Remote C−H Glycosylation by Ruthenium(II) Catalysis: Modular Assembly of meta‐C‐Aryl Glycosides

Abstract

The prevalence of C‐aryl glycosides in biologically active natural products and approved drugs has long motivated the development of efficient strategies for their selective synthesis. Cross‐couplings have been frequently used, but largely relied on palladium catalyst with prefunctionalized substrates, while ruthenium‐catalyzed C‐aryl glycoside preparation has thus far proven elusive. Herein, we disclose a versatile ruthenium(II)‐catalyzed meta‐C−H glycosylation to access meta‐C‐aryl glycosides from readily available glycosyl halide donors. The robustness of the ruthenium catalysis was reflected by mild reaction conditions, outstanding levels of anomeric selectivity and exclusive meta‐site‐selectivity.

A ruthenium(II)/phosphine catalysis system enabled versatile meta‐arene C−H glycosylation by σ‐activation. Thus, easily accessible glycosyl bromide donors furnished 1,2‐trans C‐aryl glycosides via robust ruthenium(II) catalysis under exceedingly mild conditions.

Introduction

C‐aryl glycosides represent an important carbohydrate scaffold in which the glycosidic C−C bond confers a remarkable stability to both enzymatic and chemical hydrolysis. ^[1] As a consequence, C‐aryl glycosides were widely exploited in a variety of pharmacologically relevant drugs, such as Dapagliflozin, Canagliflozin and Ipragliflozin (Scheme 1a). ^[2] For the chemical assembly of C‐aryl glycosides, transition metal‐catalyzed cross couplings with two prefunctionalized substrates, such as the Corriu‐Kumada, ^[3] Suzuki–Miyaura, ^[4] Stille, ^[5] and Negishi ^[6] couplings, were developed. ^[7] In contrast, during the recent years C−H activation has emerged as an increasing viable alternative for the late‐stage functionalization, avoiding the synthesis of the two prefunctionalized agents ^[8] In this context, palladium‐catalyzed ortho‐C−H glycosylation of arenes provided an efficient access to ortho‐C‐aryl glycosides (Scheme 1b). ^[9] Despite of indisputable advances, the selective installation of carbohydrates at a distal position of arenes remains to be in high demand via meta‐C−H functionalization.[ ¹⁰ , ¹¹ , ¹² , ¹³ ] Unlike the proximal C−H glycosylation of arenes, ^[14] the distal C−H glycosylation is significantly more challenging and limited methods are available. Combined with the role of a palladium(II) catalyst as Lewis acid promoter for glycosyl chloride donor activation, ^[15] a Catellani‐type reaction was recently designed to achieve meta‐C−H glycosylation. ^[16] Given the unique power of ruthenium catalysis for meta‐C−H functionalization, ^[17] we wondered whether a ruthenium‐catalyzed C−H glycosylation could be amenable for the particularly challenging construction of meta‐C‐aryl glycosides (Scheme 1c). ^[18] As a result, we herein disclose our findings on meta‐C‐aryl glycoside synthesis with the salient features comprising (a) unprecedented ruthenium‐catalyzed σ‐activation for meta‐C−H glycosylation of arenes, (b) high levels of site‐, chemo‐ and stereoselectivities, and (c) exceedingly mild reaction conditions applicable for late‐stage functionalization of drug scaffolds.

Scheme 1.

Selected C‐aryl glycosides and methods for C‐aryl glycoside synthesis.

Results and Discussion

We initiated our studies for the meta‐C(sp²)−H glycosylation with mannosyl bromide donor 2 a as the glycosylation reagent (Table 1). The reaction with [RuCl₂(p‐cymene)]₂ as the catalyst and MesCO₂H as the additive failed to deliver the desired product 3 (entry 2). Instead, P(4‐CF₃−C₆H₄)₃ used as ligand provided meta‐glycosylation product 3 in 29 % yield at 100 °C (entry 3). ^[12g] Decreasing the reaction temperature improved the catalytic efficiency, with 60 °C being the best choice to give the product 3 in 75 % yield with exclusive α‐selectivity (entries 1–3). Next, [RuCl₂(PPh₃)₃] as catalyst in the absence of P(4‐CF₃−C₆H₄)₃ was tested, delivering product in 37 % yield (entry 4). ^[19] When replacing the P(4‐CF₃−C₆H₄)₃ with different phosphine ligands, the yield could not be improved (entries 5–7). An optimization of the base demonstrate that K₂CO₃ was the best of choice (entries 8). A set of typical solvents, such as NMP, toluene and THF, was probed, but with limited success (entry 9). Control experiments verified the essential roles of the ruthenium catalyst and the phosphine ligand (entry 10 and 11).

Table 1.

