Alessandro Dimasi‡
a,
Mattia Failla‡a,
Arianna Montolia,
Andrea Citarellaa,
Paolo Ronchib,
Daniele Passarellaa and
Valerio Fasano*a
aDepartment of Chemistry, Università degli Studi di Milano, Via Camillo Golgi, 19, 20133 Milano, Italy. E-mail: valerio.fasano@unimi.it; Web: https://www.fasanolab.com
bMedicinal Chemistry and Drug Design Technologies Department, Global Research and Preclinical Development, Chiesi Farmaceutici S.p.A, Largo Francesco Belloli 11/a, 43126 Parma, Italy
First published on 13th February 2024
Caerulomycins, natural alkaloids with antimicrobial properties, have been previously synthesized starting with highly pre-functionalized building blocks or requiring many functional group manipulations. In this work, we report the first total synthesis of caerulomycin K, a diversely trifunctionalized pyridine readily assembled in three steps exploiting the recent advancements in the C–H activation of N-heterocycles.
Specifically, the installation of the methoxy group can require four steps: formation of the N-oxide with an oxidant, nitration with concentrated H2SO4, nucleophilic aromatic substitution (SNAr) with MeONa, and removal of the oxide with Ac2O.18,19 The insertion of an ortho-pyridine group is usually more rapid but requires the use of Grignard reagents or prefunctionalized 2-bromopyridines.15,16 Finally, the insertion of the carbonyl group is achieved by oxidation of a methyl group whose installation has been obtained only with a halogen (Cl or Br) already placed in ortho-position.18,21 Given the recent advancements in selective C–H functionalizations of pyridines,4,5 it would be expected that alternative strategies should now allow a faster synthesis of caerulomycins. Herein, we report our efforts to rapidly convert a cheap monosubstituted pyridine into caerulomycin K, a recently isolated alkaloid whose total synthesis has never been reported before.
Our investigation began with the design of a synthetic route that would furnish caerulomycin K in a few steps using two C–H activations, thus avoiding highly pre-functionalized starting materials. In an initial retrosynthetic approach, we imagined that the aldoxime group could be derived from a methyl group, as reported by Quéguiner and co-workers, thus leading to trifunctionalized pyridine I (Scheme 2A).18 At this point, we envisaged that I could be accessed by a selective difunctionalization of pyridyl bis-phosphonium salt III via two sequential ligand-coupling (LC) reactions, that is formal SNAr reactions where a phosphonium group is replaced by opportune nucleophiles.22–24 Salt III could then be obtained from 2-phenylpyridine 1 via two consecutive C–P bond formation reactions, in analogy with a rare example of a pyridyl bis-phosphonium salt (Scheme 2B).25,26
Scheme 2 Reactivity of bis-phosphonium salts and application in the planned retrosynthesis of caerulomycin K. The triflate anion is not reported for clarity. |
This strategy would provide the desired product in 5 steps, whereas the ionic nature of most intermediates would reduce the need for column chromatography. Moreover, considering the wide versatility of ligand-coupling reactions,25,26 III would be a strategic intermediate for the synthesis of libraries of trifunctionalized pyridines by simply changing the order and the nature of the added nucleophiles. In the laboratory, 1 was dissolved in dichloromethane and cooled down to −78 °C, before sequentially adding Tf2O, Ph3P, and DBU (Scheme 3). In agreement with McNally's work,27 upon workup, 2 was easily precipitated out as a white powder from cold ether (88% isolated yield). Notably, the Ph3P addition occurs almost exclusively at the para-position (due to stereoelectronic reasons), thus no regioselectivity problems are encountered during this reaction.28 This was also confirmed by 31P NMR, with only a sharp singlet observed at 23.01 ppm. Repeating the procedure using now 2 as the starting material, the reaction crude revealed two new signals of similar intensity at 23.55 ppm and 15.37 ppm.
These signals were respectively assigned to the para- and ortho-phosphine of bis-phosphonium bis-triflate 3. However, the conversion was only modest by 31P NMR, with significant unreacted 2 and Ph3PO observed in the reaction mixture. The failure of the second C–P bond formation was attributed to a problematic N-activation since 2 should be less nucleophilic than 1 due to its cationic nature. This was confirmed using phosphonium 5, obtained in good yield from 3-phenylpyridine 4: moving away the phenyl ring from the ortho- to the meta-position improved the second C–P bond formation (5 less sterically encumbered than 2), yet not to a significant extent due to electronic reasons. Indeed, bis-phosphonium 6 was found as a minor component in 31P NMR spectrum of the reaction crude (signals at 21.93 ppm and 17.08 ppm, with a 3JP–P = 6.0 Hz). Deuterodephosphination29 of this reaction mixture further confirmed the poor conversion, with the isolated pyridine showing almost quantitative d-incorporation in para-position but a limited deuteration on the C6-site (Scheme 4).
Finally, attempts to use a more nucleophilic phosphine (i.e. (4-anisyl)3P) did not improve the C–P bond formation, and neither did the use of it as the first installed phosphine (see ESI†). Indeed, the use of (4-anisyl)3P mainly resulted in the formation of the corresponding phosphine oxide, as expected for electron-rich phosphines. Given the problematic separation of salts 2 and 3 and the modest conversion observed in the second step, we decided to perform one C–P bond formation at a time. Treatment of 2 with MeONa in dichloromethane gave disubstituted pyridine 7 in 53% 1H NMR yield (Scheme 5), although its isolation was complicated by co-eluting Ph3PO (the product, together with 1, of competitive protodephosphination). Before performing the second C–P bond formation on 7, the installation of a methyl group using 2 as a model compound was attempted.
