2.1. First study toward mycangimycin

When we started our studies on the synthesis of endoperoxides, we were attracted to mycangimycin, whose 1,2-dioxolane is highly similar to the THF ring, a heterocycle that has been intensively studied in our laboratory in past years [9]. The polyenic chain was presumably unstable, and we believed at the time that the antimalarial activity of mycangimycin could be attributed to the endoperoxide ring, while the antifungal activity would come from the polyenic chain, in a manner similar to amphotericin B, for instance. However, the total synthesis of mycangimycin still constitutes a challenge. On the one hand, the synthetic access to the chiral endoperoxide core requires the development of new methods. On the other hand, synthesizing the polyenic chain appears to be unattainable due to presumed instability. Some preliminary works have indeed shown that the synthesis of polyenes with defined E or Z stereochemistry becomes increasingly challenging after three or four conjugated double bonds. Thus, it was preferred to focus on the synthesis of the endoperoxide, designing a strategy to access saturated analogs of mycangimycin and evaluate their biological properties. For this reason, access to chiral endoperoxides was not a target at this stage; indeed, given the lack of appropriate methods, this was beyond the scope of this study.

After numerous pioneering experiments, it was discovered that the peroxidation of cyclopropanol is a convenient method for accessing the 1,2-dioxolane scaffold. Earlier reports indicated that cobalt [10] and manganese [11] catalysis promote this transformation. Cyclopropanol 2 could be obtained in high yield through the Kulinkovich reaction between alkene 1 and methyl palmitate, utilizing Cha’s protocol, which enabled titanocene metathesis at the terminal double bond [12] (Scheme 2). Oxygen insertion in cyclopropanol 2 proceeded in high yield following a procedure by Wu [10] that reported the use of Co(acac)₂ in EtOH. The remaining OH group in 1,2-dioxolane 3 was reduced with Et₃SiH and TfOH activation, via a peroxycarbenium intermediate [13]. By this route, the cis configuration was predominantly obtained (dr = 83:17), with the hydride reacting on the less hindered face. Fluoride cleavage of the TBDPS group (tert-butyldiphenylsilyl) and ruthenium-catalyzed oxidation afforded 6, a racemic and saturated analog of mycangimicyn.

2.2. Alkylation of 1,2-dioxolanyl acetates

Although the strategy used provided, in a few steps, a simplified analog of mycangimycin, it lacked flexibility because the Kulinkovitch reaction was not compatible with multiple unsaturations or proximal steric hindrance [14]. Inspired by Dussault and Woerpel’s pioneering works [15, 16, 17], it was found that endoperoxyacetals could serve as an excellent platform for gaining molecular diversity in the synthesis of 3,5-disubstituted-1,2-dioxolanes [18]. Indeed, it would be possible to generate peroxycarbenium ions from these precursors, which could be alkylated by various neutral silylated nucleophiles. However, it is more challenging to obtain these reactive species than regular oxycarbenium ions due to the electronegativity of the second oxygen, which decreases the donor effect required for stabilization (Figure 2). The relative stability of different oxy- and peroxycarbenium species was calculated using DFT (B3LYP/6-311G++(d,p) level of theory), revealing a significant impact of this phenomenon, with energy differences ranging from 13.4 to 21.5 kcal/mol compared to regular oxycarbenium species. The specific cyclic constraint of 1,2-dioxolane peroxycarbenium ions also makes this species more difficult to produce. The direct consequence was the difficulty of synthesizing methoxyperoxyacetal 9a–i and subsequently making it react (Scheme 3); we thus turned our attention to the synthesis of acetoxy derivatives 8a–i (Scheme 3) [19, 20]. Acetoxy is indeed a better leaving group, facilitating the promotion of peroxycarbenium ions. The synthesis of 8a–i went through the ring expansion of cyclopropanols 6a–i with molecular oxygen under Mn(acac)₃ or Co(acac)₂ catalysis in THF [10, 11], followed by an acetylation step. The conditions of acetylation were crucial; regular pyridine or DMAP-mediated reactions led to a fragmentation of the endoperoxide, whereas rare-earth metal catalysis, such as with ytterbium, produced the target acetate in high yields [21]. Cyclopropanols 6a–i were prepared following three different methods (A–C). Method A involved a Kulinkovich reaction [22], but was only applicable to monosubstituted cyclopropanols and in cases where a Grignard reagent could be prepared or did not bear highly hindered substituents. Method B offered an alternative for the last two situations. Chromium (II) chloride-mediated reductive cyclization of methacroleins [23, 24] limited the β-elimination pathway to produce 6f and allowed the introduction of a bulky tert-butyl group (6g). The synthesis of disubstituted 1,2-dioxolanes required Method C, which involved a dihalocyclopropanation and a Matteson reaction, followed by a boron oxidation [25].

