Regioselective Synthesis of a C-4′′ Carbamate, C-6′′ n-Pr Substituted Cyclitol Analogue of SL0101

Abstract

An asymmetric synthesis of two analogues of SL0101 (1) has been achieved. The effort is aimed at the discovery of inhibitors of the p90 ribosomal S6 kinase (RSK) with improved bioavailability. The route relies upon the use of the Taylor catalyst to regioselectively install C-3″ acetyl or carbamate functionality. This study led to the identification of a third-generation analogue of SL0101 with a C-4″ n-Pr-carbamate and a C-3″ acetate with improved RSK inhibitory activity.

Graphical Abstract

Treatments for metastatic breast cancer are limited to reducing disease progression.¹ In the five-year period after diagnosis, patients with triple negative breast cancer (TNBC) have a reduced survival rate compared to other subtypes.² The MEK-ERK1/2 cascade has been identified as a viable target for TNBC,³ but drugs that inhibit global regulators like MEIK cause side effects that reduce the drug’s efficacy.⁴ To circumvent this issue, we focused our efforts on the MEK downstream effector, RSK, which is active in the majority of TNBC tumors.⁵ RSK, is a Ser/Thr protein kinase, which contains an N-terminal kinase domain (NTKD) responsible for substrate phosphorylation.⁶ The flavonoid glycoside natural product SL0101 (1) has been identified as a selective inhibitor of the NTKD of RSK (K_i ≈ 1 μM).⁷

As part of a larger structure activity medicinal chemistry effort (Scheme 1),^8,9 we have identified SL0101 rhamno-pyranose (2) and carbasugar (3) analogues with C-6″- substitutions that showed improved efficacy in the in vitro kinase assays.^10,11 The carbasugar substitution (2, Y = O to 3, Y = CH₂)¹² was designed to address the short in vivo half-life of the pyranose forms. In a similar vein, we have explored the substitution of the C-3″/C-4″-acetates with nonhydrolyzable isosteres.¹³ This includes the substitution of the C-4″-acetate with a C-4″-acetamide (e.g., 2a to 2b).

Scheme 1.

SL0101 Structure—Activity Relationship

Unfortunately, these efforts to identify amide isosteres only found analogues (e.g., 2b with reduced affinity for RSK compared to 2a). In contrast, the bis-carbamate 4 showed some improvement in activity compared to 3a. With this in mind, we decided to target the regio-isomeric monoacetate/monocarbamates 5 and 6 for synthesis and evaluation as an inhibitor of RSK.¹⁴

Our de novo synthesis of SL0101 (1) and its analogues 2–6 is outlined in Scheme 2. The approach to both the pyranose and carbasugar motifs relies on the construction of the anomeric bond as C-1 via a Pd-catalyzed glycosylation/cyclitolization reaction (i.e., 8 from 9 + 10 or 11). The glycosyl-donor 10 can be made from achiral furan 12 in only three steps whereas the corresponding cyclitol donor 11 requires 9 steps from quinic acid 13. The Pd-catalyzed coupling reaction sequence that prepares both the pyran and cyclohexene in 8 allows for the selective introduction of the C-4 substituent. In contrast, the C-3″ substituent is introduced by means of an Upjohn dihydroxylation to form diol 7 and then a differentiation of the C-3″ equatorial from the C-2″ axial alcohol, which can require more steps than is ideally desired (Scheme 3). Herein we describe our efforts to explore the use of the Taylor catalyst¹⁵ to regioselectively install ester and carbamate functionality on the SL0101 rhamno-pyranose and carbasugar framework and its use in the synthesis of monoacetate/monocarbamates 5 and 6.

Scheme 2.

De Novo Approach to SL0101 and Analogues

Scheme 3.

