Simplified sugar synthesis
Dr Graeme Horne, senior chemist at MNLpharma outlines recent innovations in the synthesis of imino sugars and related carbohydrate molecules, speeding the development of drug leads
Dr Graeme Horne, senior chemist at MNLpharma outlines recent innovations in the synthesis of imino sugars and related carbohydrate molecules, speeding the development of drug leads
Imino sugars are naturally occurring sugar analogues found in plants and microbes, which have potential in a range of blockbuster markets - from metabolic disorders to oncology. To date, two imino sugars have reached the market: Bayer's Miglitol for diabetes, and UCB/Actelion's Zavesca for the genetic disorder known as Gaucher disease, with others in clinical development for a range of diseases including viral infection, metabolic disease and cancer.
The track record for this class of compounds is compelling when one considers that from an initial two or three natural imino sugars, six lead compounds were generated that have resulted in two marketed drugs, three Phase II compounds and one Phase I compound.
However, a number of chemical characteristics of this class of compound have impeded the rate of development and exploitation within the pharmaceutical and biotech industries. These molecules are highly polar and, as such, are not suited to research using the standard reagents and solvents used by medicinal chemists. A lack of chromaphores further complicates research and until recently very few imino sugars had been isolated, restricting diversity within the class.
Additionally, until now, there have been complexities in synthesising these compounds and their analogues. Traditional methods have been slow, complex, have frequently led to the manufacture of flawed compounds or complex mixtures, and have not had realistic prospects of commercial scalability. A key objective for researchers in this area of chemistry has been to simplify key synthetic steps, both as a matter for convenience and to significantly improve the commercial potential of this chemistry class.
MNLpharma has invested considerable time and effort to explore and exploit the untapped potential of these polar compounds. Having isolated relatively large numbers of imino sugars and related compounds from natural sources (over 100), the company recognised the urgent need to improve methods of synthesis if the therapeutic and commercial potential of imino sugars was to be realised. From the beginning of its interest in these molecules, MNLpharma formed an alliance with the research group of Professor George Fleet at the University of Oxford, a recognised world leader in carbohydrate chemistry, who is particularly skilled and experienced in solving synthetic conundrums in the field.
MNLpharma and Professor Fleet established a programme to generate synthetic analogues of imino sugars of interest to the company. MNLpharma had established a particular interest in the immunomodulatory activity of some molecules and the antiviral activity of others. Specifically, the company is developing an orally active immunomodulator to treat of cancer as its lead candidate (MNLP462a), and a synthetic analogue of this molecule (MNLP24) is under investigation as an immunopotentiator (adjuvant) for various immunotherapies and vaccines. Both MNLP462a and MNLP24 are members of a class of imino sugar known as the polyhdroxylated pyrrolizidines. Members of this class include the alexine, australine and casuarine alkaloids (see figure 1). The development of these compounds as therapeutic products will require the establishment of commercially scaleable synthetic routes.
The synthesis of polyhydroxylated pyrrolizidines is technically challenging, principally because of the problems associated with unwanted side reactions arising from the hydroxyl groups (epoxidations, substitutions etc.). The more hydroxyl groups in the structure, the greater the difficulty. This problem has been tackled by protecting or differentiating the reactivity of the individual hydroxyl groups but this often leads to lengthy and complicated syntheses.
Bell et al. describe the synthesis of four diastereoisomers of casuarine from heavily protected eight carbon sugar lactones (Figure 2a). Although this work represents an effective method for obtaining these diastereoisomers it is limited in its flexibility towards the synthesis of other stereoisomers. The use of heavily protected intermediates limits the versatility of such approaches: they cannot be readily adapted to the synthesis of other pyrrolizidine imino sugars and often lead to cumbersome and lengthy routes. This is primarily a consequence of the necessity for selective protection, deprotection and activation. Additionally, access to starting materials of the appropriate configuration may be limited by the availability of carbohydrate sources.
More recently, Izquierdo et al. have reported the synthesis of casuarine and its 6,7-diepi isomer from easily available d-fructose (Figure 2b). The key intermediate in this route is a suitably protected DMDP derivative obtained from d-fructose in nine steps and at a moderate yield. However, despite the ready availability of the starting carbohydrate, the synthesis suffers from a lack of selectivity in the key dihydroxylation step affording a 2:1 diastereomeric ratio of isolated diols.
Another approach to the synthesis of polyhydroxylated imino sugars has been reported by Professor Scott Denmark and co-workers, who synthesised several polyhydroxylated pyrrolizidine and indolizidine alkaloids with up to four contiguous stereocentres through tandem [4+2]/[3+2] nitroalkane cycloadditions. This methodology was later extended by the same workers to the synthesis of casuarine by the intermolecular [3+2] cycloaddition of a suitable substituted dipolarophile and a flexible, heavily substituted nitronate (Figure 2c). Although this work has been successfully applied to the synthesis of casuarine, the method lacks stereocontrol, and so requires the resolution of complex diastereomeric mixtures. Additionally, the synthesis is not readily adaptable to other stereoisomers and is not amenable to scale-up.
There is, therefore, a need for generally applicable, efficient and flexible processes for the synthesis of polyhydroxylated pyrrolizidines. Below we describe a programme of research at the University of Oxford, which led to the discovery of a novel method for the synthesis of imino sugar analogues. This method contains a important key step in the closure of the bicyclic ring of the pyrrolizidine nucleus; a stereospecific double ring closure on unprotected polyhydroxylated systems.
The ability to control the stereoselectivity of the imino sugar ring formation is fundamental to the success of our approach and has been used in the syntheses of numerous pyrrolizidine and indolizidine alkaloids. Moreover, the use of intermediates having free hydroxyl groups provides a hitherto unexploited mechanism for controlling the distribution of the product stereospecificity and its yield, via complex formation at the free hydroxyl groups.