The next objective was then to substitute the acetate function. The first set-up conditions were inspired by Woerpel and Dussault’s works [16, 17] and involved TiCl₄ or SnCl₄ as Lewis acid (Scheme 4, Conditions A) to promote the peroxycarbenium ion, enabling the addition of several nucleophile types (allylsilanes, enoxysilanes, silanes, or silylcyanides and azides) [19]. Yields were generally good, particularly with SnCl₄. However, silylketene acetals proved to be unreactive under these conditions (10m–o). Diastereoselectivity was generally low (except with TiCl₄), such as with THF rings [26]. Stoichiometry was optimum with 0.9 equiv of Lewis acid, which allowed full conversion and minimized degradation pathways. TiCl₄ was particularly reactive and consequently gave generally lower yields than SnCl₄ due to greater degradations. However, the higher diastereoselectivity toward the trans product with TiCl₄ was surprising, and it was proposed, supported by DFT calculations, that the cis diastereomer degrades faster than the trans product, explaining its prominence. Degradation pathways involve Lewis acid-mediated fragmentation of the 1,2-dioxolane ring and Kornblum–DeLaMare rearrangement [27].

Following this initial study, new conditions were needed to limit the degradation pathways and improve the reproducibility of the various alkylation reactions. Indeed, maintaining the temperature at −40 °C without proper equipment, as well as the high reactivity of TiCl₄ and SnCl₄, which can lead to overdegradation, accounts for the variability in the results. InCl₃, in combination with TMSCl [28, 29, 30], as well as rare-earth triflate [31, 32, 33] have been reported for Sakurai or Mukaiyama aldol reactions, with the main advantage being that they can be used in catalytic amounts under mild conditions. The use of InCl₃ (5 mol%) with 2.5 equiv of TMSCl showed high capacity in the substitution reaction from endoperoxyacetals 8a–g (Scheme 4, Condition B). The reaction typically completes within 1–2 h at room temperature [20]. No significant degradation was observed, unless the reaction was allowed to continue for more than 24 h. The scope of the reaction is large, with yields and selectivities comparable to those obtained with SnCl₄. However, halo-allylsilanes were unreactive with these catalysts, giving no conversion to 10f or 10i. The geminal halogen deactivates the silicon atom, which explains a lack of reactivity in this case; however, the observed reactivity with SnCl₄ or TiCl₄ is probably caused by a transmetallation with silicon to produce more reactive species. In contrast, the catalytic conditions enabled the reaction with silylketene acetals (esters 10o or thioesters 10m–n), which is of great importance considering the challenge that total synthesis of endoperoxides represents. Rare-earth metal triflates were also studied as mild Lewis acids. Among all salts tested, scandium was found to show the highest reactivity, similar to that of InCl₃/TMSCl. The main advantage of Sc(OTf)₃ is that only one reagent needs to be added to the reaction mixture; the diastereomeric ratio and yields were of the same magnitude as those obtained with either SnCl₄ or InCl₃/TMSCl. It is noteworthy that cis- and trans-1,2-dioxolanes are, in general, not separable by silica gel chromatography; therefore, all the products were isolated and characterized as mixtures.