C-3 Acylation of Pyranose Analogues

This de novo approach to the SL0101 and its analogues is both highly stereoselective and generalizable (Scheme 3). For example, the Pd-glycosylation products 14a/b could be stereoselectively reduced, acylated, and dihydroxylated to form 16a/b in excellent overall yield (54% and 70% for three steps). In contrast to the stereoselectivity, there were issues associated with the regioselective introduction of the C-3″ acetate. Our efforts to directly acylate 16a (Ac₂O/Py/DMAP) gave a 1:1:1 mixture of diacetates (17a/20a) as well as the triacetate. Using the ortho-acetate/hydrolysis protocol on 16a/b, the C-2″ acetate could be selectively installed to form 17a/b, both in near-quantitative yield as a single regioisomer, via intermediate 18. Regardless of the C-6″ substituent, the C-2″ axial acetate could be isomerized to an ~2:1 mixture of C-3″/C-2″ acetates with the more stable C-3″ acetate favored. Fortunately, the regioisomers were separable by silica gel chromatography affording diacetates 20a and 20b (62% and 55%, respectively). While this approach was able to provide enough material for initial testing, our interest in scaling up the synthesis inspired us to explore protocols for the regioselective introduction of the C-3″ acetate directly on diols 16a/b. We first turned to the stochiometric use of Bu₂SnO and Bu₂Sn(OMe)₂, unfortunately these conditions irreproducibly gave a mixture of regio-isomers 16a/b and 20a/b. To our delight, a highly regioselective and reproduceable procedure resulted when we switch to the diphenyl borinic ester catalyst developed by Taylor. Thus, in the presence of 10% Ph₂BO(CH₂)₂NH₂ both 16a and 16b reacted with AcCl to form 20a (74%) and 20b (85%) with excellent regioselectivity (>90%) via intermediate 21. It is worth noting that the same stereoelectronic effect directs the regioselectivity in both the ortho-ester and borinic ester acylation reactions (i.e., greater C- 3″ nucleophilicity in 18 and 21).

Building upon the success of the Taylor catalyst (Ph₂BO(CH₂)₂NH₂) for the regioselective C-3″ acylation of 16a/b, we decided to explore its use in the regioselective acylation and carbamate formation of the cyclitol variants (i.e., diols 23 and 28, vide infra). This effort began with diol 23 (Scheme 4), which can be easily prepared by the Upjohn dihydroxylation (1% OsO₄/NMO) of cyclohexene 22 (70%).^9e Similar to the pyran cases, the direct acylation of the diol gave a 1:1.6 mixture of diacetates 24 and 25 (entry 1) and the DBU isomerization of the mixture gave a 3:1 mixture of C-3″/C-2″ acetates with the more stable C-3″ acetate 24 being favored. In contrast to the pyranose series, the acylation via the ortho-ester formation/hydrolysis condition failed to show any significant preference for the axial acylation, affording an excellent yield of a 1:1.6 mixture of 24 and 25. However, the yield for this ortho-ester formation/hydrolysis remained high (95%), so when used in combination with the DBU isomerization this began a viable two-step method for the formation of 24. The stochiometric use of Bu₂SnO only slightly improved the selectivity for the C- 3 acetyl group (entry 4).

Scheme 4.

Regioselective C-3 Acetylation of Diol 23

The poor selectivity in both the ortho-ester and tin acetal chemistry suggested that there was a reduction in the rigidity of the carbasugar ring system that led to a diminishment in the stereoelectronic effect in the carbasugar variants of 18 and 21 (X = CH₂). Fortunately, this was less the case for the borinic ester catalysis, albeit it did require a great degree of optimization to obtain the desired regioselectivity. Thus, under our optimized conditions (entry 2), using only 10% of the Taylor catalyst, a good yield (72%) of a 5:1 mixture of 24 and 25 could be achieved, reproducibly affording the desired regioisomer in 60% yield. Interestingly, we were not able to find conditions with the more reactive tricyclic borinic acid catalyst that would give as high a regioselectivity (entries 9–10) as the commercially available catalyst.

We next turned to the regioselective installation of an n-propyl carbamate to the C-3″ position of diol 23 (Scheme 5). Exposure of diol 23 to n-propyl-isocyanate and DBU gave a slow reaction. Increasing the reaction temperature to 45 °C gave a 1:1 ratio of carbamates 26 and 27 in 38% yield. Unfortunately, the minor carbamate isomer cannot be isomerized with DBU. When the same reaction was performed in the presence of a catalytic amount of Taylor’s catalyst (10%), we observed improved yields and regioselectivity of the desired regioisomer carbamate 26. Interestingly, when the temperature was lowered the regioselectivity switched to prefer the axial carbamate isomer 27. Switching DBU for tertiary amine bases leads to slightly lower yields and regioselectivities. When the conditions that are outlined in entry 2 were scaled up, the desired isomer was isolated in a 41% yield.

Scheme 5.