Optimization of ruthenium‐catalyzed meta‐C−H glycosylation.^[a]

Entry	Deviation from the standard conditions	Yield[%][b]
1	none	75
2	MesCO2H as ligand	NR
3	At 100 °C, 80 °C, 40 °C	29/57/12
4	[RuCl2(PPh3)3] as catalyst	37[c]
5	P(4‐OMe−C6H4)3 as ligand	40
6	P(4‐F−C6H4)3 as ligand	59
7	P(3,5‐CF3−C6H3)3 as ligand	NR
8	Na2CO3/K3PO4/KOAc as bases	28/49/NR
9	NMP/toluene/THF as solvents	NR/NR/27
10	Without [RuCl2(p‐cymene)]2	NR
11	Without P(4‐CF3−C6H4)3	NR

[a] Reaction conditions: 1 a (0.10 mmol), 2 a (0.2 mmol), catalyst (5.0 mol %), ligand (10 mol %), base (2.0 equiv), solvent (1.0 mL). [b] Yield of isolated product. [c] Without P(4‐CF₃−C₆H₄)₃. NMP: N‐methyl‐2‐pyrrolidon.

With the optimized reaction conditions for the meta‐C(sp²)−H glycosylation in hand, we examined its generality (Scheme 2). ^[19] Initially, the substitution pattern on the arene moiety was tested, and para‐decorated arenes 1 b and 1 c were well tolerated (4 and 5). Electron‐rich pyridine 1 d exhibited a lower efficiency (6). When methyl substituent was installed at the meta‐position, no desired product was observed. ^[19] Then, pyrimidine derivatives 1 e–1 g were used in the meta‐C−H glycosylation and high catalytic efficiencies were observed (7–9). The electrophilic chloro‐group at the para‐position of phenyl 1 h also proved to be feasible (10), without any ortho‐arylation observed. ^[20] The meta‐C−H glycosylation was not restricted to pyridine‐guided functionalization. Indeed, a plethora of heterocycles, such as pyrazole 1 i, purine derivatives 1 j–1 l and quinoline 1 m, was identified as amenable substrates for the challenging meta‐C‐aryl glycosides assembly (11–15). In addition, fluorescent scaffolds, such as benzo[h]quinoline 1 n and benzo[c]phenanthridine 1 o afforded products 16 and 17 irrespectively in a remote C−H glycosylation manner.

Scheme 2.

Ruthenium‐catalyzed meta‐C−H glycosylation of heteroarenes.

Subsequently, the ruthenium‐catalyzed meta‐C−H glycosylation strategy was probed with different glycosyl bromides 2 (Scheme 3). Rhamnosyl bromide 2 b proved efficient to site‐ and stereo‐selectively stitch rhamnose moiety into the meta‐position of a series of heteroarenes (18–20). Diversely protected mannosyl bromides 2 c–2 e, containing acetyl and pivaloyl group, generated 21–26 with exclusive α‐anomeric selectivity.

Scheme 3.

Ruthenium‐catalyzed meta‐C−H glycosylation with different glycosyl bromide donors 2.

To gain insights into the reaction mechanism, we conducted mechanistic experiments (Scheme 4). The involvement of radical intermediates was supported by the detection of the glycosyl radical‐TEMPO adduct via high resolution mass‐spectrometry (Scheme 4a). The mannosyl radical was further substantiated by a ruthenium‐catalyzed radical relay experiment, with three‐component product 47 formed in 80 % yield as well as 10 % of direct meta‐C−H glycosylation product 7 (Scheme 4b). Based on these observations, we attempted our ruthenium catalysis with catalytic amount of phenyl pyridine in the Giese addition, but mannosyl radical conjugate addition products were not detected. ^[19] Noteworthily, there is no 1,2 acyloxyl migration process ^[21] observed in the meta‐C−H glycosylation reaction due to the difficulty to form the rigid 1,3‐dioxolanyl radical with mannosyl bromide. ^[22] To examine whether there is a neighboring effect of the C2‐benzoyl group, substrate 2 p was utilized under otherwise identical reaction conditions and product 42 was not detected (Scheme 4c). Similarly, 2‐deoxyl glycosyl bromide 2 q featuring no substituent at the C2‐position proved not suitable for the meta‐C−H glycosylation, suggesting that the C2‐carboxyl protecting group might be crucial for an efficient transformation (Scheme 4d). Glucosyl bromide 2 r failed to generate the desired meta‐C−H glycosylation product 44 (Scheme 4e). Compared to the ⁴C₁ and ¹C₄ conformers, the slightly distorted B_2,5 boat conformation of glucopyranosyl radical is more stable (Scheme 4h).[ ¹⁹ , ²³ ] The C2‐benzoyl group and lone pair electrons of the endocyclic oxygen hence may block the attack of a glucosyl radical to the para‐position of the cyclometalated C−Ru bond. ^[12j] In contrast, when acetyl protected galactosyl bromide 2 s was employed, the product 45 was formed with α‐anomeric selectivity in 42 % yield (Scheme 4f). This α‐selectivity may be caused by the C4‐acetyl group, instead of the α‐selectivity control derived from the sterically encumbered catalyst. Interestingly, when conformationally unrestricted benzoyl protected xylosyl bromide 2 t was employed, product 46 was isolated, albeit with poor stereoselectivity (Scheme 4g). We assume that the B_2,5 boat conformer of xylosyl radical is more flexible than its chair conformers. It features a planar C1‐carbon center and allows the attack from either the α‐ or the β‐side (Scheme 4h).[ ¹⁹ , ²⁴ ] The mannosyl radical possessed stable ⁴C₁‐chairlike conformation, which was stabilized by the interaction between the anomeric radical orbital (SOMO), the σ*‐orbital of the adjacent C−O bond, and the p‐orbital of a lone pair of the ring oxygen in their periplanar arrangement. ^[19] The C2 axial benzoyl group and lone pair electrons of ring oxygen force the formation of the meta‐C−H glycosylation product with complete α‐stereoselectivity (Scheme 4h). In addition, the ruthenium catalyst [Ru] was employed for the challenging meta‐C−H transformation (Scheme 4i), the meta‐glycosylation product 3 was obtained in 64 % yield, which indicates that this catalyst could be catalytically relevant.