2.3. Biological evaluation of 1,2-dioxolanes

Evaluation of twenty-three 1,2-dioxolanes, which constitute analogs of 1,2-mycangimycin, was performed against four different infectious agents [34]. As mycangimycin possesses strong antifungal activity, our 1,2-dioxolanes were evaluated against Candida albicans CAAL93 and Aspergillus fumigatus ASFU76 strains. However, none of them exhibited any significant antifungal activity. This finding supports the hypothesis that this activity is related to the polyenic chain of mycangimycin. Antimalarial and antileishmanial activities were evaluated on Plasmodium falciparum and Leishmania Donovani. The saturated analog of mycangimycin 5 did not exhibit significant antimalarial activity, suggesting that the side chain also plays a crucial role in the antimalarial activity. Nevertheless, interesting results were found for dioxolanes 8g and 10a, which possess activity in the same range as chloroquine, as well as high selectivity indexes (SI) (Figure 3). Four other compounds were found to be quite active against L. Donovani in its axenic amastigote form. 10f and 10i bear a vinyl halide function, while 10m and 10n possess a thioester function. However, the selectivity index is disappointing overall and these compounds lost most of their activity against intramacrophage amastigotes, suggesting that they cannot properly cross the macrophage and parasitophorous vacuole.

3.1. Oxidative ring expansion of cyclobutanols

After studying 1,2-dioxolanes, we were also interested in synthesizing 1,2-dioxanes. As outlined in the introduction, this heterocycle is predominant among the natural endoperoxides. Our aim was to apply the same strategy to obtain and functionalize the endoperoxide ring as outlined in Scheme 1, i.e., oxygen insertion on a strained ring and addition to a peroxycarbenium ion. Although the insertion of molecular oxygen was a known transformation, its application to cyclobutanols had not been reported previously. The strain energy of cyclobutane compared to cyclopropane is somewhat similar (26.90 versus 28.13 kcal/mol, respectively) [35]. However, the rate of ring opening of methyl radicals is considerably different, making the radical opening of methylcyclopropyl radical very fast [36], whereas the reaction speed is moderate in the case of methylcyclobutyl radical [37, 38] (Figure 4). Improvement can be achieved through α-substitution, which stabilizes the newly formed radical. Similar behavior can be anticipated regarding the relative speed of ring opening with alkoxy radicals of strained rings, suggesting that the radical insertion of molecular oxygen into cyclobutanol would be more challenging.

The various studies on the insertion of molecular oxygen into cyclobutanols implied the synthesis of such compounds. The general synthetic route is based on a 3-step procedure from aldehydes or ketones 11a–i (Scheme 5). The first step consisted of the Wittig olefination with an in-situ-formed cyclopropylphosphonium ylide. Cyclopropylidenes 12a–i underwent epoxidation and rearrangement into cyclobutanones 13a–i [39]. Following, secondary cyclobutanols 14a–i were obtained by reduction of 13a–i with NaBH₄, while tertiary cyclobutanols 15a–d were synthesized by addition of a Grignard or lithium reagent. Optimal conditions for oxygen insertion were identified as 5 mol% Co(acac)₂ in MeCN at 40 °C (8–48 h); Mn(II) or Mn(III) acetylacetonate showed no reaction [40]. As anticipated, the process was significantly more challenging compared to cyclopropanols due to the slower ring opening of cyclobutyloxy radicals, which required higher catalyst loading (5 mol% versus 0.5 mol%), higher temperatures (40 °C versus room temperature), and longer reaction times (8–48 h versus 1–2 h). The substitution also significantly affects the reaction rate. It works particularly well when the 2-position is disubstituted (16a–c, 16f), indicating that the cyclobutane scission is faster in that case; eight hours are generally sufficient for full conversion. With one alkyl substituent, the reaction was more difficult (16d) and required up to 48 h. The tertiary cyclobutanols were also more difficult to react, requiring longer reaction times and giving generally lower yields (17a–c). The steric hindrance appeared to be the important factor, as the rate of ring scission is little affected by the 1-substitution (see Figure 4); Cobalt needs to approach the alcohol function to abstract the hydrogen. The more significant steric effect of the ethyl group makes the reaction even more challenging to achieve (17a). However, the combination of 1-substitution and 2-monosubstitution made the reaction unfavorable (17d, no reaction).