Regioselective C-3 Carbamate Formation

Focus was next turned toward the regioselective acylation of diol 28 with the C-4 carbamate (Scheme 6). The carbamate 28 could be easily prepared from the Pd-cyclitolization product 22 by a three-step ester hydrolysis, carbamate formation, and dihydroxylation reaction sequence (55%). Once again, acylation of diol 28 using typical acylation conditions (AcCl/ Base) led to a mixture of regioisomers. Gratifyingly, we saw only minimal effects from the carbamate substitution at C-4. Thus, application of our optimized borinic ester catalyzed acylation conditions on diol 28 gave a nearly identical yield (64%) and ratio (4.4:1) of regioisomeric monoacetates 29 and 30. As with 25 the undesired regioisomer 30 could be recycled by DBU isomerization to a 1:2.5 mixture of 29 to 30, with the more stable isomer being the major product.

Scheme 6.

Regioselective C-3 Acetylation of Diol 28

We next looked at the regioselective C-3″ carbamate installation to form bis-carbamate 4 from diol 28 (Scheme 7). Exposure of diol 28 to n-propyl-isocyanate and DBU at 45 °C gave a 1:1 ratio of carbamates 31 and 32 in 38% yield. Similarly, the minor carbamate isomer could not be isomerized with DBU. The same reaction was performed in the presence of a catalytic amount Taylor catalyst (10%) which improved the yield and regioselectivity (3:1) for the desired regioisomer carbamate 31. Thus, using entry 2, the optimized procedure, we scaled up, affording 31 in a 34% yield. Once again, when the temperature was lowered the regioselectivity switched to prefer the axial carbamate isomer 32. Surprisingly, in this case (entry 4) the undesired C-3 regioisomer was the major isomer, albeit at low conversion (17% yield). Because 32 could not be isomerized back to the desired regioisomer 31, we did not further pursue lower temperature conditions.

Scheme 7.

Regioselective C-3 Carbamate Formation

Finally, with the required carbamate regioisomers 26 and 29 in hand, we explored their conversion into the desired target compounds 5 and 6 (Scheme 8). This was accomplished by global deprotection with exhaustive hydrogenolysis. Thus, exposure of the C-3″ carbamate 26 to 1 atm of H₂ in the presence of 10% Pd/C gave the C-3″ carbamate 5 in excellent yield (71%). Similarly, exposure of the C-4″ carbamate 29 to 1 atm of H₂ in the presence of 10% Pd/C gave the C-4″ carbamate 5 in excellent yield (70%).

Scheme 8.

Preparation of SL0101 Analogues 5 and 6

The bis-carbamate and two new monocarbamate containing cyclitol SL0101 analogues 5 and 6 were evaluated as RSK2 inhibitors in an in vitro kinase assay (Table 1). In this assay, the analogue 5 with a C-3″ n-Pr-carbamate group and a C-4″ acetate had a 3-fold decreased IC₅₀ relative to SL0101. In contrast, the analogues 4 and 6 with a C-4″ n-Pr-carbamate group had a higher affinity than SL0101, with 4 having a 4-fold and 6 having a 3-fold higher affinity. The potency difference between 2a and analogues C-4″ n-Pr-carbamate 6 and the C-3″/C4″ bis-n-Pr-carbamate 4 group is moderate. Importantly, these results are the first demonstration that modifications to the rhamnose at the 4″ position can be introduced without substantially reducing RSK affinity for the compound.

Table 1.

RSK Inhibitory Activity of Carbamates 4–6

analogue	RSK2 inhibition IC50 (μM)
4	0.14	±0.06
5	1.16	±0.24
6	0.17	±0.05

^aRSK2 IC₅₀ Assay: concentration needed for 50% RSK2 inhibition (n > 2; sextuplicate; average ±95% confidence interval). The in vitro kinase assays provides a relative IC₅₀ as the absolute value depends on batch-to-batch variation of the reagents. SL0101 is used as a positive control for every assay and therefore, the potencies provided in Scheme 1 are relative to SL0101.

In conclusion the use of Taylor’s borinic ester catalyst for the regioselective introduction of acetyl and carbamate functional groups on rhamno-5a-carbasugar motifs was investigated. The Taylor borinic ester catalyst was found to be superior at controlling regioselectivity in the rhamno-carbasugar environment compared to traditional tin and orthoester methods. Conditions were found that allowed for the efficient synthesis of C-3/C-4 regioisomeric acetate/n-Pr-carbamate analogues of the RSK1/2 inhibitor, SL0101. This synthetic effort led to the discovery of a new cyclitol analogue of SL0101 with improved RSK inhibitory activity.