Scheme 4.

Summary of key mechanistic studies.

Based on our findings, a plausible catalytic cycle (Figure 1) commences with a ortho‐C−H ruthenation to form intermediate A. Subsequently, single electron transfer (SET) from the ruthenium(II) complex to the mannosyl bromide occurs, ^[19] generating ruthenium(III) intermediate B and radical C, followed by addition of the radical C to the para‐position of intermediate B to give intermediate D. The reactive triplet radical D is stabilized by singlet metallacycle E via ligand to metal charge transfer. Finally, proton abstraction and ligand exchange deliver the desired meta‐glycosylation product 3 and regenerate ruthenium(II) complex A.

Figure 1.

Proposed catalytic cycle.

The practical utility of the ruthenium‐catalyzed meta‐C−H glycosylation was illustrated by a gram‐scale synthesis of C‐aryl glycosides 3 and 7 (Scheme 5a). Likewise, a two‐step sequence enabled the efficient transformation of pyridyl group into useful 2‐formylpyrrole 36 (Scheme 5b). Late‐stage diversification of product 3 allowed the construction of fluorescent labelled C‐aryl glycosides 37 and 38 by ruthenium ^[25] and copper ^[26] catalysis (Scheme 5c, d). In addition, to enrich the structural diversity of the products, the selective arylations of the arene scaffolds were featured in the synthesis of biaryl 39 and 40 (Scheme 5e, f). In addition, the versatility of the ruthenium catalysis was mirrored by the one‐pot synthesis of product 41 in 54 % yield with the commercially available substrate 1 p and easily prepared 2 a. It is noteworthy that the synthesis of product 41 through an established conventional cross‐coupling involved multiple synthetic steps and resulted in a much lower overall yield.

Scheme 5.

Late‐stage transformation of C‐aryl glycosides and application. ^[19]

Finally, the robustness of the meta‐C‐aryl glycoside assembly was exploited for the meta‐C−H glycosylation with structurally complex glycosyl bromides (Scheme 6). Hybrid glycosyl donors bearing natural products and drug derivatives, such as indomethacin, bezafibrate, naproxen, fenofibric acid, dehydrochloric acid, ibuprofen, repaglinide, ciprofibrate, and tolmetin, were thereby selectively converted to C‐aryl glycosides 27–35, leading to highly functionalized conjugates with excellent levels of chemo‐ and stereoselectivities.

Scheme 6.

Late‐stage meta‐C−H glycosylation.

Conclusion

In summary, we have developed a ruthenium‐catalysed late‐stage C−H glycosylation to enable a platform for the assembly of biologically important meta‐C‐aryl glycosides. Mild and robust ruthenium catalysis allowed for the expedient meta‐C−H glycosylation with excellent levels of chemo‐, site‐ and stereoselectivities. Our strategy proved efficient and operationally simple while versatile glycosyl bromides were probed for the elucidation of anomeric selectivity. Overall, this meta‐C−H glycosylation strategy well complements the current established ortho‐C−H glycosylation for C‐aryl glycosides synthesis.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

J. Wu, N. Kaplaneris, J. Pöhlmann, T. Michiyuki, B. Yuan, L. Ackermann, Angew. Chem. Int. Ed. 2022, 61, e202208620; Angew. Chem. 2022, 134, e202208620.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.