3.2. Alkylation of 1,2-dioxanyl acetates

Endoperoxyacetals 16a–f were then acetylated to intermediates 18a–f to be substituted under the conditions previously developed for 1,2-dioxolanes [20] (Scheme 6). Compound 16f had difficulties reacting in acetylation due to the Lewis basicity of the carbamate function, which deactivated the acid catalyst (18f: 22% yield); all other substrates were acylated efficiently. Nevertheless, it was discovered that 1,2-dioxanes are far more stable than 1,2-dioxolanes toward organic bases; consequently, pyridine-mediated acylation could also be performed efficiently to overcome difficult cases (100% yield for 18f) [41]. Sc(OTf)₃ catalysis was then selected over indium-mediated conditions because of its ease of handling. Several types of nucleophiles could be added to acylated intermediates 18a–e; however, 18f was again unreactive due to the Boc group. The scope of the nucleophiles includes allylsilanes (19b, 19e, 19f–g, 19j–k), hydrides (19c), cyanides (19d, 19h), enolates (19a, 19i, 19l), and siloxyfuranes (vinylogous addition, 19m). Unlike 1,2-dioxolane, the addition can be more selective on this ring size, thanks to a preferred half-chair conformation on the peroxycarbenium ion, in which the substituent is in pseudoequatorial position [42]. The nucleophile can also add preferentially in the axial position because it gives a chair-like conformation to the substitution product directly. The low yield for 19f is attributed to the adjacent aromatic ring, which enhanced Kornblum–DeLaMare or Hock rearrangements [40].

Success in accessing differently substituted 1,2-dioxanes encouraged the application of the method developed to the total synthesis of natural endoperoxides. Mycaperoxides and ethyl plakortide Z were selected for their interesting biological properties and structural features, which present distinct challenges. No total synthesis of these compounds had been reported prior to our studies.

3.3. Total synthesis of mycaperoxides

As outlined in the introduction, mycaperoxides are marine endoperoxides endowed with various reported biological activities. Several natural analogs exist, whose differences rely not only on the sesquiterpene structure but also on the relative and absolute configuration of the 1,2-dioxane scaffold. Synthetic studies toward mycaperoxide B were reported by Harwood [43, 44, 45]. The total synthesis of related analogs of mycaperoxides was nevertheless accomplished by Seifert on diacarnoxide C in twenty steps [46] and by Yikang Wu on muqubilin in twenty-nine steps [47]. To quickly access the decalin subunit of mycaperoxides, the choice of sclareolide as a starting material appeared to be the most appropriate. However, only one configuration is available, which allows access to the natural configuration of mycaperoxide B (Figure 5). Mycaperoxides C, D, and G have an inverse configuration for the decalin subunit; therefore, the synthesis of their antipodes was here privileged. Indeed, no other available resource enables direct access to their natural configuration, unless one designs a complex, multistep synthesis.

