,5-acetonide 9b was formed. Refluxing the reaction mixture over 9 wt.% SO42−/Al2O3 charred the reactant and products. To achieve single isomer product acetonide 9a, the reaction mixture was stirred for 18 h at room temperature, a 43% yield of 9a could be obtained over 3 wt.% SO42−/Al2O3, which is significantly improved comparing with 27% yield through using methane sulfonic acid as catalyst. Fortunately, the significant difference in solubility of d-lyxonolatone 8 and 2,3-acetonide 9a allows them to be easily separated with liquid–liquid (ethyl acetate/water) extraction. In turn, it was possible to recover the unreacted lyxonolatone and carry out further batch reactions to improve the yield. About 80% yield was achieved after three cycles of the reaction. Compound 9a was next reacted with MsCl in pyridine solution followed by reducing the carbonyl group on the anomeric carbon of the resulting lactam with NaBH4 in methanol, then hydrolyzing over acidic Amberlite IR-120H resin to produce 1,5-dideoxy-1,5-imino-d-arabinitol 1046. The overall yield of 1,5-dideoxy-1,5-imino-d-arabinitol 10 was 48%. The yield to d-arabinitol 10 can be compared with previously reported yields of 40% from d-arabinose20.

Table 4 Acetalization of lyxonolactone with acetone.
figure b
Full size table

Conclusion

The novel strategy for synthesis of the iminosugars 1,5-dideoxy-1,5-imino-ribitol and 1,5-dideoxy-1,5-imino-dl-arabinitol synthesis was tested out in an environment friendly route starting from naturally occurring d-ribose and d-lyxose. Using ultra small gold clusters on Al2O3 (Au/Al2O3) could effectively oxidize d-ribose and d-lyxose to corresponding lactones under alkaline-free condition. SO42−/Al2O3 showed good catalytic activities to replace homogenous Concentrated HCl and MsOH catalysts for acetalization of d-ribonolactone and d-lyxonolactone. The synthetic route resulted in good overall yields of 1,5-dideoxy-1,5-imino-d-ribitol of 54%, 1,5-dideoxy-1,5-imino-d-arabinitol of 48% and 1,5-dideoxy-1,5-imino-l-arabinitol of 46%. In addition, the heterogeneous catalysts and reagents applied in this synthetic strategy show environmental benefit and utilize the principles of green chemistry.

Experimental section

See experiment detail in supplemental information.