Retrosynthetic analysis of the pyrrolizidine nucleus of MNLP462a (1, Figure 3) provides an overview of the strategy employed by MNLpharma to synthesise these complex polyhydroxylated imino sugars:
Compound 1 can be accessed through a double ring closure brought about by the hydrogenolysis of unprotected azido dimesylate 2. The resulting free amine will then undergo nucleophilic displacement of two activated hydroxyls in the form of methanesulfonyl esters to afford MNLP462a. Compound 2 is obtained from fully protected azido ester 3 through reduction of the ester moiety and activation of the resulting hydroxyl group in the form of a methanesulfonyl ester followed by deprotection. Compound 3 is accessed through Wittig extension of lactol 5 and subsequent dihydroxylation of the resulting unsaturated ester 4. 5 is arrived at via reduction of azido lactone obtained through the selective protection of l-gulonolactone, which frees the hydroxyl at C-2 for introduction of azide following a double inversion strategy
The key to the success of this strategy can be evaluated through the outcome of two vital components of the scheme; the stereoselective cis dihydroxylation of the unsaturated ester 4 and the bicyclic ring closure on the unprotected azido dimesylate 2. It is the selectivity attained in each of these steps that allows the isolation of MNLP462a in an overall yield of 7% over 16 steps - an average of over 80% per step.
olefin dihydroxylation
The dihydroxylation of olefins is a powerful technique in the synthesis of intermediates and building blocks in the formation of pharmaceuticals, agro-chemicals and other fine chemicals. The osmium tetroxide (OsO4) mediated dihydroxylation of olefins is one of the most selective and reliable organic transformations for two reasons: OsO4 reacts with virtually all olefins; and it reacts slowly with other common organic functional groups. This procedure allows for the ready conversion of olefins into the corresponding vicinal diols. However, it is often necessary to employ the dihydroxylation in an isomerically pure form and hence the requirement to control the stereochemical outcome of these reactions has led to the development of a number of asymmetric protocols. Nevertheless, in our initial evaluation of the dihydroxylation of 4 the non-asymmetric Upjohn procedure was followed (Table 1).
Despite affording a respectable yield of diol (63 %) the relative diastereoselectivity was poor (3:2 in favour of required diol 7). A change in oxidant from potassium osmate to ruthenium chloride and sodium periodate was next evaluated. Although the reaction proceeded cleanly and rapidly, virtually no diastereo-selectivity was observed. Therefore, it was decided to apply Sharpless asymmetric dihydroxylation (AD) conditions using the (DHQ)2PHAL ligand. This gave the diol in a reasonable yield of 63% with improved stereoselectivity (7:1) however prolonged treatment led to the accelerated formation of the Michael product (9) obtained by addition of the free hydroxyl at C-6 onto the double bond via a Michael addition. This side reaction was attributed to the relatively basic conditions of the Sharpless AD. Changing conditions to circumvent the Michael addition to the acid catalysed dihydroxylation (citric acid, NMO, potassium osmate) resulted in a comparable yield of isolated diol, no Michael formation but reduced diastereoselectivity (3:1).
Attempts to improve the diastereoselectivity of the acid catalysed dihydroxylation through chiral induction by substituting citric acid for L-tartaric acid yielded no improvement. However, applying a modified Sharpless AD in which the reoxidant is changed from the potassium ferricyanide/potassium carbonate system to N-methylmorpholine afforded excellent diastereoselectivity with no Michael product formation and in satisfactory yield. A change in phtalazine ligand from (DHQ)2PHAL to (DHQ)2AQN improved the selectivity further from 7:1 to 15:1 in favour of desired vicinal diol 7 with no appreciable effects upon the yield.
Whereas previous syntheses of casuarine alkaloids had furnished the bicyclic pyrrolizidine ring system either stepwise or through the use of heavily protected precursors, our route was developed with the intention of forming both rings of the MNLP462a skeleton in one step using an unprotected precursor. Consequently, hydrogenolysis of 2 followed by treatment with sodium acetate yielded MNLP462a 1 in 7% overall yield (>80% per step).
The stereospecificty of the double ring closure is exemplified by the resulting GC-MS trace of crude material (Figure 4). As shown, in spite of the significant potential for side reactions (epoxidations leading to epimers or piperidines, formation of furans or pyrans as a result of intramolecular attack by oxygen nucleophiles or the hydrolysis of the mesyl groups) the cyclisation step proceeds with excellent stereocontrol. It is this generally applicable, efficient and flexible process for the synthesis of polyhydroxylated alkaloids that enables us to access complex structural motifs with relative ease and excellent stereocontrol. Moreover, we have used this novel and innovative bicyclic ring closing approach in the synthesis of numerous epimers and analogues of MNLP462a as well as their indolizidine congeners. Furthermore, this innovation has provided a critical progression in the commercial potential for this class of compounds.
As well as a significant reduction in the number of synthetic steps required, benefits include the production of fewer and less complex mixtures and improved yields of the target product. The company regards the IP resulting from this discovery to be one of the most important components of its patent portfolio, not only for the commercial efficiencies and potential therapeutic benefits that this technology offers but also as an additional means to protect its interests in this important class of compounds.
MNLpharma can now continue to exploit imino sugars to the full and manufacturers of existing imino sugar therapeutics have a new mechanism through which they can endeavour to develop second-generation imino sugar therapeutics with improved side-effect profiles. The thin vein of compounds available in the past has been expanded to a rich seam of both natural compounds and synthetic analogues, and manufacturing innovations such as this serve to improve the chemical potential and scalability of what has been until now a challenging area of chemistry.