The first targeted analog was ent-mycaperoxide D, which bears the same decalin substructure as sclareol. The first step required the KMnO₄-mediated oxidative cleavage of the allylic alcohol, yielding methyl ketone 20 (Scheme 7). The construction of the 1,2-dioxane relied on our cobalt-mediated oxygen insertion reaction, which provided 23 in a good yield over a five-step sequence. Acetylation needed some optimization, since Lewis acid-catalyzed conditions induce the elimination of the silylated alcohol. Instead, pyridine-mediated acylation led cleanly to peroxyacetal 24. Promotion of the peroxycarbenium ion with Sc(OTf)₃ catalysis allowed the addition of silylated propionyl thioester nucleophile 25 in high yield, but thioester 26 was obtained as a mixture of eight different diastereomers. This result was anticipated; the insertion of molecular oxygen could not be stereoselective or stereoretentive due to the reaction’s radical mechanism, resulting in no control over the configuration at the C-6 position. The addition at C-2 was more selective, yielding an interesting 4:1 ratio in favor of the 3,6-cis product. However, the configuration of the methyl group at C-2 was not controlled, affording a 1:1 mixture of 2,3-anti/syn-products. Separation of most of the diastereomers, after triethylsilane (TES) cleavage, enabled the isolation with a 12.7% yield of diastereomer 27, which has the same relative configuration as natural mycaperoxide D. Two-step transesterification gave rise to the antipode of mycaperoxide D (28). Spectroscopic data matched in all points the natural product, for which the characterizations were reported as a methyl ester, except for the optical rotation with an inverted value [+68 (c 0.25, CHCl₃); lit [48]: −52 (c 0.3, CHCl₃)]. The synthesis of 28 was achieved in only eleven chemical steps from sclareol [49].

The synthesis of other mycaperoxide analogs was then investigated but required a modification of the decalin subunit. Methyl ketone 20 was treated with catalytic iodine to promote the more selective elimination of the tertiary alcohol toward 29 in an 85:15 mixture with regioisomer 30 (Scheme 7). The entire mixture was subjected to epoxidation with m-CPBA, followed by olefination to a cyclopropylidene. The two-step sequence yielded a separable mixture of diastereomers 31a and 31b, as well as regioisomer 32, which came from 30. Next, 31a and 31b underwent distinct LiAlH₄ reductions, enabling regioselective hydride addition, in accordance with the Fürst–Plattner rule [50, 51, 52].

Regioisomer 33b exhibited the decalin scaffold of ent-mycaperoxyde C and underwent a sequence similar to that used for ent-mycaperoxide D from 21 (Scheme 7). Differences lie in the protecting group (TMS, less bulky than a TES) and the choice of yttrium in replacement of scandium for the aldol reaction, which prevented the elimination of the tertiary silyl ether on the decalin, thanks to a more moderate reactivity. Separation of the diastereomers and transesterification afforded the antipode of mycaperoxide D (37). Spectroscopic data agreed with the natural product’s methyl ester, except for the optical rotation with an inverted value [+50 (c 0.198, CHCl₃); lit [48]: −71 (c 1.1, CHCl₃)]. The synthesis of 37 was achieved in fourteen steps from sclareol [49].

The synthesis of mycaperoxide B and ent-mycaperoxide G was subsequently studied from 33a, applying the same strategy (Scheme 7). Thus, thioester 40 was obtained as a mixture of eight diastereomers in a four-step sequence. Because the hydroxy group on decalin was highly prone to elimination, it was left free, and yttrium catalysis was again applied with success. The absence of a protecting group lowered the 3,6-cis:trans selectivity (dr = 3:2) for the aldol reaction, which was beneficial because mycaperoxides B and G have a 3,6-trans configuration. Separation of diastereomers followed by a saponification step afforded a mixture of mycaperoxide B (41) and its 2,3,6-epimer (42) in about 1:1 ratio (7.6%), and in a total of eleven steps. However, no conditions for further separation were found. In parallel, the diastereomeric mixture 40 was treated with SOCl₂, allowing elimination of the tertiary alcohol. Subsequently, separation of isomers by chromatography and two-step transesterification allowed the isolation of the antipode of mycaperoxide G’s methyl ester (44) and its 13,8-epimer (43) in about 1:1 ratio (3.4%) in a total of thirteen steps.

