Metathesis access to monocyclic iminocyclitol-based therapeutic agents

By focusing on recent developments on natural and non-natural azasugars (iminocyclitols), this review bolsters the case for the role of olefin metathesis reactions (RCM, CM) as key transformations in the multistep syntheses of pyrrolidine-, piperidine- and azepane-based iminocyclitols, as important therapeutic agents against a range of common diseases and as tools for studying metabolic disorders. Considerable improvements brought about by introduction of one or more metathesis steps are outlined, with emphasis on the exquisite steric control and atom-economical outcome of the overall process. The comparative performance of several established metathesis catalysts is also highlighted.

The broad biological activity of iminocyclitols has attracted growing interest in the synthesis of naturally occurring iminocyclitols and in their structural modification. Consequently, efficient and stereoselective synthetic routes have been developed, often starting from an inexpensive chiral-pool of precursors, in particular carbohydrates that share structural features with iminocyclitols. The main hurdles in this approach are the singling out of only one of the hydroxy groups in the open carbohydrate-derived intermediate, converting this hydroxy group into an amino group, and intramolecularly closing this intermediate [8,[34][35][36]. Because of the high density of functional groups, proper protection throughout the overall synthesis scheme is an important feature that must be considered carefully, with full deprotection occurring in the final step.
With the advent of well-defined Mo-and Ru-alkylidene metathesis catalysts (e.g., 1-10; Scheme 1) [37][38][39][40][41][42][43][44][45][46][47] the RCM strategy was immediately recognized as central to success in the flexible construction of N-heterocyclic compounds, including azasugars. Moreover, the metathesis approach to azasugars has greatly benefited from the vast synthetic experience acquired in RCM preparation of a host of heterocycles. Any RCM-based protocol to iminocyclitols implies three crucial stages: (i) discovery of a route to obtain stereoselectively, starting from an ordinary substrate, the N-containing prerequisite diene precursor; (ii) RCM cyclization of this diene, with an active catalyst, to access the core cyclic olefin; and (iii) dihydroxylation of the endocyclic double bond in a highly diastereoselective manner to form the target product.
In comparison to the traditional, lengthier syntheses of iminocyclitols, the metathesis approach has emerged as a highly advantageous method in terms of atom economy. However, before carrying out the RCM reaction, the basic amino group (incompatible with most metathesis catalysts because of chelation to the metal center) [48] must either be protected (as N-Boc, N-Cbz, etc.), masked by incorporation into a cleavable heteroatom-containing cycle (oxazolidine, cyclic ketal, etc.), or deactivated by conversion into amide or carbamate functions. Due to these protective groups even metathesis catalysts sensitive to functionalities can act efficiently under reaction conditions where an adequate balance between activity/stability factors has been met. In addition, the reaction conditions (temperature, solvents) currently employed in olefin metathesis reactions can be productively transferred to the metathesis steps of iminocyclitols synthesis.
By surveying the field of recent azasugar developments, this review focuses on metathesis reactions (mainly RCM, CM) as essential transformations in the multistep synthesis of mono-  cyclic iminocyclitols, while also discussing the successes and failures in effecting the above mentioned three critical stages.
New perspectives may open up for practitioners of both glycoand metathesis chemistry involved in the synthesis and development of iminocyclitols.
Biological activity of this family of iminocyclitols is dictated by the stereochemistry at all carbon atoms of the pyrrolidine ring system which can adopt either a manno or a galacto conformation, therefore inhibiting either α-mannosidases (e.g., 11-13) or α-galactosidases (e.g., 14) (Scheme 2). A characteristic feature in 11-14 is the presence of a 1,2-dihydroxyethyl side chain.
A further contribution to new pyrrolidine-based azasugars, characteristically having 1,2-dihydroxyethyl side chains and a quaternary C-atom possessing a hydroxy and a hydroxymethyl group, was made by Vankar et al. [59] (Scheme 7). By ingeniously combining a Baylis-Hillman addition with RCM as the key steps, they obtained, stereoselectively and in high yields, 1,4-dideoxy-1,4-iminohexitols 40 and 44 which showed moderate inhibition of β-galactosidase, and α-galacto-and α-mannosidases, respectively. It should be noted that diene 38 did not cyclize in the presence of 1st-generation Grubbs catalyst, even in refluxing toluene, whereas 2nd-generation Grubbs catalyst afforded (in toluene, at 60 °C) the cyclic products 39 and 43 in 89% and 86% yields, respectively. Interestingly, Upjohn dihydroxylation of 39 or 43 (OsO 4 , NMO, acetone/ H 2 O/t-BuOH; HCl, MeOH; Ac 2 O, Et 3 N, DMAP) gave only one diastereomeric diol, because the bulky acetonide group blocks the β-face of the trisubstituted double bond of the pyrrolidine ring and is thus responsible for the high diastereoselectivity.
