Total Synthesis of Phenanthroindolizidines Using Strained Azacyclic Alkynes

Abstract

We report a concise approach to phenanthroindolizidine alkaloids, wherein strained azacyclic alkynes are intercepted in Pd-catalyzed annulations. Two types of strained intermediates were evaluated: a functionalized piperidyne and a new strained intermediate, an indolizidyne. We show that each can be employed, ultimately allowing access to three natural products: tylophorine, tylocrebine, and isotylocrebine. These efforts demonstrate the successful merger of strained azacyclic alkyne chemistry with transition-metal catalysis for the construction of complex heterocycles.

Graphical Abstract

Nitrogen-containing heterocycles are commonly present in many important molecules, including natural products,¹ medicinal agents,^2,3 agrochemicals,⁴ materials,⁵ and polymers.⁶ As such, new strategies for their syntheses remain highly valuable. One promising tactic involves the use of heterocyclic arynes, such as 1–3 (Figure 1). These short-lived intermediates can be manipulated in ways reminiscent of aryne functionalization processes^7–10 but come with the added benefit of having a heteroatom incorporated in their core.^11,12 Pyridynes (e.g., 1) have been most thoroughly studied and have been used in a number of total syntheses.^7,8 For example, total syntheses of ellipticine (6) were enabled by Diels–Alder reactions with in situ-generated pyridynes.^13,14 Indolynes (e.g., 2) and their π-extended carbazolyne counterparts (e.g., 3) are also highly valuable synthetic building blocks, with several applications in total synthesis being reported.^15–25 For example, the interception of an indolyne and carbazolyne, respectively, was instrumental in enabling the total syntheses of N-methylwelwitindolinone C isothiocyanate (7)¹⁷ and tubingensin B (8).²⁴

Figure 1.

Well-studied strained azacyclic alkynes 1–3, azacyclic alkynes 4 and 5, and examples of natural products synthesized by using N-containing hetarynes.

Nonaromatic heterocyclic alkynes are far less studied compared to their aromatic counterparts. Two methodology studies on piperidynes (e.g., 4, Figure 1) are available^26,27 that demonstrate cycloadditions and nucleophilic trappings. These, as well as other scattered examples,^28–32 show that piperidynes can be used to access functionalized, sp³-rich heterocycles. However, piperidynes have never been utilized in total synthesis or metal-catalyzed annulations. Another interesting aspect of piperidyne chemistry is that the N atom bears an electron-withdrawing group (e.g., Ts and Cbz) in known trapping experiments as a means to mitigate the basicity and nucleophilicity of the piperidine nitrogen. As such, the indolizidyne intermediate 5, bearing a basic tertiary amine, has remained unknown. This was of particular interest to us given the presence of indolizidines in natural product scaffolds, including in the phenanthroindolizidine alkaloids.³³

The most well-known family member of the phenanthroindolizidine alkaloids is tylophorine (9, Scheme 1), which was first isolated from Tylophora indica in 1935.³⁴ Tylophorine (9) and many of its derivatives have garnered significant attention due to their important biological activities, including anticancer,^35,36 anti-inflammatory,³⁷ and antiviral properties,³⁸ which include significant activity against SARS-CoV-2.³⁹ Select examples of prior syntheses include the works of Georg,⁴⁰ Opatz,^41,42 Zhao,⁴³ and Chemler.⁴⁴ In considering alternative strategic disconnections of tylophorine (9), we wondered whether strained azacyclic alkyne chemistry could offer a convergent approach.

Scheme 1.

Strategies for the Synthesis of Tylophorine (9) using Strained Azacyclic Alkynes 11 or 5 in Pd-Catalyzed Annulations

Our retrosynthetic strategy for accessing tylophorine (9) is shown in Scheme 1. Disconnection of the B ring at two sites provides fragment 10 for the western portion of the molecule as well as strained intermediates 11 or 5 for the eastern portion. In the forward sense, the fragments would be coupled via a palladium-catalyzed Larock annulation.⁴⁵ Metal-mediated transformations of strained alkynes are known^46–48 and provide powerful tools for constructing multiple bonds in efficient transformations. However, examples involving strained, azacyclic alkynes are less common, and metal-catalyzed reactions of piperidynes remain unknown.

Herein, we disclose access to phenanthroindolizidines using the aforementioned Pd-mediated coupling reaction of strained azacyclic alkynes. We demonstrate that intermediates of either type 11 or 5 (see Scheme 1) can be employed in the key step. For the latter case, we prepared the first indolizidyne precursor 5 and harnessed it to synthesize three natural products, including (±)-tylophorine (9). This study illustrates the first use of a piperidyne or indolizidyne in natural product synthesis and should prompt further synthetic studies of these nontraditional intermediates.

In our first approach to tylophorine (9), we sought to utilize a piperidyne intermediate 11 (see Scheme 1), which would bear an electron-withdrawing substituent on the nitrogen (R¹), akin to known piperidynes.²⁷ This intermediate would possess functionality at R², which could allow for the late-stage introduction of the E ring. We ultimately developed the concise route shown in Scheme 2. Silyl methoxy pyridine 12, prepared in one step from commercially available 4-methoxypyridine,⁴⁹ was treated sequentially with CbzCl, Grignard reagent 13, and aqueous HCl. This delivered vinylogous amide 14 in 99% yield via nucleophilic addition into an in situ-generated pyridinium salt at C13a and hydrolysis.⁵⁰ Subsequent 1,4-reduction of 14 was achieved using L-selectride, and trapping with Tf₂O gave piperidyne precursor 15.²⁷ Notably, our strategy to elaborate 12 to 15 is scalable and should be amenable to the synthesis of other substituted piperidyne precursors.