The total synthesis of various mycaperoxides was accomplished for the first time, enabling access to these natural structures in a few steps (from eleven to fourteen steps). This work highlights the effectiveness of the strategy developed for the synthesis of 1,2 dioxanes. Although the key transformations involved in endoperoxide formation suffer from limited selectivity (radical and S_N1 reaction), this limitation is offset by the overall efficiency of the approach, particularly in terms of step economy when compared with previous reported studies in the field [43, 44, 45, 46, 47].

3.4. Total synthesis of ethyl plakortide Z

Plakortides have the unique characteristic of showing an ethyl substituent at the C-4 position of the 1,2-dioxane ring, which presents a new challenge in total synthesis. Gemma and Campiani reported the total synthesis of 9,10-dihydroplakortin and its 6-epi analog in twenty-three steps, utilizing Mukaiyama peroxysilylation and iterative Evans aldol reaction as key transformations [53]. Despite a moderate cytotoxic activity, ethyl plakortide Z proved to be a suitable target for validating our synthetic strategy within this class of compounds; the side chain is simple and the 3,4-trans configuration seemed more adapted to the peroxycarbenium-mediated aldol reaction.

Our designed pathway for synthesizing ethyl plakortide Z began with the preparation of cyclobutanone 46 in two steps (Scheme 8). The first key step was controlling the ethyl group at position C-4, which necessitated the utilization of an Enders hydrazone to produce 48. Hydrazone cleavage with ozone, followed by reduction with NaBH₄ yielded 49. The second key step involved the cobalt-mediated insertion of oxygen, which proceeded exclusively on the most substituted position to produce 50 as a C-6 diastereomeric mixture. Activating the acetal function to an acetate enabled the thioacetate insertion with Sc(OTf)₃ catalysis in the final key step. The reaction proceeded with a moderate diastereoselectivity (dr = 4:1) toward the 3,4-trans configuration. The desired diastereomer could be isolated from the other three by preparative HPLC, and a two-step transesterification provided ethyl plakortide Z as a pure enantiomer; the spectral data of the synthetic ethyl plakortide Z matched those of the reported natural one [54].

Ethyl plakortide Z was synthesized in only eleven steps, yielding 4.2% overall, constituting a significant advancement for this class of compound. However, this stereodivergent strategy presents significant limitations, notably the requirement for laborious chromatographic separations of diastereoisomers, arising from insufficient control over peroxy bond formation. A major advance would be designing asymmetric versions of the key reactions, specifically the oxygen insertion and Mukaiyama aldol reactions. In the first case, the radical pathway would make it difficult to develop such a reaction with current knowledge in the field. However, the ionic interaction in the final key step would certainly enable asymmetric control with a chiral counteranion.

4.1. Pioneering works

The lack of readily available asymmetric methods for forming peroxyl bonds is a major obstacle to the total synthesis of endoperoxides. Past known methods relied either on: S_N2 displacement of hydroperoxides to epoxides, halides, or sulfonates [46, 47, 55]; kinetic resolution of hydroperoxides by reduction [56, 57, 58]; asymmetric organocatalyzed 1,4-addition of a hydroperoxyl function to enones or enals [13, 59, 60].

The similarity between ether-based oxycarbenium ions and peroxycarbenium species suggests that asymmetric methods used for ethers could also be applied to peroxyl analogs. Indeed, in recent years, List et al. have demonstrated that new organocatalysts with high acidity and confinement properties, called imidodiphosphorimidates (IDPi), could be used in a large range of reactions involving reactive cationic species [61, 62]. One major breakthrough involved the synthesis of chiral ethers from cyclic acetals 54a–c with high yields, high enantiomeric ratios, and low catalyst loadings of IDPi-H 53 (Scheme 9) [63]. Consequently, we thought that this asymmetric method could be applied to endoperoxides, although some additional difficulties needed to be overcome, including the fragility of the peroxyl bond and the reduced stabilization of the peroxycarbenium species compared to classical oxycarbenium ions (Figure 2).