The starting point in the synthesis of (+)-broussonetine G, 53, was the same annulated oxazolone 48 which, after conversion into the Weinreb amide 51, was coupled with the alkyl bromide substituted spiro compound 52 (Scheme 9).
In fact, the case of broussonetines is much more complicated. This subgroup is currently represented by 30 reported examples, all isolated from plant species and used in folk medicine in China and Japan. Most broussonetines display marked inhibitory activities on various glycosidase types, with selectivities differing from that of other standard iminosugars such as DNJ. In the majority of the broussonetines (54, Scheme 10), a Scheme 9: Synthesis of (+)-broussonetine G (53).
Scheme 11: Synthesis of broussonetines by cross-metathesis. common polyhydroxylated pyrrolidine building block (possibly prepared via protocols including RCM) is bound to a side chain fragment of 13 C-atoms, diversely functionalized. For the introduction of the appropriate side chain, cross-metathesis appeared to be the most versatile method, permitting access to many members of this family, both naturally occurring and analogues. Two types of metathesis processes, RCM and CM, can be thus advantageously intertwined in the synthesis of broussonetines.
For instance, the syntheses of broussonetines C, D, M, O and P were completed by Falomir, Marco et al. [62,63] in a convergent, stereocontrolled way starting from commercial D-serine (55) as the chiral precursor and by applying the critical step of cross-metathesis (the first-ever synthesis of broussonetines O and P) (Scheme 11).
The cross-metathesis reaction was promoted by the 2nd-generation Grubbs catalyst (5, in CH 2 Cl 2 , by heating under reflux in a N 2 atmosphere for 24 h or by heating for 1 h at 100 °C under microwave irradiation). As expected in a cross-metathesis process, a mixture of three products (CM product plus the two homo-metathesis products, all in both stereoisomeric forms) was obtained. Homo-metathesis products from either 56 or the alkene were recycled in the cross-metathesis stage to provide an additional amount of the useful product 57, thus enhancing the overall yield.

Piperidine-based iminocyclitols
During the last decade, polyhydroxylated piperidines have been the target of much cutting-edge synthesis work [8]. Such com- pounds are of special interest as therapeutic agents and as tools for the study of cellular mechanisms and metabolic diseases. From this class, nojirimycin (NJ, trivial name for 5-amino-5deoxy-D-glucopyranose) (59), the first alkaloid discovered that mimicks a sugar (originally isolated from Streptomyces filtrate but also found in other bacterial cultures and plant sources), is a potent glycosidase inhibitor. In aqueous solution nojirimycin exists in both the α-and β-forms, each of which is responsible for inhibition of α-or β-glucosidase, respectively. Similar to its other congeners, mannonojirimycin (60; MJ or nojirimycin B) and galactonojirimycin (61; GJ or galactostatin), nojirimycin is unstable because hemiacetal structures can be adopted [8]   An improvement in the selectivity and efficiency of the total synthesis of (+)-1-deoxynojirimycin (62) (24% overall yield) was made by Poisson et al. [65], who developed a one-pot tandem protocol involving enol ether RCM/hydroboration/oxidation, which gave the best results when the Hoveyda-Grubbs catalyst 6 was used in the RCM (Scheme 15).