Scheme 2.

Concise Total Synthesis of Tylophorine (9) Using Piperidyne 16

With 15 in hand, we directed our attention to the key Pd-catalyzed annulation with known biaryl bromide 10 (see Scheme 1). In previous studies, we have found that metal-catalyzed trappings of heterocyclic strained intermediates require careful optimization.⁴⁸ This is due to the inherent requirement that the transient, strained intermediate intercepts a species only present in catalytic quantities. Indeed, the coupling proved to be challenging using conditions developed for benzyne annulations. Ultimately, we found that the reaction between piperidyne precursor 15 and biaryl bromide 10 proceeded most effectively when using 10 mol % Pd(dba)₂ and P(o-tolyl)₃, 1.2 equiv of cesium carbonate, 10 equiv of cesium fluoride, and a 1:20 ratio⁵¹ of acetonitrile to toluene at 110 °C. After the coupling was complete, quenching with aqueous acid at 23 °C facilitated removal of the acetal protecting group. Annulated product 17 bearing a pendant aldehyde was obtained in a 60% yield. To complete the total synthesis of tylophorine (9), 17 was treated with H₂ and Pd/C in methanol. This allowed for hydrogenolysis, followed by reductive amination, to deliver (±)-tylophorine (9) in 90% yield. The route is four steps from 12 and establishes the viability of using piperidynes in total synthesis.

After establishing that a strained, azacyclic alkyne could be used in a metal-mediated Larock annulation to access the phenanthroindolizidine core, we sought to synthesize an indolizidyne precursor. Such an intermediate could offer a unique approach to phenanthroindolizidines that relies on strained intermediate trapping as the final step. Although many of our attempts proved unsuccessful, the route described in Scheme 3 showed some initial promise. The strategy parallels the general approach shown in Scheme 2 but with provisions to introduce the necessary five-membered ring. Treatment of pyridine 12 with allyl bromide gave pyridinium salt 18, which was directly subjected to vinylmagnesium bromide. Subsequent hydrolysis afforded product 19 bearing allyl and vinyl substituents on N and C13a, respectively. With the hope of accessing silyl triflate 20, 19 was treated with L-selectride, followed by Comins’ reagent. However, instead of obtaining desired silyl triflate 20, fragmentation product 22 was isolated in 71% yield. It is hypothesized that silyl triflate 20 initially forms but undergoes rapid and irreversible loss of triflate with C–C bond cleavage to form 21. Upon aqueous workup, iminium 21 is hydrolyzed to give 22.⁵²

Scheme 3.

Unsuccessful Attempt toward Accessing an Indolizidyne Precursor

To circumvent the undesired pathway shown in Scheme 3, we targeted silyl tosylate 25 as a precursor to 5 (Scheme 4). The lower leaving group ability of tosylates in comparison to triflates was expected to suppress the undesired fragmentation; in addition, silyl tosylates have been used as precursors to other strained intermediates.⁵³ We were pleased to find that treatment of 19 with L-selectride, followed by p-toluenesulfonic anhydride (Ts₂O), gave 23 in 80% yield. Fragmentation product 22 was not observed. Next, the basic amine in 23 was protonated with p-toluenesulfonic acid (TsOH) prior to treatment with a Grubbs II catalyst, which yielded metathesis product 24 in 63% yield. Hydrogenation yielded desired indolizidyne precursor 25.

Scheme 4.

Synthesis of Indolizidyne Precursor 25

With access to silyl tosylate 25, we were eager to evaluate indolizidyne generation and the key Pd-catalyzed annulation, as this would provide direct access to tylophorine (9) (Scheme 5). We first attempted the key step using conditions developed for annulating piperidyne 16 (see Scheme 2) but only observed trace conversion to tylophorine (9). We ultimately found that by altering the solvent to 1:4 DMF/PhMe and adding tetrabutylammonium triflate to modulate fluoride solubility, the yield of 9 increased to 57%. This result underscores the inherent challenge of intercepting strained intermediates in metal-catalyzed processes, as one must identify conditions that allow for the simultaneous generation of 26 and 5 at low concentration, followed by a productive biomolecular reaction and catalyst turnover. It is also notable that the reaction proceeds in the presence of the basic tertiary amine. Only one example of a piperidyne possessing a basic tertiary amine has been proposed in the literature, arising from a flash vacuum pyrolysis experiment.⁵⁴ Moreover, the catalyst activity in the annulation is not detrimentally hampered by the presence of the basic amine.

Scheme 5.

Pd-Catalyzed Annulation of Indolizidyne 5 to Furnish Tylophorine (9)

As highlighted in Scheme 6, the indolizidyne annulation provides a strategy to directly access other phenanthroindolizidines as a final step. Use of biaryl bromide 27 in the annulation reaction with indolizidyne precursor 25 furnished a 1:1 ratio of tylocrebine (28) and isotylocrebine (29) in 68% yield. Both of these natural products demonstrate anticancer properties.^55,56 Overall, the indolizidyne approach obviates the need for protecting groups and late-stage manipulations while providing advances in azacyclic alkyne chemistry.

Scheme 6.

Indolizidyne Trapping Provides Tylocrebine(28) and Isotylocrebine (29)

This study demonstrates that strained azacyclic alkynes (i.e., piperidynes and indolizidynes) serve as valuable building blocks in total synthesis. We hope these efforts will prompt further investigation and use of strained azacyclic alkynes for the synthesis of complex molecules.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.