4.2. Synthesis of chiral endoperoxides

Searches for optimal reaction conditions led to the selection of IDPi catalyst 53 as the most versatile and selective catalyst for our substrates [64], in accordance with List’s previous work on cyclic ethers that possess similar geometries [63] (Scheme 10). The reactions proceeded more selectively in Et₂O. Other nonpolar solvents, used to maximize intermolecular interactions between the two ionic species, can also be employed, such as methylcyclohexane or n-pentane, which provided results similar to those obtained with Et₂O. However, their limited solubilizing power prevents their general use across a broad range of substrates. Other solvents, such as dichloromethane (DCM) or toluene, led to no reactivity or degradation. Temperature played a crucial role; carrying out the reaction at low temperature (−90 °C) diminished acid-catalyzed peroxide degradation pathways, increasing the isolated yields dramatically, albeit at the cost of a long reaction time (2–5 days). Nevertheless, increasing the reaction temperature to −78 °C has been used as a compromise for less reactive nucleophiles or substrates. Another important observation is that the reaction does not tolerate variation of the leaving group, with the acetate moiety remaining the only one capable of promoting this transformation. A similar trend is observed for the silylated nucleophiles. Various protecting groups, differing in both steric and electronic properties, were evaluated, but only the TBS group demonstrated satisfactory reactivity and reproducibility.

This unprecedented use of IDPi catalysts for the functionalization of 1,2-dioxane acetates has enabled the preparation of a wide variety of products by tuning the nucleophile (Scheme 10). In this context, several classes of nucleophiles were explored, including silyl ketene acetals, as well as aromatic and cyclic silyl enol ethers. In all cases, the corresponding addition products 55a–f, 56a–c, 57a–c, 58a were obtained in good yields and with remarkable enantioselectivities (ee = 93–99%) [64]. In the case of cyclic enol ethers, an outstanding diastereoselectivity can be observed in favor of the anti product (dr = 90:10–95:5). As anticipated, the 1,2-dioxolanes are more difficult to produce due to their lesser stability. Ketone adducts generally led to degradations, but the silyl ketene acetals allowed in this instance the isolation of functionalized 1,2-dioxolanes 59b and 60b in very good yields and high enantioselectivity (ee = 93 and 85%, respectively). The reaction was also successfully examined with vinylogous nucleophiles, such as siloxyfuranes [64].

The proposed reaction mechanism (Scheme 11) is in line with some kinetic investigations and previous studies [62, 63, 64]. The initial step involves a protodesilylation between the nucleophile and IDPi-H, generating silylium species IDPi-TBS, the actual Lewis acid. It then promotes departure of the acetate group from 18a, leading to peroxycarbenium ion I. The chiral environment brought by the chiral counteranion allows a controlled nucleophilic addition to the most accessible face of intermediate I. Adduct II can subsequently release IDPi-TBS after providing 55a, thereby sustaining the catalytic cycle.

4.3. Second-generation total synthesis of ethyl plakortide Z

The aforementioned asymmetric reaction was successfully applied to the total synthesis of ethyl plakortide Z, by employing acetate intermediate 50 [64] (Scheme 12). The use of IDPi catalyst (S,S)-53, in combination with silylated nucleophile 61, enabled direct access to ethyl plakortide Z with an excellent diastereomeric ratio of 18:1 in favor of the 3,4-trans configuration. As a reminder, Sc(OTf)₃ provided the 3,4-trans configuration with a moderate 4:1 ratio in our previous studies and required an additional transesterification step [54] (Scheme 8). Interestingly, the use of (R,R)-53 catalyst decreased the ratio to 1.3:1 without reversing the selectivity toward the 3,4-cis product. This result suggests that the substrate’s chirality predominantly controls the substitution, rather than the catalyst, in this example; the catalyst only amplifies or diminishes the original selectivity.