Interestingly, in this case the asymmetric synthesis of the protected RCM precursor 78 started from a non-chiral source, the alcohol 75, and proceeded with complete stereocontrol over the 11 steps involved. All attempts to achieve metathesis on another diene precursor having an endocyclic N-atom (the result of N-alkylation of 77 with 3-iodo-2-(methoxymethyloxy)prop-1-ene) led to either recovery of the starting material or olefin isomerization, even in the presence of a number of ruthenium hydride traps. Satisfactory results in RCM were, however, obtained from 78: in the presence of the 2nd-generation Grubbs catalyst 5 and benzoquinone in refluxing toluene, 78 was converted into the cyclized enol ether 79 in 70% yield, while with the Hoveyda-Grubbs catalyst (6, 10 mol %; benzoquinone 10 mol %; in refluxing toluene) 79 was obtained in 85% yield. The three reaction steps leading from 78 to 80, i.e., RCM/ hydroboration/oxidation, could be accomplished in one-pot to afford the product as a single isomer (all-trans triol). The prepared (+)-1-deoxynojirimycin (62) displayed spectroscopic data which perfectly matched those of the natural product.
A similar methodology was used by Han [70] to prepare 5-des(hydroxymethyl)-1-deoxynojirimycin (114) and its mannose analogue 111 (as HCl salts) in a highly stereoselective mode starting from a different common olefin, 107 (Scheme 20). In this case, RCM was promoted by the 2ndgeneration Grubbs catalyst 5 which ensured a high yield of the ring closure (89%) under milder conditions (CH 2 Cl 2 ): all attempts to employ the 1st-generation Grubbs catalyst 2 in RCM failed, supposedly because of an unfavourable steric environment during generation of the Ru-carbene species from 109, as compared to 98 (distinct N-protective groups). Cyclic sulfate chemistry was again invoked for effectively performing the synthesis of 114.
Introducing a genereal strategy for synthesis of deoxyazasugars based on cheap D-glucose, Ghosh et al. also laid groundwork for the preparation of D-1-deoxygulonojirimycin (91) (previously communicated by Takahata [67]; Scheme 17) and L-1deoxyallonojirimycin (122) (Scheme 21) starting from protected diacetone glucose 115 [71]. Different pathways were devised for 91 and 122 via the epimeric RCM precursors 117 and 120, respectively. High yielding cyclization of these dienes, in the presence of the 1st-generation Grubbs catalyst 2 (10 mol %, in CH 2 Cl 2 , under argon, 24 h at 50 °C), led to 118 and 121 with preserved configurations at the stereogenic centre, which therefore allowed the desired stereochemistry in the isomeric final products 91 and 122.
Adenophorine (α-1-deoxy-1-C-methylhomonojirimycin) is a further important iminocyclitol in whose synthesis RCM proved helpful. (+)-Adenophorine (135), a naturally occurring iminocy-clitol with a lipophilic substituent at the anomeric position, is active on α-glucosidase which is a valid proof that α-alkylation at C1 does not supress the glycosidase inhibitory effect. Its lack of activity on β-galactosidase once again indicates that the relative position of hydroxy substituents is critical for selectivity. In the seminal work by Lebreton and coworkers [73], the first asymmetric total synthesis of (+)-adenophorine was achieved in 14 steps (3.5% overall yield, Scheme 23), starting from the Garner's aldehyde 69. RCM is essential for construction of the 6-membered N-heterocycle in 133. Protection of the amino alcohols trans-132 and cis-132, as the corresponding trans and cis oxazolidinones, afforded a mixture of diastereomers that were not separable on silica gel. After effecting RCM (2ndgeneration Grubbs catalyst 5, 5 mol %) on this mixture, separation of the diastereomers by flash chromatography was possible, affording the pure tetrahydropyridine derivative trans-133 in 74% yield. Successive epoxidations on enantiopure trans-133 and then 134, followed each time by regioselective epoxide opening (with a selenium-boron complex and water, respectively), gave finally 135 with good stereoselectivity. This overall synthesis demonstrates rigorous control at every stage of both the steric configuration of the starting materials and the steric effects induced by substituents attached to the piperidine moiety.

Azepane-based iminocyclitols
Iminocyclitols incorporating the azepane ring system are more flexible than the parent pyrrolidine and piperidine iminosugars, and they adopt quasi-flattened, low-energy conformations which can potentially lead to a more favourable binding with the active site of enzymes. The unusual spatial distribution of the hydroxy groups in these compounds should generate new inhibitory profiles. According to in vitro assays, seven- membered ring iminocyclitols are noted inhibitors of α-mannosidase, an enzyme that plays important roles in glycoprotein biosynthesis. Derivatives of this class bearing hydroxymethyl groups at C-6 have been shown to inhibit powerfully lysosomal α-mannosidase while displaying varying potencies toward α-1,6-mannosidase. On the other hand, N-alkylated polyhydroxylated azepanes with the D-glucose or L-idose configuration proved to be potent β-glucosidase inhibitors that showed only weak activity towards α-glucosidase and α-mannosidase [78][79][80]. Malto-oligosaccharides and analogues of di-and trisaccharides containing polyhydroxylated azepane moieties are glucosidase or HIV/FIV-protease blockers, or both.