Scheme 5 Ligand-coupling reactions with phosphonium salt 2. 1H NMR yield determined using CH2Br2 as internal standard (in brackets, isolated yield). |
Indeed, while the replacement of Ph3P with chalcogens/pnictogen nucleophiles (–OR, –SR, –NR2) is relatively straightforward,27,30,31 the installation of alkyl or aryl groups via ligand-coupling requires additional manipulations.32–34 However, the direct use of organolithium has been shown successful in a couple of cases (e.g. ArLi), hence we hoped the use of MeLi or MeMgBr would avoid extra steps.27 Unfortunately, treatment of 2 with these organometallics provided equimolar amounts of 2,4-diphenylpyridine 8 and Ph2P(O)Me (Scheme 5), the latter observable in the 1H NMR spectrum (2.01 ppm, d, 2JP–H = 13.2 Hz, 3H, Me). This result highlights how, in contrast to alkoxides, the phenyl ring has a higher migration aptitude than a methyl group during the ligand-coupling of the phosphorane intermediate,35,36 thus leading to 8 and Ph2PMe (then oxidized during the workup). An alternative approach to the use of phosphonium salts would be an ortho-halogenation followed by Negishi coupling with MeZnCl, in analogy with the reported synthesis of caerulomycin E.18 These halogenations (Reissert–Henze reactions) require the use of N-oxides, easily made upon treatment of pyridines with an oxidant such as H2O2 or m-chloroperbenzoic acid (mCPBA).37 Initially, using 1-O as a model substrate, activation with Tf2O and bromination with tetra-n-butylammonium bromide (TBAB) was attempted, in analogy with the ortho-bromination of quinolines reported by Baran and co-workers (Scheme 6).38 However, no desired product was observed, but only a mixture of brominated pyridines, probably due to some Br2 formed from the residual oxidant (mCPBA) still present in 1-O. In contrast to pyridines, the successful ortho-bromination observed by Baran for quinolines reflects the lower loss in resonance stabilization typical of bicyclic aromatics (i.e. naphthalene vs. benzene). Chlorination of 1-O with POCl3 gave better results (9 isolated in 55% yield), but the need for harsh conditions (neat POCl3 refluxing at 106 °C) somehow defeated our original purpose for a short and mild synthesis, thus a completely different strategy was considered.
Minisci-type chemistry is an excellent method for ortho-functionalizations of pyridines, especially employing nucleophilic carbon-based radicals (ideal for the synthesis of caerulomycin K).39,40 Moreover, starting with a 4-substituted pyridine would prevent regioselectivity issues (C2 vs. C6) typical of unsymmetrical starting materials. For the ortho-arylation of pyridines, Baran and others have shown how aryl boronic esters, in combination with AgNO3, Na2S2O8, and TFA, are excellent aryl radical precursors.41,42 To install a carbonyl group, Angeles, Yeung, and colleagues have used 1,3,5-trioxanes as an aldehyde equivalent in Minisci-type carbonylation of pyridines.43 Based on this precedent, a Minisci arylation of 4-chloropyridine 10 was performed (Scheme 7). In this case, an excess of phenylboronic acid (1.5 equiv.) was needed to compensate for competitive protodeboronation, whereas a higher loading of AgNO3 allowed the isolation of product 11 in 56% yield. A second Minisci reaction was then performed on this pyridine using 1,3,5-trioxane in the presence of (nBu4N)2S2O8. A successful ortho-alkylation gave product 12 in 50% yield (a value in agreement with previous reports),44–47 whereas a subsequent nucleophilic aromatic substitution allowed to access trifunctionalized pyridine 13 almost quantitatively. It has to be noted that starting from 4-methoxypyridine 14 would shorten the synthesis, but the electron-donating effect of the methoxy group will negatively affect both steps since Minisci reactions are based on the addition of nucleophilic radicals. The final conversion of 13 into caerulomycin K was achieved in a one-pot procedure by treatment with HCl (to reveal the aldehyde functionality), followed by condensation with NH2OH. Importantly, 12 could be directly converted into caerulomycin K without the need for isolation of 13, further simplifying the synthesis (route in green). Therefore, this three-step total synthesis (overall yield of 10%) represents the first synthesis of caerulomycin K as well as a potential alternative to the synthesis of caerulomycins.
In conclusion, the first total synthesis of caerulomycin K has been reported. Starting from monofunctionalized pyridines, the first strategy looked at a double C–H activation by means of phosphonium chemistry. However, a poor conversion of the second C–P bond formation and a problematic ortho-methylation, including via halogenation, prompted the search for a better alternative. This was achieved by sequential Minisci ortho-arylation and ortho-alkylation, with the latter converted in one pot into the desired oxime. Compared to previously reported caerulomycin syntheses, this novel approach does not require highly pre-functionalized starting materials.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ra00589a |
‡ Equal contribution. |
This journal is © The Royal Society of Chemistry 2024 |