As for the previous classes, in the synthesis of seven-membered iminocyclitols RCM provides a focal point in ring closure being responsible for constructing the azepane framework. For example, 1,6-dideoxy-1,6-iminoheptitols 148 and 149, that can be viewed as higher homologues of fagomine and nojirimycin, respectively, are easily accessed from the protected diene 146. RCM of this diene with 1st-generation Grubbs catalyst (2, CH 2 Cl 2 , 45 °C) gives the common N-heterocyclic intermediate 147 (91% yield, Scheme 25). Hydrogenation of the latter gives the iminocyclitol 148 whereas its cisselective dihydroxylation affords the pentahydroxy derivative 149.

Scheme 27: Representative azepane-based iminocyclitols.
Starting from L-serine 150, Lin et al. [81] devised a refined method for the synthesis of structurally diverse stereoisomers of polyhydroxyazepanes. In their complex strategy, RCM (1stgeneration Grubbs catalyst, 10 mol %, CH 2 Cl 2 , reflux, 12 h) plays a significant role by leading to a panel of oxazolidinyl azacyclic products (e. g., 152 and 154). Remarkably, the authors expertly arranged the positions of the double bonds involved in RCM on the one hand by addition of alkenyl nucleophiles (with different lengths) on aldehyde intermediates, and on the other hand by placing the second double bond at a different distance relative to the nitrogen atom (Scheme 26). There are two advantageous follow-ups: (i) a desired location of the double bond in the azacyclic RCM product, and therefore of the hydroxyls in the final iminocyclitol products, and (ii) possible extension of the methodology to the construction of other ring sizes (5-to 8-membered). This versatile approach, featuring the basic sequence metathesis/dihydroxylation, led in good yields to a number of stereoisomers of seven-membered iminocyclitols exhibiting glycosidase inhibitory properties (Scheme 27). Of the compounds shown in Scheme 27, compound 161 with L-configuration at C-6 exhibited the highest inhibition.

Conclusion
The paper introduces the broad scope of olefin metathesis as a key reaction in synthetic strategies for the preparation of monocyclic iminocyclitols. In comparison with earlier well-established protocols, olefin metathesis (RCM, CM) offers shorter, simpler and atom-economical routes, and preserving at the same time the carefully designed and worked for stereochemistry of the precursors. Whereas RCM is the method of choice for constructing the pyrrolidine, piperidine or azepane cores of monocyclic iminocyclitols, CM rewardingly permits access to a collection of new iminocyclitols simply by using one heterocyclic intermediate endowed with an olefinic side-chain and changing only its olefin partner. The reaction conditions applied in these crucial steps are rather conventional for metathesis processes, with the choice of the temperature and solvent (refluxing CH 2 Cl 2 or toluene) being dictated by steric demands, and hence energetics, for ring-closing or cross-coupling. While the 1st-and 2nd-generation Grubbs catalysts (5-10 mol %) are the catalysts most frequently employed, the 2nd-generation Grubbs and Hoveyda-Grubbs catalysts perform better when harsher conditions are required. Despite the various functionalities existing on the metathesis precursors and products, sensitive metathesis catalysts are quite productive due to inventive protection/deprotection at the O-and N-heteroatoms. Such delicate operations are skillfully conceived so as to either maintain or reverse the geometry at stereogenic centres, as required. In the ensemble of stereocontrolled reactions concentrating on the economical achievement of the targeted number and relative positions of hydroxy, hydroxyalkyl or other substituents, i.e., the overall structure that hinges on the biological activity, metathesis is surely a fine addition which is bound to succeed in creating novel azasugars with a larger therapeutic window. The metathesis approach may ultimately yield benefits for patients suffering from metabolic disorders, cancer and viral diseases.